1 \input texinfo @c -*-texinfo-*-
2 @comment %**start of header
3 @setfilename ../../info/eintr
4 @c setfilename emacs-lisp-intro.info
5 @c sethtmlfilename emacs-lisp-intro.html
6 @settitle Programming in Emacs Lisp
11 @include emacsver.texi
13 @c ================ How to Print a Book in Various Sizes ================
15 @c This book can be printed in any of three different sizes.
16 @c Set the following @-commands appropriately.
26 @c European A4 size paper:
31 @c (Note: if you edit the book so as to change the length of the
32 @c table of contents, you may have to change the value of `pageno' below.)
34 @c <<<< For hard copy printing, this file is now
35 @c set for smallbook, which works for all sizes
36 @c of paper, and with PostScript figures >>>>
44 @c ================ Included Figures ================
46 @c If you clear this, the figures will be printed as ASCII diagrams
47 @c rather than PostScript/PDF.
48 @c (This is not relevant to Info, since Info only handles ASCII.)
49 @set print-postscript-figures
50 @c clear print-postscript-figures
52 @comment %**end of header
54 @c per rms and peterb, use 10pt fonts for the main text, mostly to
55 @c save on paper cost.
56 @c Do this inside @tex for now, so current makeinfo does not complain.
62 \global\hbadness=6666 % don't worry about not-too-underfull boxes
65 @c These refer to the printed book sold by the FSF.
66 @set edition-number 3.10
67 @set update-date 28 October 2009
69 @c For next or subsequent edition:
70 @c create function using with-output-to-temp-buffer
71 @c create a major mode, with keymaps
72 @c run an asynchronous process, like grep or diff
74 @c For 8.5 by 11 inch format: do not use such a small amount of
75 @c whitespace between paragraphs as smallbook format
78 \global\parskip 6pt plus 1pt
82 @c For all sized formats: print within-book cross
83 @c reference with ``...'' rather than [...]
85 @c This works with the texinfo.tex file, version 2003-05-04.08,
86 @c in the Texinfo version 4.6 of the 2003 Jun 13 distribution.
89 \if \xrefprintnodename
90 \global\def\xrefprintnodename#1{\unskip, ``#1''}
92 \global\def\xrefprintnodename#1{ ``#1''}
94 % \global\def\xrefprintnodename#1{, ``#1''}
97 @c ----------------------------------------------------
99 @dircategory GNU Emacs Lisp
101 * Emacs Lisp Intro: (eintr).
102 A simple introduction to Emacs Lisp programming.
106 This is an @cite{Introduction to Programming in Emacs Lisp}, for
107 people who are not programmers.
110 Edition @value{edition-number}, @value{update-date}
113 Distributed with Emacs version @value{EMACSVER}.
117 <p>The homepage for GNU Emacs is at
118 <a href="http://www.gnu.org/software/emacs/">http://www.gnu.org/software/emacs/</a>.
119 <br>To view this manual in other formats, click
120 <a href="/software/emacs/emacs-lisp-intro/emacs-lisp-intro.html">here</a>.
124 Copyright @copyright{} 1990--1995, 1997, 2001--2013 Free Software
131 GNU Press, @hfill @uref{http://www.fsf.org/licensing/gnu-press/}@*
132 a division of the @hfill email: @email{sales@@fsf.org}@*
133 Free Software Foundation, Inc. @hfill Tel: +1 (617) 542-5942@*
134 51 Franklin Street, Fifth Floor @hfill Fax: +1 (617) 542-2652@*
135 Boston, MA 02110-1301 USA
139 Printed copies available from @uref{http://shop.fsf.org/}. Published by:
142 GNU Press, http://www.fsf.org/licensing/gnu-press/
143 a division of the email: sales@@fsf.org
144 Free Software Foundation, Inc. Tel: +1 (617) 542-5942
145 51 Franklin Street, Fifth Floor Fax: +1 (617) 542-2652
146 Boston, MA 02110-1301 USA
153 Permission is granted to copy, distribute and/or modify this document
154 under the terms of the GNU Free Documentation License, Version 1.3 or
155 any later version published by the Free Software Foundation; there
156 being no Invariant Section, with the Front-Cover Texts being ``A GNU
157 Manual'', and with the Back-Cover Texts as in (a) below. A copy of
158 the license is included in the section entitled ``GNU Free
159 Documentation License''.
161 (a) The FSF's Back-Cover Text is: ``You have the freedom to
162 copy and modify this GNU manual. Buying copies from the FSF
163 supports it in developing GNU and promoting software freedom.''
166 @c half title; two lines here, so do not use `shorttitlepage'
169 \hbox{}\vskip 1.5in \chaprm \centerline{An Introduction to}%
171 {\begingroup\hbox{}\vskip 0.25in \chaprm%
172 \centerline{Programming in Emacs Lisp}%
173 \endgroup\page\hbox{}\page}
178 @center @titlefont{An Introduction to}
180 @center @titlefont{Programming in Emacs Lisp}
182 @center Revised Third Edition
184 @center by Robert J. Chassell
187 @vskip 0pt plus 1filll
193 @evenheading @thispage @| @| @thischapter
194 @oddheading @thissection @| @| @thispage
198 @c Keep T.O.C. short by tightening up for largebook
201 \global\parskip 2pt plus 1pt
202 \global\advance\baselineskip by -1pt
212 @top An Introduction to Programming in Emacs Lisp
216 This master menu first lists each chapter and index; then it lists
217 every node in every chapter.
220 @c >>>> Set pageno appropriately <<<<
222 @c The first page of the Preface is a roman numeral; it is the first
223 @c right handed page after the Table of Contents; hence the following
224 @c setting must be for an odd negative number.
227 @c global@pageno = -11
230 @set COUNT-WORDS count-words-example
231 @c Length of variable name chosen so that things still line up when expanded.
234 * Preface:: What to look for.
235 * List Processing:: What is Lisp?
236 * Practicing Evaluation:: Running several programs.
237 * Writing Defuns:: How to write function definitions.
238 * Buffer Walk Through:: Exploring a few buffer-related functions.
239 * More Complex:: A few, even more complex functions.
240 * Narrowing & Widening:: Restricting your and Emacs attention to
242 * car cdr & cons:: Fundamental functions in Lisp.
243 * Cutting & Storing Text:: Removing text and saving it.
244 * List Implementation:: How lists are implemented in the computer.
245 * Yanking:: Pasting stored text.
246 * Loops & Recursion:: How to repeat a process.
247 * Regexp Search:: Regular expression searches.
248 * Counting Words:: A review of repetition and regexps.
249 * Words in a defun:: Counting words in a @code{defun}.
250 * Readying a Graph:: A prototype graph printing function.
251 * Emacs Initialization:: How to write a @file{.emacs} file.
252 * Debugging:: How to run the Emacs Lisp debuggers.
253 * Conclusion:: Now you have the basics.
254 * the-the:: An appendix: how to find reduplicated words.
255 * Kill Ring:: An appendix: how the kill ring works.
256 * Full Graph:: How to create a graph with labeled axes.
257 * Free Software and Free Manuals::
258 * GNU Free Documentation License::
263 --- The Detailed Node Listing ---
267 * Why:: Why learn Emacs Lisp?
268 * On Reading this Text:: Read, gain familiarity, pick up habits....
269 * Who You Are:: For whom this is written.
271 * Note for Novices:: You can read this as a novice.
276 * Lisp Lists:: What are lists?
277 * Run a Program:: Any list in Lisp is a program ready to run.
278 * Making Errors:: Generating an error message.
279 * Names & Definitions:: Names of symbols and function definitions.
280 * Lisp Interpreter:: What the Lisp interpreter does.
281 * Evaluation:: Running a program.
282 * Variables:: Returning a value from a variable.
283 * Arguments:: Passing information to a function.
284 * set & setq:: Setting the value of a variable.
285 * Summary:: The major points.
286 * Error Message Exercises::
290 * Numbers Lists:: List have numbers, other lists, in them.
291 * Lisp Atoms:: Elemental entities.
292 * Whitespace in Lists:: Formatting lists to be readable.
293 * Typing Lists:: How GNU Emacs helps you type lists.
297 * Complications:: Variables, Special forms, Lists within.
298 * Byte Compiling:: Specially processing code for speed.
302 * How the Interpreter Acts:: Returns and Side Effects...
303 * Evaluating Inner Lists:: Lists within lists...
307 * fill-column Example::
308 * Void Function:: The error message for a symbol
310 * Void Variable:: The error message for a symbol without a value.
314 * Data types:: Types of data passed to a function.
315 * Args as Variable or List:: An argument can be the value
316 of a variable or list.
317 * Variable Number of Arguments:: Some functions may take a
318 variable number of arguments.
319 * Wrong Type of Argument:: Passing an argument of the wrong type
321 * message:: A useful function for sending messages.
323 Setting the Value of a Variable
325 * Using set:: Setting values.
326 * Using setq:: Setting a quoted value.
327 * Counting:: Using @code{setq} to count.
329 Practicing Evaluation
331 * How to Evaluate:: Typing editing commands or @kbd{C-x C-e}
333 * Buffer Names:: Buffers and files are different.
334 * Getting Buffers:: Getting a buffer itself, not merely its name.
335 * Switching Buffers:: How to change to another buffer.
336 * Buffer Size & Locations:: Where point is located and the size of
338 * Evaluation Exercise::
340 How To Write Function Definitions
342 * Primitive Functions::
343 * defun:: The @code{defun} macro.
344 * Install:: Install a function definition.
345 * Interactive:: Making a function interactive.
346 * Interactive Options:: Different options for @code{interactive}.
347 * Permanent Installation:: Installing code permanently.
348 * let:: Creating and initializing local variables.
350 * else:: If--then--else expressions.
351 * Truth & Falsehood:: What Lisp considers false and true.
352 * save-excursion:: Keeping track of point, mark, and buffer.
356 Install a Function Definition
358 * Effect of installation::
359 * Change a defun:: How to change a function definition.
361 Make a Function Interactive
363 * Interactive multiply-by-seven:: An overview.
364 * multiply-by-seven in detail:: The interactive version.
368 * Prevent confusion::
369 * Parts of let Expression::
370 * Sample let Expression::
371 * Uninitialized let Variables::
373 The @code{if} Special Form
375 * if in more detail::
376 * type-of-animal in detail:: An example of an @code{if} expression.
378 Truth and Falsehood in Emacs Lisp
380 * nil explained:: @code{nil} has two meanings.
382 @code{save-excursion}
384 * Point and mark:: A review of various locations.
385 * Template for save-excursion::
387 A Few Buffer--Related Functions
389 * Finding More:: How to find more information.
390 * simplified-beginning-of-buffer:: Shows @code{goto-char},
391 @code{point-min}, and @code{push-mark}.
392 * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}.
393 * append-to-buffer:: Uses @code{save-excursion} and
394 @code{insert-buffer-substring}.
395 * Buffer Related Review:: Review.
398 The Definition of @code{mark-whole-buffer}
400 * mark-whole-buffer overview::
401 * Body of mark-whole-buffer:: Only three lines of code.
403 The Definition of @code{append-to-buffer}
405 * append-to-buffer overview::
406 * append interactive:: A two part interactive expression.
407 * append-to-buffer body:: Incorporates a @code{let} expression.
408 * append save-excursion:: How the @code{save-excursion} works.
410 A Few More Complex Functions
412 * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}.
413 * insert-buffer:: Read-only, and with @code{or}.
414 * beginning-of-buffer:: Shows @code{goto-char},
415 @code{point-min}, and @code{push-mark}.
416 * Second Buffer Related Review::
417 * optional Exercise::
419 The Definition of @code{insert-buffer}
421 * insert-buffer code::
422 * insert-buffer interactive:: When you can read, but not write.
423 * insert-buffer body:: The body has an @code{or} and a @code{let}.
424 * if & or:: Using an @code{if} instead of an @code{or}.
425 * Insert or:: How the @code{or} expression works.
426 * Insert let:: Two @code{save-excursion} expressions.
427 * New insert-buffer::
429 The Interactive Expression in @code{insert-buffer}
431 * Read-only buffer:: When a buffer cannot be modified.
432 * b for interactive:: An existing buffer or else its name.
434 Complete Definition of @code{beginning-of-buffer}
436 * Optional Arguments::
437 * beginning-of-buffer opt arg:: Example with optional argument.
438 * beginning-of-buffer complete::
440 @code{beginning-of-buffer} with an Argument
442 * Disentangle beginning-of-buffer::
443 * Large buffer case::
444 * Small buffer case::
446 Narrowing and Widening
448 * Narrowing advantages:: The advantages of narrowing
449 * save-restriction:: The @code{save-restriction} special form.
450 * what-line:: The number of the line that point is on.
453 @code{car}, @code{cdr}, @code{cons}: Fundamental Functions
455 * Strange Names:: An historical aside: why the strange names?
456 * car & cdr:: Functions for extracting part of a list.
457 * cons:: Constructing a list.
458 * nthcdr:: Calling @code{cdr} repeatedly.
460 * setcar:: Changing the first element of a list.
461 * setcdr:: Changing the rest of a list.
467 * length:: How to find the length of a list.
469 Cutting and Storing Text
471 * Storing Text:: Text is stored in a list.
472 * zap-to-char:: Cutting out text up to a character.
473 * kill-region:: Cutting text out of a region.
474 * copy-region-as-kill:: A definition for copying text.
475 * Digression into C:: Minor note on C programming language macros.
476 * defvar:: How to give a variable an initial value.
477 * cons & search-fwd Review::
482 * Complete zap-to-char:: The complete implementation.
483 * zap-to-char interactive:: A three part interactive expression.
484 * zap-to-char body:: A short overview.
485 * search-forward:: How to search for a string.
486 * progn:: The @code{progn} special form.
487 * Summing up zap-to-char:: Using @code{point} and @code{search-forward}.
491 * Complete kill-region:: The function definition.
492 * condition-case:: Dealing with a problem.
495 @code{copy-region-as-kill}
497 * Complete copy-region-as-kill:: The complete function definition.
498 * copy-region-as-kill body:: The body of @code{copy-region-as-kill}.
500 The Body of @code{copy-region-as-kill}
502 * last-command & this-command::
503 * kill-append function::
504 * kill-new function::
506 Initializing a Variable with @code{defvar}
508 * See variable current value::
509 * defvar and asterisk::
511 How Lists are Implemented
514 * Symbols as Chest:: Exploring a powerful metaphor.
519 * Kill Ring Overview::
520 * kill-ring-yank-pointer:: The kill ring is a list.
521 * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable.
525 * while:: Causing a stretch of code to repeat.
527 * Recursion:: Causing a function to call itself.
532 * Looping with while:: Repeat so long as test returns true.
533 * Loop Example:: A @code{while} loop that uses a list.
534 * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}.
535 * Incrementing Loop:: A loop with an incrementing counter.
536 * Incrementing Loop Details::
537 * Decrementing Loop:: A loop with a decrementing counter.
539 Details of an Incrementing Loop
541 * Incrementing Example:: Counting pebbles in a triangle.
542 * Inc Example parts:: The parts of the function definition.
543 * Inc Example altogether:: Putting the function definition together.
545 Loop with a Decrementing Counter
547 * Decrementing Example:: More pebbles on the beach.
548 * Dec Example parts:: The parts of the function definition.
549 * Dec Example altogether:: Putting the function definition together.
551 Save your time: @code{dolist} and @code{dotimes}
558 * Building Robots:: Same model, different serial number ...
559 * Recursive Definition Parts:: Walk until you stop ...
560 * Recursion with list:: Using a list as the test whether to recurse.
561 * Recursive triangle function::
562 * Recursion with cond::
563 * Recursive Patterns:: Often used templates.
564 * No Deferment:: Don't store up work ...
565 * No deferment solution::
567 Recursion in Place of a Counter
569 * Recursive Example arg of 1 or 2::
570 * Recursive Example arg of 3 or 4::
578 Regular Expression Searches
580 * sentence-end:: The regular expression for @code{sentence-end}.
581 * re-search-forward:: Very similar to @code{search-forward}.
582 * forward-sentence:: A straightforward example of regexp search.
583 * forward-paragraph:: A somewhat complex example.
584 * etags:: How to create your own @file{TAGS} table.
586 * re-search Exercises::
588 @code{forward-sentence}
590 * Complete forward-sentence::
591 * fwd-sentence while loops:: Two @code{while} loops.
592 * fwd-sentence re-search:: A regular expression search.
594 @code{forward-paragraph}: a Goldmine of Functions
596 * forward-paragraph in brief:: Key parts of the function definition.
597 * fwd-para let:: The @code{let*} expression.
598 * fwd-para while:: The forward motion @code{while} loop.
600 Counting: Repetition and Regexps
603 * @value{COUNT-WORDS}:: Use a regexp, but find a problem.
604 * recursive-count-words:: Start with case of no words in region.
605 * Counting Exercise::
607 The @code{@value{COUNT-WORDS}} Function
609 * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop.
610 * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}.
612 Counting Words in a @code{defun}
614 * Divide and Conquer::
615 * Words and Symbols:: What to count?
616 * Syntax:: What constitutes a word or symbol?
617 * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}.
618 * Several defuns:: Counting several defuns in a file.
619 * Find a File:: Do you want to look at a file?
620 * lengths-list-file:: A list of the lengths of many definitions.
621 * Several files:: Counting in definitions in different files.
622 * Several files recursively:: Recursively counting in different files.
623 * Prepare the data:: Prepare the data for display in a graph.
625 Count Words in @code{defuns} in Different Files
627 * lengths-list-many-files:: Return a list of the lengths of defuns.
628 * append:: Attach one list to another.
630 Prepare the Data for Display in a Graph
632 * Data for Display in Detail::
633 * Sorting:: Sorting lists.
634 * Files List:: Making a list of files.
635 * Counting function definitions::
639 * Columns of a graph::
640 * graph-body-print:: How to print the body of a graph.
641 * recursive-graph-body-print::
643 * Line Graph Exercise::
645 Your @file{.emacs} File
647 * Default Configuration::
648 * Site-wide Init:: You can write site-wide init files.
649 * defcustom:: Emacs will write code for you.
650 * Beginning a .emacs File:: How to write a @code{.emacs file}.
651 * Text and Auto-fill:: Automatically wrap lines.
652 * Mail Aliases:: Use abbreviations for email addresses.
653 * Indent Tabs Mode:: Don't use tabs with @TeX{}
654 * Keybindings:: Create some personal keybindings.
655 * Keymaps:: More about key binding.
656 * Loading Files:: Load (i.e., evaluate) files automatically.
657 * Autoload:: Make functions available.
658 * Simple Extension:: Define a function; bind it to a key.
659 * X11 Colors:: Colors in X.
661 * Mode Line:: How to customize your mode line.
665 * debug:: How to use the built-in debugger.
666 * debug-on-entry:: Start debugging when you call a function.
667 * debug-on-quit:: Start debugging when you quit with @kbd{C-g}.
668 * edebug:: How to use Edebug, a source level debugger.
669 * Debugging Exercises::
671 Handling the Kill Ring
673 * What the Kill Ring Does::
675 * yank:: Paste a copy of a clipped element.
676 * yank-pop:: Insert element pointed to.
679 The @code{current-kill} Function
681 * Code for current-kill::
682 * Understanding current-kill::
684 @code{current-kill} in Outline
686 * Body of current-kill::
687 * Digression concerning error:: How to mislead humans, but not computers.
688 * Determining the Element::
690 A Graph with Labeled Axes
693 * print-graph Varlist:: @code{let} expression in @code{print-graph}.
694 * print-Y-axis:: Print a label for the vertical axis.
695 * print-X-axis:: Print a horizontal label.
696 * Print Whole Graph:: The function to print a complete graph.
698 The @code{print-Y-axis} Function
700 * print-Y-axis in Detail::
701 * Height of label:: What height for the Y axis?
702 * Compute a Remainder:: How to compute the remainder of a division.
703 * Y Axis Element:: Construct a line for the Y axis.
704 * Y-axis-column:: Generate a list of Y axis labels.
705 * print-Y-axis Penultimate:: A not quite final version.
707 The @code{print-X-axis} Function
709 * Similarities differences:: Much like @code{print-Y-axis}, but not exactly.
710 * X Axis Tic Marks:: Create tic marks for the horizontal axis.
712 Printing the Whole Graph
714 * The final version:: A few changes.
715 * Test print-graph:: Run a short test.
716 * Graphing words in defuns:: Executing the final code.
717 * lambda:: How to write an anonymous function.
718 * mapcar:: Apply a function to elements of a list.
719 * Another Bug:: Yet another bug @dots{} most insidious.
720 * Final printed graph:: The graph itself!
728 Most of the GNU Emacs integrated environment is written in the programming
729 language called Emacs Lisp. The code written in this programming
730 language is the software---the sets of instructions---that tell the
731 computer what to do when you give it commands. Emacs is designed so
732 that you can write new code in Emacs Lisp and easily install it as an
733 extension to the editor.
735 (GNU Emacs is sometimes called an ``extensible editor'', but it does
736 much more than provide editing capabilities. It is better to refer to
737 Emacs as an ``extensible computing environment''. However, that
738 phrase is quite a mouthful. It is easier to refer to Emacs simply as
739 an editor. Moreover, everything you do in Emacs---find the Mayan date
740 and phases of the moon, simplify polynomials, debug code, manage
741 files, read letters, write books---all these activities are kinds of
742 editing in the most general sense of the word.)
745 * Why:: Why learn Emacs Lisp?
746 * On Reading this Text:: Read, gain familiarity, pick up habits....
747 * Who You Are:: For whom this is written.
749 * Note for Novices:: You can read this as a novice.
755 @unnumberedsec Why Study Emacs Lisp?
758 Although Emacs Lisp is usually thought of in association only with Emacs,
759 it is a full computer programming language. You can use Emacs Lisp as
760 you would any other programming language.
762 Perhaps you want to understand programming; perhaps you want to extend
763 Emacs; or perhaps you want to become a programmer. This introduction to
764 Emacs Lisp is designed to get you started: to guide you in learning the
765 fundamentals of programming, and more importantly, to show you how you
766 can teach yourself to go further.
768 @node On Reading this Text
769 @unnumberedsec On Reading this Text
771 All through this document, you will see little sample programs you can
772 run inside of Emacs. If you read this document in Info inside of GNU
773 Emacs, you can run the programs as they appear. (This is easy to do and
774 is explained when the examples are presented.) Alternatively, you can
775 read this introduction as a printed book while sitting beside a computer
776 running Emacs. (This is what I like to do; I like printed books.) If
777 you don't have a running Emacs beside you, you can still read this book,
778 but in this case, it is best to treat it as a novel or as a travel guide
779 to a country not yet visited: interesting, but not the same as being
782 Much of this introduction is dedicated to walkthroughs or guided tours
783 of code used in GNU Emacs. These tours are designed for two purposes:
784 first, to give you familiarity with real, working code (code you use
785 every day); and, second, to give you familiarity with the way Emacs
786 works. It is interesting to see how a working environment is
789 hope that you will pick up the habit of browsing through source code.
790 You can learn from it and mine it for ideas. Having GNU Emacs is like
791 having a dragon's cave of treasures.
793 In addition to learning about Emacs as an editor and Emacs Lisp as a
794 programming language, the examples and guided tours will give you an
795 opportunity to get acquainted with Emacs as a Lisp programming
796 environment. GNU Emacs supports programming and provides tools that
797 you will want to become comfortable using, such as @kbd{M-.} (the key
798 which invokes the @code{find-tag} command). You will also learn about
799 buffers and other objects that are part of the environment.
800 Learning about these features of Emacs is like learning new routes
801 around your home town.
804 In addition, I have written several programs as extended examples.
805 Although these are examples, the programs are real. I use them.
806 Other people use them. You may use them. Beyond the fragments of
807 programs used for illustrations, there is very little in here that is
808 `just for teaching purposes'; what you see is used. This is a great
809 advantage of Emacs Lisp: it is easy to learn to use it for work.
812 Finally, I hope to convey some of the skills for using Emacs to
813 learn aspects of programming that you don't know. You can often use
814 Emacs to help you understand what puzzles you or to find out how to do
815 something new. This self-reliance is not only a pleasure, but an
819 @unnumberedsec For Whom This is Written
821 This text is written as an elementary introduction for people who are
822 not programmers. If you are a programmer, you may not be satisfied with
823 this primer. The reason is that you may have become expert at reading
824 reference manuals and be put off by the way this text is organized.
826 An expert programmer who reviewed this text said to me:
829 @i{I prefer to learn from reference manuals. I ``dive into'' each
830 paragraph, and ``come up for air'' between paragraphs.}
832 @i{When I get to the end of a paragraph, I assume that that subject is
833 done, finished, that I know everything I need (with the
834 possible exception of the case when the next paragraph starts talking
835 about it in more detail). I expect that a well written reference manual
836 will not have a lot of redundancy, and that it will have excellent
837 pointers to the (one) place where the information I want is.}
840 This introduction is not written for this person!
842 Firstly, I try to say everything at least three times: first, to
843 introduce it; second, to show it in context; and third, to show it in a
844 different context, or to review it.
846 Secondly, I hardly ever put all the information about a subject in one
847 place, much less in one paragraph. To my way of thinking, that imposes
848 too heavy a burden on the reader. Instead I try to explain only what
849 you need to know at the time. (Sometimes I include a little extra
850 information so you won't be surprised later when the additional
851 information is formally introduced.)
853 When you read this text, you are not expected to learn everything the
854 first time. Frequently, you need only make, as it were, a `nodding
855 acquaintance' with some of the items mentioned. My hope is that I have
856 structured the text and given you enough hints that you will be alert to
857 what is important, and concentrate on it.
859 You will need to ``dive into'' some paragraphs; there is no other way
860 to read them. But I have tried to keep down the number of such
861 paragraphs. This book is intended as an approachable hill, rather than
862 as a daunting mountain.
864 This introduction to @cite{Programming in Emacs Lisp} has a companion
867 @cite{The GNU Emacs Lisp Reference Manual}.
870 @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU
871 Emacs Lisp Reference Manual}.
873 The reference manual has more detail than this introduction. In the
874 reference manual, all the information about one topic is concentrated
875 in one place. You should turn to it if you are like the programmer
876 quoted above. And, of course, after you have read this
877 @cite{Introduction}, you will find the @cite{Reference Manual} useful
878 when you are writing your own programs.
881 @unnumberedsec Lisp History
884 Lisp was first developed in the late 1950s at the Massachusetts
885 Institute of Technology for research in artificial intelligence. The
886 great power of the Lisp language makes it superior for other purposes as
887 well, such as writing editor commands and integrated environments.
891 GNU Emacs Lisp is largely inspired by Maclisp, which was written at MIT
892 in the 1960s. It is somewhat inspired by Common Lisp, which became a
893 standard in the 1980s. However, Emacs Lisp is much simpler than Common
894 Lisp. (The standard Emacs distribution contains an optional extensions
895 file, @file{cl.el}, that adds many Common Lisp features to Emacs Lisp.)
897 @node Note for Novices
898 @unnumberedsec A Note for Novices
900 If you don't know GNU Emacs, you can still read this document
901 profitably. However, I recommend you learn Emacs, if only to learn to
902 move around your computer screen. You can teach yourself how to use
903 Emacs with the on-line tutorial. To use it, type @kbd{C-h t}. (This
904 means you press and release the @key{CTRL} key and the @kbd{h} at the
905 same time, and then press and release @kbd{t}.)
907 Also, I often refer to one of Emacs's standard commands by listing the
908 keys which you press to invoke the command and then giving the name of
909 the command in parentheses, like this: @kbd{M-C-\}
910 (@code{indent-region}). What this means is that the
911 @code{indent-region} command is customarily invoked by typing
912 @kbd{M-C-\}. (You can, if you wish, change the keys that are typed to
913 invoke the command; this is called @dfn{rebinding}. @xref{Keymaps, ,
914 Keymaps}.) The abbreviation @kbd{M-C-\} means that you type your
915 @key{META} key, @key{CTRL} key and @key{\} key all at the same time.
916 (On many modern keyboards the @key{META} key is labeled
918 Sometimes a combination like this is called a keychord, since it is
919 similar to the way you play a chord on a piano. If your keyboard does
920 not have a @key{META} key, the @key{ESC} key prefix is used in place
921 of it. In this case, @kbd{M-C-\} means that you press and release your
922 @key{ESC} key and then type the @key{CTRL} key and the @key{\} key at
923 the same time. But usually @kbd{M-C-\} means press the @key{CTRL} key
924 along with the key that is labeled @key{ALT} and, at the same time,
925 press the @key{\} key.
927 In addition to typing a lone keychord, you can prefix what you type
928 with @kbd{C-u}, which is called the `universal argument'. The
929 @kbd{C-u} keychord passes an argument to the subsequent command.
930 Thus, to indent a region of plain text by 6 spaces, mark the region,
931 and then type @w{@kbd{C-u 6 M-C-\}}. (If you do not specify a number,
932 Emacs either passes the number 4 to the command or otherwise runs the
933 command differently than it would otherwise.) @xref{Arguments, ,
934 Numeric Arguments, emacs, The GNU Emacs Manual}.
936 If you are reading this in Info using GNU Emacs, you can read through
937 this whole document just by pressing the space bar, @key{SPC}.
938 (To learn about Info, type @kbd{C-h i} and then select Info.)
940 A note on terminology: when I use the word Lisp alone, I often am
941 referring to the various dialects of Lisp in general, but when I speak
942 of Emacs Lisp, I am referring to GNU Emacs Lisp in particular.
945 @unnumberedsec Thank You
947 My thanks to all who helped me with this book. My especial thanks to
948 @r{Jim Blandy}, @r{Noah Friedman}, @w{Jim Kingdon}, @r{Roland
949 McGrath}, @w{Frank Ritter}, @w{Randy Smith}, @w{Richard M.
950 Stallman}, and @w{Melissa Weisshaus}. My thanks also go to both
951 @w{Philip Johnson} and @w{David Stampe} for their patient
952 encouragement. My mistakes are my own.
959 @c ================ Beginning of main text ================
961 @c Start main text on right-hand (verso) page
964 \par\vfill\supereject
967 \par\vfill\supereject
969 \par\vfill\supereject
971 \par\vfill\supereject
975 @c Note: this resetting of the page number back to 1 causes TeX to gripe
976 @c about already having seen page numbers 1-4 before (in the preface):
977 @c pdfTeX warning (ext4): destination with the same identifier (name{1})
978 @c has been already used, duplicate ignored
979 @c I guess that is harmless (what happens if a later part of the text
980 @c makes a link to something in the first 4 pages though?).
981 @c E.g., note that the Emacs manual has a preface, but does not bother
982 @c resetting the page numbers back to 1 after that.
985 @evenheading @thispage @| @| @thischapter
986 @oddheading @thissection @| @| @thispage
990 @node List Processing
991 @chapter List Processing
993 To the untutored eye, Lisp is a strange programming language. In Lisp
994 code there are parentheses everywhere. Some people even claim that
995 the name stands for `Lots of Isolated Silly Parentheses'. But the
996 claim is unwarranted. Lisp stands for LISt Processing, and the
997 programming language handles @emph{lists} (and lists of lists) by
998 putting them between parentheses. The parentheses mark the boundaries
999 of the list. Sometimes a list is preceded by a single apostrophe or
1000 quotation mark, @samp{'}@footnote{The single apostrophe or quotation
1001 mark is an abbreviation for the function @code{quote}; you need not
1002 think about functions now; functions are defined in @ref{Making
1003 Errors, , Generate an Error Message}.} Lists are the basis of Lisp.
1006 * Lisp Lists:: What are lists?
1007 * Run a Program:: Any list in Lisp is a program ready to run.
1008 * Making Errors:: Generating an error message.
1009 * Names & Definitions:: Names of symbols and function definitions.
1010 * Lisp Interpreter:: What the Lisp interpreter does.
1011 * Evaluation:: Running a program.
1012 * Variables:: Returning a value from a variable.
1013 * Arguments:: Passing information to a function.
1014 * set & setq:: Setting the value of a variable.
1015 * Summary:: The major points.
1016 * Error Message Exercises::
1023 In Lisp, a list looks like this: @code{'(rose violet daisy buttercup)}.
1024 This list is preceded by a single apostrophe. It could just as well be
1025 written as follows, which looks more like the kind of list you are likely
1026 to be familiar with:
1038 The elements of this list are the names of the four different flowers,
1039 separated from each other by whitespace and surrounded by parentheses,
1040 like flowers in a field with a stone wall around them.
1041 @cindex Flowers in a field
1044 * Numbers Lists:: List have numbers, other lists, in them.
1045 * Lisp Atoms:: Elemental entities.
1046 * Whitespace in Lists:: Formatting lists to be readable.
1047 * Typing Lists:: How GNU Emacs helps you type lists.
1052 @unnumberedsubsec Numbers, Lists inside of Lists
1055 Lists can also have numbers in them, as in this list: @code{(+ 2 2)}.
1056 This list has a plus-sign, @samp{+}, followed by two @samp{2}s, each
1057 separated by whitespace.
1059 In Lisp, both data and programs are represented the same way; that is,
1060 they are both lists of words, numbers, or other lists, separated by
1061 whitespace and surrounded by parentheses. (Since a program looks like
1062 data, one program may easily serve as data for another; this is a very
1063 powerful feature of Lisp.) (Incidentally, these two parenthetical
1064 remarks are @emph{not} Lisp lists, because they contain @samp{;} and
1065 @samp{.} as punctuation marks.)
1068 Here is another list, this time with a list inside of it:
1071 '(this list has (a list inside of it))
1074 The components of this list are the words @samp{this}, @samp{list},
1075 @samp{has}, and the list @samp{(a list inside of it)}. The interior
1076 list is made up of the words @samp{a}, @samp{list}, @samp{inside},
1077 @samp{of}, @samp{it}.
1080 @subsection Lisp Atoms
1083 In Lisp, what we have been calling words are called @dfn{atoms}. This
1084 term comes from the historical meaning of the word atom, which means
1085 `indivisible'. As far as Lisp is concerned, the words we have been
1086 using in the lists cannot be divided into any smaller parts and still
1087 mean the same thing as part of a program; likewise with numbers and
1088 single character symbols like @samp{+}. On the other hand, unlike an
1089 ancient atom, a list can be split into parts. (@xref{car cdr & cons,
1090 , @code{car} @code{cdr} & @code{cons} Fundamental Functions}.)
1092 In a list, atoms are separated from each other by whitespace. They can be
1093 right next to a parenthesis.
1095 @cindex @samp{empty list} defined
1096 Technically speaking, a list in Lisp consists of parentheses surrounding
1097 atoms separated by whitespace or surrounding other lists or surrounding
1098 both atoms and other lists. A list can have just one atom in it or
1099 have nothing in it at all. A list with nothing in it looks like this:
1100 @code{()}, and is called the @dfn{empty list}. Unlike anything else, an
1101 empty list is considered both an atom and a list at the same time.
1103 @cindex Symbolic expressions, introduced
1104 @cindex @samp{expression} defined
1105 @cindex @samp{form} defined
1106 The printed representation of both atoms and lists are called
1107 @dfn{symbolic expressions} or, more concisely, @dfn{s-expressions}.
1108 The word @dfn{expression} by itself can refer to either the printed
1109 representation, or to the atom or list as it is held internally in the
1110 computer. Often, people use the term @dfn{expression}
1111 indiscriminately. (Also, in many texts, the word @dfn{form} is used
1112 as a synonym for expression.)
1114 Incidentally, the atoms that make up our universe were named such when
1115 they were thought to be indivisible; but it has been found that physical
1116 atoms are not indivisible. Parts can split off an atom or it can
1117 fission into two parts of roughly equal size. Physical atoms were named
1118 prematurely, before their truer nature was found. In Lisp, certain
1119 kinds of atom, such as an array, can be separated into parts; but the
1120 mechanism for doing this is different from the mechanism for splitting a
1121 list. As far as list operations are concerned, the atoms of a list are
1124 As in English, the meanings of the component letters of a Lisp atom
1125 are different from the meaning the letters make as a word. For
1126 example, the word for the South American sloth, the @samp{ai}, is
1127 completely different from the two words, @samp{a}, and @samp{i}.
1129 There are many kinds of atom in nature but only a few in Lisp: for
1130 example, @dfn{numbers}, such as 37, 511, or 1729, and @dfn{symbols}, such
1131 as @samp{+}, @samp{foo}, or @samp{forward-line}. The words we have
1132 listed in the examples above are all symbols. In everyday Lisp
1133 conversation, the word ``atom'' is not often used, because programmers
1134 usually try to be more specific about what kind of atom they are dealing
1135 with. Lisp programming is mostly about symbols (and sometimes numbers)
1136 within lists. (Incidentally, the preceding three word parenthetical
1137 remark is a proper list in Lisp, since it consists of atoms, which in
1138 this case are symbols, separated by whitespace and enclosed by
1139 parentheses, without any non-Lisp punctuation.)
1142 Text between double quotation marks---even sentences or
1143 paragraphs---is also an atom. Here is an example:
1144 @cindex Text between double quotation marks
1147 '(this list includes "text between quotation marks.")
1150 @cindex @samp{string} defined
1152 In Lisp, all of the quoted text including the punctuation mark and the
1153 blank spaces is a single atom. This kind of atom is called a
1154 @dfn{string} (for `string of characters') and is the sort of thing that
1155 is used for messages that a computer can print for a human to read.
1156 Strings are a different kind of atom than numbers or symbols and are
1159 @node Whitespace in Lists
1160 @subsection Whitespace in Lists
1161 @cindex Whitespace in lists
1164 The amount of whitespace in a list does not matter. From the point of view
1165 of the Lisp language,
1176 is exactly the same as this:
1179 '(this list looks like this)
1182 Both examples show what to Lisp is the same list, the list made up of
1183 the symbols @samp{this}, @samp{list}, @samp{looks}, @samp{like}, and
1184 @samp{this} in that order.
1186 Extra whitespace and newlines are designed to make a list more readable
1187 by humans. When Lisp reads the expression, it gets rid of all the extra
1188 whitespace (but it needs to have at least one space between atoms in
1189 order to tell them apart.)
1191 Odd as it seems, the examples we have seen cover almost all of what Lisp
1192 lists look like! Every other list in Lisp looks more or less like one
1193 of these examples, except that the list may be longer and more complex.
1194 In brief, a list is between parentheses, a string is between quotation
1195 marks, a symbol looks like a word, and a number looks like a number.
1196 (For certain situations, square brackets, dots and a few other special
1197 characters may be used; however, we will go quite far without them.)
1200 @subsection GNU Emacs Helps You Type Lists
1201 @cindex Help typing lists
1202 @cindex Formatting help
1204 When you type a Lisp expression in GNU Emacs using either Lisp
1205 Interaction mode or Emacs Lisp mode, you have available to you several
1206 commands to format the Lisp expression so it is easy to read. For
1207 example, pressing the @key{TAB} key automatically indents the line the
1208 cursor is on by the right amount. A command to properly indent the
1209 code in a region is customarily bound to @kbd{M-C-\}. Indentation is
1210 designed so that you can see which elements of a list belong to which
1211 list---elements of a sub-list are indented more than the elements of
1214 In addition, when you type a closing parenthesis, Emacs momentarily
1215 jumps the cursor back to the matching opening parenthesis, so you can
1216 see which one it is. This is very useful, since every list you type
1217 in Lisp must have its closing parenthesis match its opening
1218 parenthesis. (@xref{Major Modes, , Major Modes, emacs, The GNU Emacs
1219 Manual}, for more information about Emacs's modes.)
1222 @section Run a Program
1223 @cindex Run a program
1224 @cindex Program, running one
1226 @cindex @samp{evaluate} defined
1227 A list in Lisp---any list---is a program ready to run. If you run it
1228 (for which the Lisp jargon is @dfn{evaluate}), the computer will do one
1229 of three things: do nothing except return to you the list itself; send
1230 you an error message; or, treat the first symbol in the list as a
1231 command to do something. (Usually, of course, it is the last of these
1232 three things that you really want!)
1234 @c use code for the single apostrophe, not samp.
1235 The single apostrophe, @code{'}, that I put in front of some of the
1236 example lists in preceding sections is called a @dfn{quote}; when it
1237 precedes a list, it tells Lisp to do nothing with the list, other than
1238 take it as it is written. But if there is no quote preceding a list,
1239 the first item of the list is special: it is a command for the computer
1240 to obey. (In Lisp, these commands are called @emph{functions}.) The list
1241 @code{(+ 2 2)} shown above did not have a quote in front of it, so Lisp
1242 understands that the @code{+} is an instruction to do something with the
1243 rest of the list: add the numbers that follow.
1246 If you are reading this inside of GNU Emacs in Info, here is how you can
1247 evaluate such a list: place your cursor immediately after the right
1248 hand parenthesis of the following list and then type @kbd{C-x C-e}:
1254 @c use code for the number four, not samp.
1256 You will see the number @code{4} appear in the echo area. (In the
1257 jargon, what you have just done is ``evaluate the list.'' The echo area
1258 is the line at the bottom of the screen that displays or ``echoes''
1259 text.) Now try the same thing with a quoted list: place the cursor
1260 right after the following list and type @kbd{C-x C-e}:
1263 '(this is a quoted list)
1267 You will see @code{(this is a quoted list)} appear in the echo area.
1269 @cindex Lisp interpreter, explained
1270 @cindex Interpreter, Lisp, explained
1271 In both cases, what you are doing is giving a command to the program
1272 inside of GNU Emacs called the @dfn{Lisp interpreter}---giving the
1273 interpreter a command to evaluate the expression. The name of the Lisp
1274 interpreter comes from the word for the task done by a human who comes
1275 up with the meaning of an expression---who ``interprets'' it.
1277 You can also evaluate an atom that is not part of a list---one that is
1278 not surrounded by parentheses; again, the Lisp interpreter translates
1279 from the humanly readable expression to the language of the computer.
1280 But before discussing this (@pxref{Variables}), we will discuss what the
1281 Lisp interpreter does when you make an error.
1284 @section Generate an Error Message
1285 @cindex Generate an error message
1286 @cindex Error message generation
1288 Partly so you won't worry if you do it accidentally, we will now give
1289 a command to the Lisp interpreter that generates an error message.
1290 This is a harmless activity; and indeed, we will often try to generate
1291 error messages intentionally. Once you understand the jargon, error
1292 messages can be informative. Instead of being called ``error''
1293 messages, they should be called ``help'' messages. They are like
1294 signposts to a traveler in a strange country; deciphering them can be
1295 hard, but once understood, they can point the way.
1297 The error message is generated by a built-in GNU Emacs debugger. We
1298 will `enter the debugger'. You get out of the debugger by typing @code{q}.
1300 What we will do is evaluate a list that is not quoted and does not
1301 have a meaningful command as its first element. Here is a list almost
1302 exactly the same as the one we just used, but without the single-quote
1303 in front of it. Position the cursor right after it and type @kbd{C-x
1307 (this is an unquoted list)
1312 What you see depends on which version of Emacs you are running. GNU
1313 Emacs version 22 provides more information than version 20 and before.
1314 First, the more recent result of generating an error; then the
1315 earlier, version 20 result.
1319 In GNU Emacs version 22, a @file{*Backtrace*} window will open up and
1320 you will see the following in it:
1323 A @file{*Backtrace*} window will open up and you should see the
1328 ---------- Buffer: *Backtrace* ----------
1329 Debugger entered--Lisp error: (void-function this)
1330 (this is an unquoted list)
1331 eval((this is an unquoted list))
1332 eval-last-sexp-1(nil)
1334 call-interactively(eval-last-sexp)
1335 ---------- Buffer: *Backtrace* ----------
1341 Your cursor will be in this window (you may have to wait a few seconds
1342 before it becomes visible). To quit the debugger and make the
1343 debugger window go away, type:
1350 Please type @kbd{q} right now, so you become confident that you can
1351 get out of the debugger. Then, type @kbd{C-x C-e} again to re-enter
1354 @cindex @samp{function} defined
1355 Based on what we already know, we can almost read this error message.
1357 You read the @file{*Backtrace*} buffer from the bottom up; it tells
1358 you what Emacs did. When you typed @kbd{C-x C-e}, you made an
1359 interactive call to the command @code{eval-last-sexp}. @code{eval} is
1360 an abbreviation for `evaluate' and @code{sexp} is an abbreviation for
1361 `symbolic expression'. The command means `evaluate last symbolic
1362 expression', which is the expression just before your cursor.
1364 Each line above tells you what the Lisp interpreter evaluated next.
1365 The most recent action is at the top. The buffer is called the
1366 @file{*Backtrace*} buffer because it enables you to track Emacs
1370 At the top of the @file{*Backtrace*} buffer, you see the line:
1373 Debugger entered--Lisp error: (void-function this)
1377 The Lisp interpreter tried to evaluate the first atom of the list, the
1378 word @samp{this}. It is this action that generated the error message
1379 @samp{void-function this}.
1381 The message contains the words @samp{void-function} and @samp{this}.
1383 @cindex @samp{function} defined
1384 The word @samp{function} was mentioned once before. It is a very
1385 important word. For our purposes, we can define it by saying that a
1386 @dfn{function} is a set of instructions to the computer that tell the
1387 computer to do something.
1389 Now we can begin to understand the error message: @samp{void-function
1390 this}. The function (that is, the word @samp{this}) does not have a
1391 definition of any set of instructions for the computer to carry out.
1393 The slightly odd word, @samp{void-function}, is designed to cover the
1394 way Emacs Lisp is implemented, which is that when a symbol does not
1395 have a function definition attached to it, the place that should
1396 contain the instructions is `void'.
1398 On the other hand, since we were able to add 2 plus 2 successfully, by
1399 evaluating @code{(+ 2 2)}, we can infer that the symbol @code{+} must
1400 have a set of instructions for the computer to obey and those
1401 instructions must be to add the numbers that follow the @code{+}.
1403 It is possible to prevent Emacs entering the debugger in cases like
1404 this. We do not explain how to do that here, but we will mention what
1405 the result looks like, because you may encounter a similar situation
1406 if there is a bug in some Emacs code that you are using. In such
1407 cases, you will see only one line of error message; it will appear in
1408 the echo area and look like this:
1411 Symbol's function definition is void:@: this
1416 (Also, your terminal may beep at you---some do, some don't; and others
1417 blink. This is just a device to get your attention.)
1419 The message goes away as soon as you type a key, even just to
1422 We know the meaning of the word @samp{Symbol}. It refers to the first
1423 atom of the list, the word @samp{this}. The word @samp{function}
1424 refers to the instructions that tell the computer what to do.
1425 (Technically, the symbol tells the computer where to find the
1426 instructions, but this is a complication we can ignore for the
1429 The error message can be understood: @samp{Symbol's function
1430 definition is void:@: this}. The symbol (that is, the word
1431 @samp{this}) lacks instructions for the computer to carry out.
1433 @node Names & Definitions
1434 @section Symbol Names and Function Definitions
1435 @cindex Symbol names
1437 We can articulate another characteristic of Lisp based on what we have
1438 discussed so far---an important characteristic: a symbol, like
1439 @code{+}, is not itself the set of instructions for the computer to
1440 carry out. Instead, the symbol is used, perhaps temporarily, as a way
1441 of locating the definition or set of instructions. What we see is the
1442 name through which the instructions can be found. Names of people
1443 work the same way. I can be referred to as @samp{Bob}; however, I am
1444 not the letters @samp{B}, @samp{o}, @samp{b} but am, or was, the
1445 consciousness consistently associated with a particular life-form.
1446 The name is not me, but it can be used to refer to me.
1448 In Lisp, one set of instructions can be attached to several names.
1449 For example, the computer instructions for adding numbers can be
1450 linked to the symbol @code{plus} as well as to the symbol @code{+}
1451 (and are in some dialects of Lisp). Among humans, I can be referred
1452 to as @samp{Robert} as well as @samp{Bob} and by other words as well.
1454 On the other hand, a symbol can have only one function definition
1455 attached to it at a time. Otherwise, the computer would be confused as
1456 to which definition to use. If this were the case among people, only
1457 one person in the world could be named @samp{Bob}. However, the function
1458 definition to which the name refers can be changed readily.
1459 (@xref{Install, , Install a Function Definition}.)
1461 Since Emacs Lisp is large, it is customary to name symbols in a way
1462 that identifies the part of Emacs to which the function belongs.
1463 Thus, all the names for functions that deal with Texinfo start with
1464 @samp{texinfo-} and those for functions that deal with reading mail
1465 start with @samp{rmail-}.
1467 @node Lisp Interpreter
1468 @section The Lisp Interpreter
1469 @cindex Lisp interpreter, what it does
1470 @cindex Interpreter, what it does
1472 Based on what we have seen, we can now start to figure out what the
1473 Lisp interpreter does when we command it to evaluate a list.
1474 First, it looks to see whether there is a quote before the list; if
1475 there is, the interpreter just gives us the list. On the other
1476 hand, if there is no quote, the interpreter looks at the first element
1477 in the list and sees whether it has a function definition. If it does,
1478 the interpreter carries out the instructions in the function definition.
1479 Otherwise, the interpreter prints an error message.
1481 This is how Lisp works. Simple. There are added complications which we
1482 will get to in a minute, but these are the fundamentals. Of course, to
1483 write Lisp programs, you need to know how to write function definitions
1484 and attach them to names, and how to do this without confusing either
1485 yourself or the computer.
1488 * Complications:: Variables, Special forms, Lists within.
1489 * Byte Compiling:: Specially processing code for speed.
1494 @unnumberedsubsec Complications
1497 Now, for the first complication. In addition to lists, the Lisp
1498 interpreter can evaluate a symbol that is not quoted and does not have
1499 parentheses around it. The Lisp interpreter will attempt to determine
1500 the symbol's value as a @dfn{variable}. This situation is described
1501 in the section on variables. (@xref{Variables}.)
1503 @cindex Special form
1504 The second complication occurs because some functions are unusual and
1505 do not work in the usual manner. Those that don't are called
1506 @dfn{special forms}. They are used for special jobs, like defining a
1507 function, and there are not many of them. In the next few chapters,
1508 you will be introduced to several of the more important special forms.
1510 As well as special forms, there are also @dfn{macros}. A macro
1511 is a construct defined in Lisp, which differs from a function in that it
1512 translates a Lisp expression into another expression that is to be
1513 evaluated in place of the original expression. (@xref{Lisp macro}.)
1515 For the purposes of this introduction, you do not need to worry too much
1516 about whether something is a special form, macro, or ordinary function.
1517 For example, @code{if} is a special form (@pxref{if}), but @code{when}
1518 is a macro (@pxref{Lisp macro}). In earlier versions of Emacs,
1519 @code{defun} was a special form, but now it is a macro (@pxref{defun}).
1520 It still behaves in the same way.
1522 The final complication is this: if the function that the
1523 Lisp interpreter is looking at is not a special form, and if it is part
1524 of a list, the Lisp interpreter looks to see whether the list has a list
1525 inside of it. If there is an inner list, the Lisp interpreter first
1526 figures out what it should do with the inside list, and then it works on
1527 the outside list. If there is yet another list embedded inside the
1528 inner list, it works on that one first, and so on. It always works on
1529 the innermost list first. The interpreter works on the innermost list
1530 first, to evaluate the result of that list. The result may be
1531 used by the enclosing expression.
1533 Otherwise, the interpreter works left to right, from one expression to
1536 @node Byte Compiling
1537 @subsection Byte Compiling
1538 @cindex Byte compiling
1540 One other aspect of interpreting: the Lisp interpreter is able to
1541 interpret two kinds of entity: humanly readable code, on which we will
1542 focus exclusively, and specially processed code, called @dfn{byte
1543 compiled} code, which is not humanly readable. Byte compiled code
1544 runs faster than humanly readable code.
1546 You can transform humanly readable code into byte compiled code by
1547 running one of the compile commands such as @code{byte-compile-file}.
1548 Byte compiled code is usually stored in a file that ends with a
1549 @file{.elc} extension rather than a @file{.el} extension. You will
1550 see both kinds of file in the @file{emacs/lisp} directory; the files
1551 to read are those with @file{.el} extensions.
1553 As a practical matter, for most things you might do to customize or
1554 extend Emacs, you do not need to byte compile; and I will not discuss
1555 the topic here. @xref{Byte Compilation, , Byte Compilation, elisp,
1556 The GNU Emacs Lisp Reference Manual}, for a full description of byte
1563 When the Lisp interpreter works on an expression, the term for the
1564 activity is called @dfn{evaluation}. We say that the interpreter
1565 `evaluates the expression'. I've used this term several times before.
1566 The word comes from its use in everyday language, `to ascertain the
1567 value or amount of; to appraise', according to @cite{Webster's New
1568 Collegiate Dictionary}.
1571 * How the Interpreter Acts:: Returns and Side Effects...
1572 * Evaluating Inner Lists:: Lists within lists...
1576 @node How the Interpreter Acts
1577 @unnumberedsubsec How the Lisp Interpreter Acts
1580 @cindex @samp{returned value} explained
1581 After evaluating an expression, the Lisp interpreter will most likely
1582 @dfn{return} the value that the computer produces by carrying out the
1583 instructions it found in the function definition, or perhaps it will
1584 give up on that function and produce an error message. (The interpreter
1585 may also find itself tossed, so to speak, to a different function or it
1586 may attempt to repeat continually what it is doing for ever and ever in
1587 what is called an `infinite loop'. These actions are less common; and
1588 we can ignore them.) Most frequently, the interpreter returns a value.
1590 @cindex @samp{side effect} defined
1591 At the same time the interpreter returns a value, it may do something
1592 else as well, such as move a cursor or copy a file; this other kind of
1593 action is called a @dfn{side effect}. Actions that we humans think are
1594 important, such as printing results, are often ``side effects'' to the
1595 Lisp interpreter. The jargon can sound peculiar, but it turns out that
1596 it is fairly easy to learn to use side effects.
1598 In summary, evaluating a symbolic expression most commonly causes the
1599 Lisp interpreter to return a value and perhaps carry out a side effect;
1600 or else produce an error.
1602 @node Evaluating Inner Lists
1603 @subsection Evaluating Inner Lists
1604 @cindex Inner list evaluation
1605 @cindex Evaluating inner lists
1607 If evaluation applies to a list that is inside another list, the outer
1608 list may use the value returned by the first evaluation as information
1609 when the outer list is evaluated. This explains why inner expressions
1610 are evaluated first: the values they return are used by the outer
1614 We can investigate this process by evaluating another addition example.
1615 Place your cursor after the following expression and type @kbd{C-x C-e}:
1622 The number 8 will appear in the echo area.
1624 What happens is that the Lisp interpreter first evaluates the inner
1625 expression, @code{(+ 3 3)}, for which the value 6 is returned; then it
1626 evaluates the outer expression as if it were written @code{(+ 2 6)}, which
1627 returns the value 8. Since there are no more enclosing expressions to
1628 evaluate, the interpreter prints that value in the echo area.
1630 Now it is easy to understand the name of the command invoked by the
1631 keystrokes @kbd{C-x C-e}: the name is @code{eval-last-sexp}. The
1632 letters @code{sexp} are an abbreviation for `symbolic expression', and
1633 @code{eval} is an abbreviation for `evaluate'. The command means
1634 `evaluate last symbolic expression'.
1636 As an experiment, you can try evaluating the expression by putting the
1637 cursor at the beginning of the next line immediately following the
1638 expression, or inside the expression.
1641 Here is another copy of the expression:
1648 If you place the cursor at the beginning of the blank line that
1649 immediately follows the expression and type @kbd{C-x C-e}, you will
1650 still get the value 8 printed in the echo area. Now try putting the
1651 cursor inside the expression. If you put it right after the next to
1652 last parenthesis (so it appears to sit on top of the last parenthesis),
1653 you will get a 6 printed in the echo area! This is because the command
1654 evaluates the expression @code{(+ 3 3)}.
1656 Now put the cursor immediately after a number. Type @kbd{C-x C-e} and
1657 you will get the number itself. In Lisp, if you evaluate a number, you
1658 get the number itself---this is how numbers differ from symbols. If you
1659 evaluate a list starting with a symbol like @code{+}, you will get a
1660 value returned that is the result of the computer carrying out the
1661 instructions in the function definition attached to that name. If a
1662 symbol by itself is evaluated, something different happens, as we will
1663 see in the next section.
1669 In Emacs Lisp, a symbol can have a value attached to it just as it can
1670 have a function definition attached to it. The two are different.
1671 The function definition is a set of instructions that a computer will
1672 obey. A value, on the other hand, is something, such as number or a
1673 name, that can vary (which is why such a symbol is called a variable).
1674 The value of a symbol can be any expression in Lisp, such as a symbol,
1675 number, list, or string. A symbol that has a value is often called a
1678 A symbol can have both a function definition and a value attached to
1679 it at the same time. Or it can have just one or the other.
1680 The two are separate. This is somewhat similar
1681 to the way the name Cambridge can refer to the city in Massachusetts
1682 and have some information attached to the name as well, such as
1683 ``great programming center''.
1686 (Incidentally, in Emacs Lisp, a symbol can have two
1687 other things attached to it, too: a property list and a documentation
1688 string; these are discussed later.)
1691 Another way to think about this is to imagine a symbol as being a chest
1692 of drawers. The function definition is put in one drawer, the value in
1693 another, and so on. What is put in the drawer holding the value can be
1694 changed without affecting the contents of the drawer holding the
1695 function definition, and vice-verse.
1698 * fill-column Example::
1699 * Void Function:: The error message for a symbol
1701 * Void Variable:: The error message for a symbol without a value.
1705 @node fill-column Example
1706 @unnumberedsubsec @code{fill-column}, an Example Variable
1709 @findex fill-column, @r{an example variable}
1710 @cindex Example variable, @code{fill-column}
1711 @cindex Variable, example of, @code{fill-column}
1712 The variable @code{fill-column} illustrates a symbol with a value
1713 attached to it: in every GNU Emacs buffer, this symbol is set to some
1714 value, usually 72 or 70, but sometimes to some other value. To find the
1715 value of this symbol, evaluate it by itself. If you are reading this in
1716 Info inside of GNU Emacs, you can do this by putting the cursor after
1717 the symbol and typing @kbd{C-x C-e}:
1724 After I typed @kbd{C-x C-e}, Emacs printed the number 72 in my echo
1725 area. This is the value for which @code{fill-column} is set for me as I
1726 write this. It may be different for you in your Info buffer. Notice
1727 that the value returned as a variable is printed in exactly the same way
1728 as the value returned by a function carrying out its instructions. From
1729 the point of view of the Lisp interpreter, a value returned is a value
1730 returned. What kind of expression it came from ceases to matter once
1733 A symbol can have any value attached to it or, to use the jargon, we can
1734 @dfn{bind} the variable to a value: to a number, such as 72; to a
1735 string, @code{"such as this"}; to a list, such as @code{(spruce pine
1736 oak)}; we can even bind a variable to a function definition.
1738 A symbol can be bound to a value in several ways. @xref{set & setq, ,
1739 Setting the Value of a Variable}, for information about one way to do
1743 @subsection Error Message for a Symbol Without a Function
1744 @cindex Symbol without function error
1745 @cindex Error for symbol without function
1747 When we evaluated @code{fill-column} to find its value as a variable,
1748 we did not place parentheses around the word. This is because we did
1749 not intend to use it as a function name.
1751 If @code{fill-column} were the first or only element of a list, the
1752 Lisp interpreter would attempt to find the function definition
1753 attached to it. But @code{fill-column} has no function definition.
1754 Try evaluating this:
1762 You will create a @file{*Backtrace*} buffer that says:
1766 ---------- Buffer: *Backtrace* ----------
1767 Debugger entered--Lisp error: (void-function fill-column)
1770 eval-last-sexp-1(nil)
1772 call-interactively(eval-last-sexp)
1773 ---------- Buffer: *Backtrace* ----------
1778 (Remember, to quit the debugger and make the debugger window go away,
1779 type @kbd{q} in the @file{*Backtrace*} buffer.)
1783 In GNU Emacs 20 and before, you will produce an error message that says:
1786 Symbol's function definition is void:@: fill-column
1790 (The message will go away as soon as you move the cursor or type
1795 @subsection Error Message for a Symbol Without a Value
1796 @cindex Symbol without value error
1797 @cindex Error for symbol without value
1799 If you attempt to evaluate a symbol that does not have a value bound to
1800 it, you will receive an error message. You can see this by
1801 experimenting with our 2 plus 2 addition. In the following expression,
1802 put your cursor right after the @code{+}, before the first number 2,
1811 In GNU Emacs 22, you will create a @file{*Backtrace*} buffer that
1816 ---------- Buffer: *Backtrace* ----------
1817 Debugger entered--Lisp error: (void-variable +)
1819 eval-last-sexp-1(nil)
1821 call-interactively(eval-last-sexp)
1822 ---------- Buffer: *Backtrace* ----------
1827 (Again, you can quit the debugger by
1828 typing @kbd{q} in the @file{*Backtrace*} buffer.)
1830 This backtrace is different from the very first error message we saw,
1831 which said, @samp{Debugger entered--Lisp error: (void-function this)}.
1832 In this case, the function does not have a value as a variable; while
1833 in the other error message, the function (the word `this') did not
1836 In this experiment with the @code{+}, what we did was cause the Lisp
1837 interpreter to evaluate the @code{+} and look for the value of the
1838 variable instead of the function definition. We did this by placing the
1839 cursor right after the symbol rather than after the parenthesis of the
1840 enclosing list as we did before. As a consequence, the Lisp interpreter
1841 evaluated the preceding s-expression, which in this case was
1844 Since @code{+} does not have a value bound to it, just the function
1845 definition, the error message reported that the symbol's value as a
1850 In GNU Emacs version 20 and before, your error message will say:
1853 Symbol's value as variable is void:@: +
1857 The meaning is the same as in GNU Emacs 22.
1863 @cindex Passing information to functions
1865 To see how information is passed to functions, let's look again at
1866 our old standby, the addition of two plus two. In Lisp, this is written
1873 If you evaluate this expression, the number 4 will appear in your echo
1874 area. What the Lisp interpreter does is add the numbers that follow
1877 @cindex @samp{argument} defined
1878 The numbers added by @code{+} are called the @dfn{arguments} of the
1879 function @code{+}. These numbers are the information that is given to
1880 or @dfn{passed} to the function.
1882 The word `argument' comes from the way it is used in mathematics and
1883 does not refer to a disputation between two people; instead it refers to
1884 the information presented to the function, in this case, to the
1885 @code{+}. In Lisp, the arguments to a function are the atoms or lists
1886 that follow the function. The values returned by the evaluation of
1887 these atoms or lists are passed to the function. Different functions
1888 require different numbers of arguments; some functions require none at
1889 all.@footnote{It is curious to track the path by which the word `argument'
1890 came to have two different meanings, one in mathematics and the other in
1891 everyday English. According to the @cite{Oxford English Dictionary},
1892 the word derives from the Latin for @samp{to make clear, prove}; thus it
1893 came to mean, by one thread of derivation, `the evidence offered as
1894 proof', which is to say, `the information offered', which led to its
1895 meaning in Lisp. But in the other thread of derivation, it came to mean
1896 `to assert in a manner against which others may make counter
1897 assertions', which led to the meaning of the word as a disputation.
1898 (Note here that the English word has two different definitions attached
1899 to it at the same time. By contrast, in Emacs Lisp, a symbol cannot
1900 have two different function definitions at the same time.)}
1903 * Data types:: Types of data passed to a function.
1904 * Args as Variable or List:: An argument can be the value
1905 of a variable or list.
1906 * Variable Number of Arguments:: Some functions may take a
1907 variable number of arguments.
1908 * Wrong Type of Argument:: Passing an argument of the wrong type
1910 * message:: A useful function for sending messages.
1914 @subsection Arguments' Data Types
1916 @cindex Types of data
1917 @cindex Arguments' data types
1919 The type of data that should be passed to a function depends on what
1920 kind of information it uses. The arguments to a function such as
1921 @code{+} must have values that are numbers, since @code{+} adds numbers.
1922 Other functions use different kinds of data for their arguments.
1926 For example, the @code{concat} function links together or unites two or
1927 more strings of text to produce a string. The arguments are strings.
1928 Concatenating the two character strings @code{abc}, @code{def} produces
1929 the single string @code{abcdef}. This can be seen by evaluating the
1933 (concat "abc" "def")
1937 The value produced by evaluating this expression is @code{"abcdef"}.
1939 A function such as @code{substring} uses both a string and numbers as
1940 arguments. The function returns a part of the string, a substring of
1941 the first argument. This function takes three arguments. Its first
1942 argument is the string of characters, the second and third arguments are
1943 numbers that indicate the beginning and end of the substring. The
1944 numbers are a count of the number of characters (including spaces and
1945 punctuation) from the beginning of the string.
1948 For example, if you evaluate the following:
1951 (substring "The quick brown fox jumped." 16 19)
1955 you will see @code{"fox"} appear in the echo area. The arguments are the
1956 string and the two numbers.
1958 Note that the string passed to @code{substring} is a single atom even
1959 though it is made up of several words separated by spaces. Lisp counts
1960 everything between the two quotation marks as part of the string,
1961 including the spaces. You can think of the @code{substring} function as
1962 a kind of `atom smasher' since it takes an otherwise indivisible atom
1963 and extracts a part. However, @code{substring} is only able to extract
1964 a substring from an argument that is a string, not from another type of
1965 atom such as a number or symbol.
1967 @node Args as Variable or List
1968 @subsection An Argument as the Value of a Variable or List
1970 An argument can be a symbol that returns a value when it is evaluated.
1971 For example, when the symbol @code{fill-column} by itself is evaluated,
1972 it returns a number. This number can be used in an addition.
1975 Position the cursor after the following expression and type @kbd{C-x
1983 The value will be a number two more than what you get by evaluating
1984 @code{fill-column} alone. For me, this is 74, because my value of
1985 @code{fill-column} is 72.
1987 As we have just seen, an argument can be a symbol that returns a value
1988 when evaluated. In addition, an argument can be a list that returns a
1989 value when it is evaluated. For example, in the following expression,
1990 the arguments to the function @code{concat} are the strings
1991 @w{@code{"The "}} and @w{@code{" red foxes."}} and the list
1992 @code{(number-to-string (+ 2 fill-column))}.
1994 @c For GNU Emacs 22, need number-to-string
1996 (concat "The " (number-to-string (+ 2 fill-column)) " red foxes.")
2000 If you evaluate this expression---and if, as with my Emacs,
2001 @code{fill-column} evaluates to 72---@code{"The 74 red foxes."} will
2002 appear in the echo area. (Note that you must put spaces after the
2003 word @samp{The} and before the word @samp{red} so they will appear in
2004 the final string. The function @code{number-to-string} converts the
2005 integer that the addition function returns to a string.
2006 @code{number-to-string} is also known as @code{int-to-string}.)
2008 @node Variable Number of Arguments
2009 @subsection Variable Number of Arguments
2010 @cindex Variable number of arguments
2011 @cindex Arguments, variable number of
2013 Some functions, such as @code{concat}, @code{+} or @code{*}, take any
2014 number of arguments. (The @code{*} is the symbol for multiplication.)
2015 This can be seen by evaluating each of the following expressions in
2016 the usual way. What you will see in the echo area is printed in this
2017 text after @samp{@result{}}, which you may read as `evaluates to'.
2020 In the first set, the functions have no arguments:
2031 In this set, the functions have one argument each:
2042 In this set, the functions have three arguments each:
2046 (+ 3 4 5) @result{} 12
2048 (* 3 4 5) @result{} 60
2052 @node Wrong Type of Argument
2053 @subsection Using the Wrong Type Object as an Argument
2054 @cindex Wrong type of argument
2055 @cindex Argument, wrong type of
2057 When a function is passed an argument of the wrong type, the Lisp
2058 interpreter produces an error message. For example, the @code{+}
2059 function expects the values of its arguments to be numbers. As an
2060 experiment we can pass it the quoted symbol @code{hello} instead of a
2061 number. Position the cursor after the following expression and type
2069 When you do this you will generate an error message. What has happened
2070 is that @code{+} has tried to add the 2 to the value returned by
2071 @code{'hello}, but the value returned by @code{'hello} is the symbol
2072 @code{hello}, not a number. Only numbers can be added. So @code{+}
2073 could not carry out its addition.
2076 You will create and enter a @file{*Backtrace*} buffer that says:
2081 ---------- Buffer: *Backtrace* ----------
2082 Debugger entered--Lisp error:
2083 (wrong-type-argument number-or-marker-p hello)
2085 eval((+ 2 (quote hello)))
2086 eval-last-sexp-1(nil)
2088 call-interactively(eval-last-sexp)
2089 ---------- Buffer: *Backtrace* ----------
2094 As usual, the error message tries to be helpful and makes sense after you
2095 learn how to read it.@footnote{@code{(quote hello)} is an expansion of
2096 the abbreviation @code{'hello}.}
2098 The first part of the error message is straightforward; it says
2099 @samp{wrong type argument}. Next comes the mysterious jargon word
2100 @w{@samp{number-or-marker-p}}. This word is trying to tell you what
2101 kind of argument the @code{+} expected.
2103 The symbol @code{number-or-marker-p} says that the Lisp interpreter is
2104 trying to determine whether the information presented it (the value of
2105 the argument) is a number or a marker (a special object representing a
2106 buffer position). What it does is test to see whether the @code{+} is
2107 being given numbers to add. It also tests to see whether the
2108 argument is something called a marker, which is a specific feature of
2109 Emacs Lisp. (In Emacs, locations in a buffer are recorded as markers.
2110 When the mark is set with the @kbd{C-@@} or @kbd{C-@key{SPC}} command,
2111 its position is kept as a marker. The mark can be considered a
2112 number---the number of characters the location is from the beginning
2113 of the buffer.) In Emacs Lisp, @code{+} can be used to add the
2114 numeric value of marker positions as numbers.
2116 The @samp{p} of @code{number-or-marker-p} is the embodiment of a
2117 practice started in the early days of Lisp programming. The @samp{p}
2118 stands for `predicate'. In the jargon used by the early Lisp
2119 researchers, a predicate refers to a function to determine whether some
2120 property is true or false. So the @samp{p} tells us that
2121 @code{number-or-marker-p} is the name of a function that determines
2122 whether it is true or false that the argument supplied is a number or
2123 a marker. Other Lisp symbols that end in @samp{p} include @code{zerop},
2124 a function that tests whether its argument has the value of zero, and
2125 @code{listp}, a function that tests whether its argument is a list.
2127 Finally, the last part of the error message is the symbol @code{hello}.
2128 This is the value of the argument that was passed to @code{+}. If the
2129 addition had been passed the correct type of object, the value passed
2130 would have been a number, such as 37, rather than a symbol like
2131 @code{hello}. But then you would not have got the error message.
2135 In GNU Emacs version 20 and before, the echo area displays an error
2139 Wrong type argument:@: number-or-marker-p, hello
2142 This says, in different words, the same as the top line of the
2143 @file{*Backtrace*} buffer.
2147 @subsection The @code{message} Function
2150 Like @code{+}, the @code{message} function takes a variable number of
2151 arguments. It is used to send messages to the user and is so useful
2152 that we will describe it here.
2155 A message is printed in the echo area. For example, you can print a
2156 message in your echo area by evaluating the following list:
2159 (message "This message appears in the echo area!")
2162 The whole string between double quotation marks is a single argument
2163 and is printed @i{in toto}. (Note that in this example, the message
2164 itself will appear in the echo area within double quotes; that is
2165 because you see the value returned by the @code{message} function. In
2166 most uses of @code{message} in programs that you write, the text will
2167 be printed in the echo area as a side-effect, without the quotes.
2168 @xref{multiply-by-seven in detail, , @code{multiply-by-seven} in
2169 detail}, for an example of this.)
2171 However, if there is a @samp{%s} in the quoted string of characters, the
2172 @code{message} function does not print the @samp{%s} as such, but looks
2173 to the argument that follows the string. It evaluates the second
2174 argument and prints the value at the location in the string where the
2178 You can see this by positioning the cursor after the following
2179 expression and typing @kbd{C-x C-e}:
2182 (message "The name of this buffer is: %s." (buffer-name))
2186 In Info, @code{"The name of this buffer is: *info*."} will appear in the
2187 echo area. The function @code{buffer-name} returns the name of the
2188 buffer as a string, which the @code{message} function inserts in place
2191 To print a value as an integer, use @samp{%d} in the same way as
2192 @samp{%s}. For example, to print a message in the echo area that
2193 states the value of the @code{fill-column}, evaluate the following:
2196 (message "The value of fill-column is %d." fill-column)
2200 On my system, when I evaluate this list, @code{"The value of
2201 fill-column is 72."} appears in my echo area@footnote{Actually, you
2202 can use @code{%s} to print a number. It is non-specific. @code{%d}
2203 prints only the part of a number left of a decimal point, and not
2204 anything that is not a number.}.
2206 If there is more than one @samp{%s} in the quoted string, the value of
2207 the first argument following the quoted string is printed at the
2208 location of the first @samp{%s} and the value of the second argument is
2209 printed at the location of the second @samp{%s}, and so on.
2212 For example, if you evaluate the following,
2216 (message "There are %d %s in the office!"
2217 (- fill-column 14) "pink elephants")
2222 a rather whimsical message will appear in your echo area. On my system
2223 it says, @code{"There are 58 pink elephants in the office!"}.
2225 The expression @code{(- fill-column 14)} is evaluated and the resulting
2226 number is inserted in place of the @samp{%d}; and the string in double
2227 quotes, @code{"pink elephants"}, is treated as a single argument and
2228 inserted in place of the @samp{%s}. (That is to say, a string between
2229 double quotes evaluates to itself, like a number.)
2231 Finally, here is a somewhat complex example that not only illustrates
2232 the computation of a number, but also shows how you can use an
2233 expression within an expression to generate the text that is substituted
2238 (message "He saw %d %s"
2242 "The quick brown foxes jumped." 16 21)
2247 In this example, @code{message} has three arguments: the string,
2248 @code{"He saw %d %s"}, the expression, @code{(- fill-column 32)}, and
2249 the expression beginning with the function @code{concat}. The value
2250 resulting from the evaluation of @code{(- fill-column 32)} is inserted
2251 in place of the @samp{%d}; and the value returned by the expression
2252 beginning with @code{concat} is inserted in place of the @samp{%s}.
2254 When your fill column is 70 and you evaluate the expression, the
2255 message @code{"He saw 38 red foxes leaping."} appears in your echo
2259 @section Setting the Value of a Variable
2260 @cindex Variable, setting value
2261 @cindex Setting value of variable
2263 @cindex @samp{bind} defined
2264 There are several ways by which a variable can be given a value. One of
2265 the ways is to use either the function @code{set} or the function
2266 @code{setq}. Another way is to use @code{let} (@pxref{let}). (The
2267 jargon for this process is to @dfn{bind} a variable to a value.)
2269 The following sections not only describe how @code{set} and @code{setq}
2270 work but also illustrate how arguments are passed.
2273 * Using set:: Setting values.
2274 * Using setq:: Setting a quoted value.
2275 * Counting:: Using @code{setq} to count.
2279 @subsection Using @code{set}
2282 To set the value of the symbol @code{flowers} to the list @code{'(rose
2283 violet daisy buttercup)}, evaluate the following expression by
2284 positioning the cursor after the expression and typing @kbd{C-x C-e}.
2287 (set 'flowers '(rose violet daisy buttercup))
2291 The list @code{(rose violet daisy buttercup)} will appear in the echo
2292 area. This is what is @emph{returned} by the @code{set} function. As a
2293 side effect, the symbol @code{flowers} is bound to the list; that is,
2294 the symbol @code{flowers}, which can be viewed as a variable, is given
2295 the list as its value. (This process, by the way, illustrates how a
2296 side effect to the Lisp interpreter, setting the value, can be the
2297 primary effect that we humans are interested in. This is because every
2298 Lisp function must return a value if it does not get an error, but it
2299 will only have a side effect if it is designed to have one.)
2301 After evaluating the @code{set} expression, you can evaluate the symbol
2302 @code{flowers} and it will return the value you just set. Here is the
2303 symbol. Place your cursor after it and type @kbd{C-x C-e}.
2310 When you evaluate @code{flowers}, the list
2311 @code{(rose violet daisy buttercup)} appears in the echo area.
2313 Incidentally, if you evaluate @code{'flowers}, the variable with a quote
2314 in front of it, what you will see in the echo area is the symbol itself,
2315 @code{flowers}. Here is the quoted symbol, so you can try this:
2321 Note also, that when you use @code{set}, you need to quote both
2322 arguments to @code{set}, unless you want them evaluated. Since we do
2323 not want either argument evaluated, neither the variable
2324 @code{flowers} nor the list @code{(rose violet daisy buttercup)}, both
2325 are quoted. (When you use @code{set} without quoting its first
2326 argument, the first argument is evaluated before anything else is
2327 done. If you did this and @code{flowers} did not have a value
2328 already, you would get an error message that the @samp{Symbol's value
2329 as variable is void}; on the other hand, if @code{flowers} did return
2330 a value after it was evaluated, the @code{set} would attempt to set
2331 the value that was returned. There are situations where this is the
2332 right thing for the function to do; but such situations are rare.)
2335 @subsection Using @code{setq}
2338 As a practical matter, you almost always quote the first argument to
2339 @code{set}. The combination of @code{set} and a quoted first argument
2340 is so common that it has its own name: the special form @code{setq}.
2341 This special form is just like @code{set} except that the first argument
2342 is quoted automatically, so you don't need to type the quote mark
2343 yourself. Also, as an added convenience, @code{setq} permits you to set
2344 several different variables to different values, all in one expression.
2346 To set the value of the variable @code{carnivores} to the list
2347 @code{'(lion tiger leopard)} using @code{setq}, the following expression
2351 (setq carnivores '(lion tiger leopard))
2355 This is exactly the same as using @code{set} except the first argument
2356 is automatically quoted by @code{setq}. (The @samp{q} in @code{setq}
2357 means @code{quote}.)
2360 With @code{set}, the expression would look like this:
2363 (set 'carnivores '(lion tiger leopard))
2366 Also, @code{setq} can be used to assign different values to
2367 different variables. The first argument is bound to the value
2368 of the second argument, the third argument is bound to the value of the
2369 fourth argument, and so on. For example, you could use the following to
2370 assign a list of trees to the symbol @code{trees} and a list of herbivores
2371 to the symbol @code{herbivores}:
2375 (setq trees '(pine fir oak maple)
2376 herbivores '(gazelle antelope zebra))
2381 (The expression could just as well have been on one line, but it might
2382 not have fit on a page; and humans find it easier to read nicely
2385 Although I have been using the term `assign', there is another way of
2386 thinking about the workings of @code{set} and @code{setq}; and that is to
2387 say that @code{set} and @code{setq} make the symbol @emph{point} to the
2388 list. This latter way of thinking is very common and in forthcoming
2389 chapters we shall come upon at least one symbol that has `pointer' as
2390 part of its name. The name is chosen because the symbol has a value,
2391 specifically a list, attached to it; or, expressed another way,
2392 the symbol is set to ``point'' to the list.
2395 @subsection Counting
2398 Here is an example that shows how to use @code{setq} in a counter. You
2399 might use this to count how many times a part of your program repeats
2400 itself. First set a variable to zero; then add one to the number each
2401 time the program repeats itself. To do this, you need a variable that
2402 serves as a counter, and two expressions: an initial @code{setq}
2403 expression that sets the counter variable to zero; and a second
2404 @code{setq} expression that increments the counter each time it is
2409 (setq counter 0) ; @r{Let's call this the initializer.}
2411 (setq counter (+ counter 1)) ; @r{This is the incrementer.}
2413 counter ; @r{This is the counter.}
2418 (The text following the @samp{;} are comments. @xref{Change a
2419 defun, , Change a Function Definition}.)
2421 If you evaluate the first of these expressions, the initializer,
2422 @code{(setq counter 0)}, and then evaluate the third expression,
2423 @code{counter}, the number @code{0} will appear in the echo area. If
2424 you then evaluate the second expression, the incrementer, @code{(setq
2425 counter (+ counter 1))}, the counter will get the value 1. So if you
2426 again evaluate @code{counter}, the number @code{1} will appear in the
2427 echo area. Each time you evaluate the second expression, the value of
2428 the counter will be incremented.
2430 When you evaluate the incrementer, @code{(setq counter (+ counter 1))},
2431 the Lisp interpreter first evaluates the innermost list; this is the
2432 addition. In order to evaluate this list, it must evaluate the variable
2433 @code{counter} and the number @code{1}. When it evaluates the variable
2434 @code{counter}, it receives its current value. It passes this value and
2435 the number @code{1} to the @code{+} which adds them together. The sum
2436 is then returned as the value of the inner list and passed to the
2437 @code{setq} which sets the variable @code{counter} to this new value.
2438 Thus, the value of the variable, @code{counter}, is changed.
2443 Learning Lisp is like climbing a hill in which the first part is the
2444 steepest. You have now climbed the most difficult part; what remains
2445 becomes easier as you progress onwards.
2453 Lisp programs are made up of expressions, which are lists or single atoms.
2456 Lists are made up of zero or more atoms or inner lists, separated by whitespace and
2457 surrounded by parentheses. A list can be empty.
2460 Atoms are multi-character symbols, like @code{forward-paragraph}, single
2461 character symbols like @code{+}, strings of characters between double
2462 quotation marks, or numbers.
2465 A number evaluates to itself.
2468 A string between double quotes also evaluates to itself.
2471 When you evaluate a symbol by itself, its value is returned.
2474 When you evaluate a list, the Lisp interpreter looks at the first symbol
2475 in the list and then at the function definition bound to that symbol.
2476 Then the instructions in the function definition are carried out.
2479 A single quotation mark,
2486 , tells the Lisp interpreter that it should
2487 return the following expression as written, and not evaluate it as it
2488 would if the quote were not there.
2491 Arguments are the information passed to a function. The arguments to a
2492 function are computed by evaluating the rest of the elements of the list
2493 of which the function is the first element.
2496 A function always returns a value when it is evaluated (unless it gets
2497 an error); in addition, it may also carry out some action called a
2498 ``side effect''. In many cases, a function's primary purpose is to
2499 create a side effect.
2502 @node Error Message Exercises
2505 A few simple exercises:
2509 Generate an error message by evaluating an appropriate symbol that is
2510 not within parentheses.
2513 Generate an error message by evaluating an appropriate symbol that is
2514 between parentheses.
2517 Create a counter that increments by two rather than one.
2520 Write an expression that prints a message in the echo area when
2524 @node Practicing Evaluation
2525 @chapter Practicing Evaluation
2526 @cindex Practicing evaluation
2527 @cindex Evaluation practice
2529 Before learning how to write a function definition in Emacs Lisp, it is
2530 useful to spend a little time evaluating various expressions that have
2531 already been written. These expressions will be lists with the
2532 functions as their first (and often only) element. Since some of the
2533 functions associated with buffers are both simple and interesting, we
2534 will start with those. In this section, we will evaluate a few of
2535 these. In another section, we will study the code of several other
2536 buffer-related functions, to see how they were written.
2539 * How to Evaluate:: Typing editing commands or @kbd{C-x C-e}
2541 * Buffer Names:: Buffers and files are different.
2542 * Getting Buffers:: Getting a buffer itself, not merely its name.
2543 * Switching Buffers:: How to change to another buffer.
2544 * Buffer Size & Locations:: Where point is located and the size of
2546 * Evaluation Exercise::
2550 @node How to Evaluate
2551 @unnumberedsec How to Evaluate
2554 @i{Whenever you give an editing command} to Emacs Lisp, such as the
2555 command to move the cursor or to scroll the screen, @i{you are evaluating
2556 an expression,} the first element of which is a function. @i{This is
2559 @cindex @samp{interactive function} defined
2560 @cindex @samp{command} defined
2561 When you type keys, you cause the Lisp interpreter to evaluate an
2562 expression and that is how you get your results. Even typing plain text
2563 involves evaluating an Emacs Lisp function, in this case, one that uses
2564 @code{self-insert-command}, which simply inserts the character you
2565 typed. The functions you evaluate by typing keystrokes are called
2566 @dfn{interactive} functions, or @dfn{commands}; how you make a function
2567 interactive will be illustrated in the chapter on how to write function
2568 definitions. @xref{Interactive, , Making a Function Interactive}.
2570 In addition to typing keyboard commands, we have seen a second way to
2571 evaluate an expression: by positioning the cursor after a list and
2572 typing @kbd{C-x C-e}. This is what we will do in the rest of this
2573 section. There are other ways to evaluate an expression as well; these
2574 will be described as we come to them.
2576 Besides being used for practicing evaluation, the functions shown in the
2577 next few sections are important in their own right. A study of these
2578 functions makes clear the distinction between buffers and files, how to
2579 switch to a buffer, and how to determine a location within it.
2582 @section Buffer Names
2584 @findex buffer-file-name
2586 The two functions, @code{buffer-name} and @code{buffer-file-name}, show
2587 the difference between a file and a buffer. When you evaluate the
2588 following expression, @code{(buffer-name)}, the name of the buffer
2589 appears in the echo area. When you evaluate @code{(buffer-file-name)},
2590 the name of the file to which the buffer refers appears in the echo
2591 area. Usually, the name returned by @code{(buffer-name)} is the same as
2592 the name of the file to which it refers, and the name returned by
2593 @code{(buffer-file-name)} is the full path-name of the file.
2595 A file and a buffer are two different entities. A file is information
2596 recorded permanently in the computer (unless you delete it). A buffer,
2597 on the other hand, is information inside of Emacs that will vanish at
2598 the end of the editing session (or when you kill the buffer). Usually,
2599 a buffer contains information that you have copied from a file; we say
2600 the buffer is @dfn{visiting} that file. This copy is what you work on
2601 and modify. Changes to the buffer do not change the file, until you
2602 save the buffer. When you save the buffer, the buffer is copied to the file
2603 and is thus saved permanently.
2606 If you are reading this in Info inside of GNU Emacs, you can evaluate
2607 each of the following expressions by positioning the cursor after it and
2608 typing @kbd{C-x C-e}.
2619 When I do this in Info, the value returned by evaluating
2620 @code{(buffer-name)} is @file{"*info*"}, and the value returned by
2621 evaluating @code{(buffer-file-name)} is @file{nil}.
2623 On the other hand, while I am writing this document, the value
2624 returned by evaluating @code{(buffer-name)} is
2625 @file{"introduction.texinfo"}, and the value returned by evaluating
2626 @code{(buffer-file-name)} is
2627 @file{"/gnu/work/intro/introduction.texinfo"}.
2629 @cindex @code{nil}, history of word
2630 The former is the name of the buffer and the latter is the name of the
2631 file. In Info, the buffer name is @file{"*info*"}. Info does not
2632 point to any file, so the result of evaluating
2633 @code{(buffer-file-name)} is @file{nil}. The symbol @code{nil} is
2634 from the Latin word for `nothing'; in this case, it means that the
2635 buffer is not associated with any file. (In Lisp, @code{nil} is also
2636 used to mean `false' and is a synonym for the empty list, @code{()}.)
2638 When I am writing, the name of my buffer is
2639 @file{"introduction.texinfo"}. The name of the file to which it
2640 points is @file{"/gnu/work/intro/introduction.texinfo"}.
2642 (In the expressions, the parentheses tell the Lisp interpreter to
2643 treat @w{@code{buffer-name}} and @w{@code{buffer-file-name}} as
2644 functions; without the parentheses, the interpreter would attempt to
2645 evaluate the symbols as variables. @xref{Variables}.)
2647 In spite of the distinction between files and buffers, you will often
2648 find that people refer to a file when they mean a buffer and vice-verse.
2649 Indeed, most people say, ``I am editing a file,'' rather than saying,
2650 ``I am editing a buffer which I will soon save to a file.'' It is
2651 almost always clear from context what people mean. When dealing with
2652 computer programs, however, it is important to keep the distinction in mind,
2653 since the computer is not as smart as a person.
2655 @cindex Buffer, history of word
2656 The word `buffer', by the way, comes from the meaning of the word as a
2657 cushion that deadens the force of a collision. In early computers, a
2658 buffer cushioned the interaction between files and the computer's
2659 central processing unit. The drums or tapes that held a file and the
2660 central processing unit were pieces of equipment that were very
2661 different from each other, working at their own speeds, in spurts. The
2662 buffer made it possible for them to work together effectively.
2663 Eventually, the buffer grew from being an intermediary, a temporary
2664 holding place, to being the place where work is done. This
2665 transformation is rather like that of a small seaport that grew into a
2666 great city: once it was merely the place where cargo was warehoused
2667 temporarily before being loaded onto ships; then it became a business
2668 and cultural center in its own right.
2670 Not all buffers are associated with files. For example, a
2671 @file{*scratch*} buffer does not visit any file. Similarly, a
2672 @file{*Help*} buffer is not associated with any file.
2674 In the old days, when you lacked a @file{~/.emacs} file and started an
2675 Emacs session by typing the command @code{emacs} alone, without naming
2676 any files, Emacs started with the @file{*scratch*} buffer visible.
2677 Nowadays, you will see a splash screen. You can follow one of the
2678 commands suggested on the splash screen, visit a file, or press the
2679 spacebar to reach the @file{*scratch*} buffer.
2681 If you switch to the @file{*scratch*} buffer, type
2682 @code{(buffer-name)}, position the cursor after it, and then type
2683 @kbd{C-x C-e} to evaluate the expression. The name @code{"*scratch*"}
2684 will be returned and will appear in the echo area. @code{"*scratch*"}
2685 is the name of the buffer. When you type @code{(buffer-file-name)} in
2686 the @file{*scratch*} buffer and evaluate that, @code{nil} will appear
2687 in the echo area, just as it does when you evaluate
2688 @code{(buffer-file-name)} in Info.
2690 Incidentally, if you are in the @file{*scratch*} buffer and want the
2691 value returned by an expression to appear in the @file{*scratch*}
2692 buffer itself rather than in the echo area, type @kbd{C-u C-x C-e}
2693 instead of @kbd{C-x C-e}. This causes the value returned to appear
2694 after the expression. The buffer will look like this:
2697 (buffer-name)"*scratch*"
2701 You cannot do this in Info since Info is read-only and it will not allow
2702 you to change the contents of the buffer. But you can do this in any
2703 buffer you can edit; and when you write code or documentation (such as
2704 this book), this feature is very useful.
2706 @node Getting Buffers
2707 @section Getting Buffers
2708 @findex current-buffer
2709 @findex other-buffer
2710 @cindex Getting a buffer
2712 The @code{buffer-name} function returns the @emph{name} of the buffer;
2713 to get the buffer @emph{itself}, a different function is needed: the
2714 @code{current-buffer} function. If you use this function in code, what
2715 you get is the buffer itself.
2717 A name and the object or entity to which the name refers are different
2718 from each other. You are not your name. You are a person to whom
2719 others refer by name. If you ask to speak to George and someone hands you
2720 a card with the letters @samp{G}, @samp{e}, @samp{o}, @samp{r},
2721 @samp{g}, and @samp{e} written on it, you might be amused, but you would
2722 not be satisfied. You do not want to speak to the name, but to the
2723 person to whom the name refers. A buffer is similar: the name of the
2724 scratch buffer is @file{*scratch*}, but the name is not the buffer. To
2725 get a buffer itself, you need to use a function such as
2726 @code{current-buffer}.
2728 However, there is a slight complication: if you evaluate
2729 @code{current-buffer} in an expression on its own, as we will do here,
2730 what you see is a printed representation of the name of the buffer
2731 without the contents of the buffer. Emacs works this way for two
2732 reasons: the buffer may be thousands of lines long---too long to be
2733 conveniently displayed; and, another buffer may have the same contents
2734 but a different name, and it is important to distinguish between them.
2737 Here is an expression containing the function:
2744 If you evaluate this expression in Info in Emacs in the usual way,
2745 @file{#<buffer *info*>} will appear in the echo area. The special
2746 format indicates that the buffer itself is being returned, rather than
2749 Incidentally, while you can type a number or symbol into a program, you
2750 cannot do that with the printed representation of a buffer: the only way
2751 to get a buffer itself is with a function such as @code{current-buffer}.
2753 A related function is @code{other-buffer}. This returns the most
2754 recently selected buffer other than the one you are in currently, not
2755 a printed representation of its name. If you have recently switched
2756 back and forth from the @file{*scratch*} buffer, @code{other-buffer}
2757 will return that buffer.
2760 You can see this by evaluating the expression:
2767 You should see @file{#<buffer *scratch*>} appear in the echo area, or
2768 the name of whatever other buffer you switched back from most
2769 recently@footnote{Actually, by default, if the buffer from which you
2770 just switched is visible to you in another window, @code{other-buffer}
2771 will choose the most recent buffer that you cannot see; this is a
2772 subtlety that I often forget.}.
2774 @node Switching Buffers
2775 @section Switching Buffers
2776 @findex switch-to-buffer
2778 @cindex Switching to a buffer
2780 The @code{other-buffer} function actually provides a buffer when it is
2781 used as an argument to a function that requires one. We can see this
2782 by using @code{other-buffer} and @code{switch-to-buffer} to switch to a
2785 But first, a brief introduction to the @code{switch-to-buffer}
2786 function. When you switched back and forth from Info to the
2787 @file{*scratch*} buffer to evaluate @code{(buffer-name)}, you most
2788 likely typed @kbd{C-x b} and then typed @file{*scratch*}@footnote{Or
2789 rather, to save typing, you probably only typed @kbd{RET} if the
2790 default buffer was @file{*scratch*}, or if it was different, then you
2791 typed just part of the name, such as @code{*sc}, pressed your
2792 @kbd{TAB} key to cause it to expand to the full name, and then typed
2793 @kbd{RET}.} when prompted in the minibuffer for the name of
2794 the buffer to which you wanted to switch. The keystrokes, @kbd{C-x
2795 b}, cause the Lisp interpreter to evaluate the interactive function
2796 @code{switch-to-buffer}. As we said before, this is how Emacs works:
2797 different keystrokes call or run different functions. For example,
2798 @kbd{C-f} calls @code{forward-char}, @kbd{M-e} calls
2799 @code{forward-sentence}, and so on.
2801 By writing @code{switch-to-buffer} in an expression, and giving it a
2802 buffer to switch to, we can switch buffers just the way @kbd{C-x b}
2806 (switch-to-buffer (other-buffer))
2810 The symbol @code{switch-to-buffer} is the first element of the list,
2811 so the Lisp interpreter will treat it as a function and carry out the
2812 instructions that are attached to it. But before doing that, the
2813 interpreter will note that @code{other-buffer} is inside parentheses
2814 and work on that symbol first. @code{other-buffer} is the first (and
2815 in this case, the only) element of this list, so the Lisp interpreter
2816 calls or runs the function. It returns another buffer. Next, the
2817 interpreter runs @code{switch-to-buffer}, passing to it, as an
2818 argument, the other buffer, which is what Emacs will switch to. If
2819 you are reading this in Info, try this now. Evaluate the expression.
2820 (To get back, type @kbd{C-x b @key{RET}}.)@footnote{Remember, this
2821 expression will move you to your most recent other buffer that you
2822 cannot see. If you really want to go to your most recently selected
2823 buffer, even if you can still see it, you need to evaluate the
2824 following more complex expression:
2827 (switch-to-buffer (other-buffer (current-buffer) t))
2831 In this case, the first argument to @code{other-buffer} tells it which
2832 buffer to skip---the current one---and the second argument tells
2833 @code{other-buffer} it is OK to switch to a visible buffer.
2834 In regular use, @code{switch-to-buffer} takes you to an invisible
2835 window since you would most likely use @kbd{C-x o} (@code{other-window})
2836 to go to another visible buffer.}
2838 In the programming examples in later sections of this document, you will
2839 see the function @code{set-buffer} more often than
2840 @code{switch-to-buffer}. This is because of a difference between
2841 computer programs and humans: humans have eyes and expect to see the
2842 buffer on which they are working on their computer terminals. This is
2843 so obvious, it almost goes without saying. However, programs do not
2844 have eyes. When a computer program works on a buffer, that buffer does
2845 not need to be visible on the screen.
2847 @code{switch-to-buffer} is designed for humans and does two different
2848 things: it switches the buffer to which Emacs's attention is directed; and
2849 it switches the buffer displayed in the window to the new buffer.
2850 @code{set-buffer}, on the other hand, does only one thing: it switches
2851 the attention of the computer program to a different buffer. The buffer
2852 on the screen remains unchanged (of course, normally nothing happens
2853 there until the command finishes running).
2855 @cindex @samp{call} defined
2856 Also, we have just introduced another jargon term, the word @dfn{call}.
2857 When you evaluate a list in which the first symbol is a function, you
2858 are calling that function. The use of the term comes from the notion of
2859 the function as an entity that can do something for you if you `call'
2860 it---just as a plumber is an entity who can fix a leak if you call him
2863 @node Buffer Size & Locations
2864 @section Buffer Size and the Location of Point
2865 @cindex Size of buffer
2867 @cindex Point location
2868 @cindex Location of point
2870 Finally, let's look at several rather simple functions,
2871 @code{buffer-size}, @code{point}, @code{point-min}, and
2872 @code{point-max}. These give information about the size of a buffer and
2873 the location of point within it.
2875 The function @code{buffer-size} tells you the size of the current
2876 buffer; that is, the function returns a count of the number of
2877 characters in the buffer.
2884 You can evaluate this in the usual way, by positioning the
2885 cursor after the expression and typing @kbd{C-x C-e}.
2887 @cindex @samp{point} defined
2888 In Emacs, the current position of the cursor is called @dfn{point}.
2889 The expression @code{(point)} returns a number that tells you where the
2890 cursor is located as a count of the number of characters from the
2891 beginning of the buffer up to point.
2894 You can see the character count for point in this buffer by evaluating
2895 the following expression in the usual way:
2902 As I write this, the value of @code{point} is 65724. The @code{point}
2903 function is frequently used in some of the examples later in this
2907 The value of point depends, of course, on its location within the
2908 buffer. If you evaluate point in this spot, the number will be larger:
2915 For me, the value of point in this location is 66043, which means that
2916 there are 319 characters (including spaces) between the two
2917 expressions. (Doubtless, you will see different numbers, since I will
2918 have edited this since I first evaluated point.)
2920 @cindex @samp{narrowing} defined
2921 The function @code{point-min} is somewhat similar to @code{point}, but
2922 it returns the value of the minimum permissible value of point in the
2923 current buffer. This is the number 1 unless @dfn{narrowing} is in
2924 effect. (Narrowing is a mechanism whereby you can restrict yourself,
2925 or a program, to operations on just a part of a buffer.
2926 @xref{Narrowing & Widening, , Narrowing and Widening}.) Likewise, the
2927 function @code{point-max} returns the value of the maximum permissible
2928 value of point in the current buffer.
2930 @node Evaluation Exercise
2933 Find a file with which you are working and move towards its middle.
2934 Find its buffer name, file name, length, and your position in the file.
2936 @node Writing Defuns
2937 @chapter How To Write Function Definitions
2938 @cindex Definition writing
2939 @cindex Function definition writing
2940 @cindex Writing a function definition
2942 When the Lisp interpreter evaluates a list, it looks to see whether the
2943 first symbol on the list has a function definition attached to it; or,
2944 put another way, whether the symbol points to a function definition. If
2945 it does, the computer carries out the instructions in the definition. A
2946 symbol that has a function definition is called, simply, a function
2947 (although, properly speaking, the definition is the function and the
2948 symbol refers to it.)
2951 * Primitive Functions::
2952 * defun:: The @code{defun} macro.
2953 * Install:: Install a function definition.
2954 * Interactive:: Making a function interactive.
2955 * Interactive Options:: Different options for @code{interactive}.
2956 * Permanent Installation:: Installing code permanently.
2957 * let:: Creating and initializing local variables.
2959 * else:: If--then--else expressions.
2960 * Truth & Falsehood:: What Lisp considers false and true.
2961 * save-excursion:: Keeping track of point, mark, and buffer.
2967 @node Primitive Functions
2968 @unnumberedsec An Aside about Primitive Functions
2970 @cindex Primitive functions
2971 @cindex Functions, primitive
2973 @cindex C language primitives
2974 @cindex Primitives written in C
2975 All functions are defined in terms of other functions, except for a few
2976 @dfn{primitive} functions that are written in the C programming
2977 language. When you write functions' definitions, you will write them in
2978 Emacs Lisp and use other functions as your building blocks. Some of the
2979 functions you will use will themselves be written in Emacs Lisp (perhaps
2980 by you) and some will be primitives written in C@. The primitive
2981 functions are used exactly like those written in Emacs Lisp and behave
2982 like them. They are written in C so we can easily run GNU Emacs on any
2983 computer that has sufficient power and can run C.
2985 Let me re-emphasize this: when you write code in Emacs Lisp, you do not
2986 distinguish between the use of functions written in C and the use of
2987 functions written in Emacs Lisp. The difference is irrelevant. I
2988 mention the distinction only because it is interesting to know. Indeed,
2989 unless you investigate, you won't know whether an already-written
2990 function is written in Emacs Lisp or C.
2993 @section The @code{defun} Macro
2996 @cindex @samp{function definition} defined
2997 In Lisp, a symbol such as @code{mark-whole-buffer} has code attached to
2998 it that tells the computer what to do when the function is called.
2999 This code is called the @dfn{function definition} and is created by
3000 evaluating a Lisp expression that starts with the symbol @code{defun}
3001 (which is an abbreviation for @emph{define function}).
3003 In subsequent sections, we will look at function definitions from the
3004 Emacs source code, such as @code{mark-whole-buffer}. In this section,
3005 we will describe a simple function definition so you can see how it
3006 looks. This function definition uses arithmetic because it makes for a
3007 simple example. Some people dislike examples using arithmetic; however,
3008 if you are such a person, do not despair. Hardly any of the code we
3009 will study in the remainder of this introduction involves arithmetic or
3010 mathematics. The examples mostly involve text in one way or another.
3012 A function definition has up to five parts following the word
3017 The name of the symbol to which the function definition should be
3021 A list of the arguments that will be passed to the function. If no
3022 arguments will be passed to the function, this is an empty list,
3026 Documentation describing the function. (Technically optional, but
3027 strongly recommended.)
3030 Optionally, an expression to make the function interactive so you can
3031 use it by typing @kbd{M-x} and then the name of the function; or by
3032 typing an appropriate key or keychord.
3034 @cindex @samp{body} defined
3036 The code that instructs the computer what to do: the @dfn{body} of the
3037 function definition.
3040 It is helpful to think of the five parts of a function definition as
3041 being organized in a template, with slots for each part:
3045 (defun @var{function-name} (@var{arguments}@dots{})
3046 "@var{optional-documentation}@dots{}"
3047 (interactive @var{argument-passing-info}) ; @r{optional}
3052 As an example, here is the code for a function that multiplies its
3053 argument by 7. (This example is not interactive. @xref{Interactive,
3054 , Making a Function Interactive}, for that information.)
3058 (defun multiply-by-seven (number)
3059 "Multiply NUMBER by seven."
3064 This definition begins with a parenthesis and the symbol @code{defun},
3065 followed by the name of the function.
3067 @cindex @samp{argument list} defined
3068 The name of the function is followed by a list that contains the
3069 arguments that will be passed to the function. This list is called
3070 the @dfn{argument list}. In this example, the list has only one
3071 element, the symbol, @code{number}. When the function is used, the
3072 symbol will be bound to the value that is used as the argument to the
3075 Instead of choosing the word @code{number} for the name of the argument,
3076 I could have picked any other name. For example, I could have chosen
3077 the word @code{multiplicand}. I picked the word `number' because it
3078 tells what kind of value is intended for this slot; but I could just as
3079 well have chosen the word `multiplicand' to indicate the role that the
3080 value placed in this slot will play in the workings of the function. I
3081 could have called it @code{foogle}, but that would have been a bad
3082 choice because it would not tell humans what it means. The choice of
3083 name is up to the programmer and should be chosen to make the meaning of
3086 Indeed, you can choose any name you wish for a symbol in an argument
3087 list, even the name of a symbol used in some other function: the name
3088 you use in an argument list is private to that particular definition.
3089 In that definition, the name refers to a different entity than any use
3090 of the same name outside the function definition. Suppose you have a
3091 nick-name `Shorty' in your family; when your family members refer to
3092 `Shorty', they mean you. But outside your family, in a movie, for
3093 example, the name `Shorty' refers to someone else. Because a name in an
3094 argument list is private to the function definition, you can change the
3095 value of such a symbol inside the body of a function without changing
3096 its value outside the function. The effect is similar to that produced
3097 by a @code{let} expression. (@xref{let, , @code{let}}.)
3100 Note also that we discuss the word `number' in two different ways: as a
3101 symbol that appears in the code, and as the name of something that will
3102 be replaced by a something else during the evaluation of the function.
3103 In the first case, @code{number} is a symbol, not a number; it happens
3104 that within the function, it is a variable who value is the number in
3105 question, but our primary interest in it is as a symbol. On the other
3106 hand, when we are talking about the function, our interest is that we
3107 will substitute a number for the word @var{number}. To keep this
3108 distinction clear, we use different typography for the two
3109 circumstances. When we talk about this function, or about how it works,
3110 we refer to this number by writing @var{number}. In the function
3111 itself, we refer to it by writing @code{number}.
3114 The argument list is followed by the documentation string that
3115 describes the function. This is what you see when you type
3116 @w{@kbd{C-h f}} and the name of a function. Incidentally, when you
3117 write a documentation string like this, you should make the first line
3118 a complete sentence since some commands, such as @code{apropos}, print
3119 only the first line of a multi-line documentation string. Also, you
3120 should not indent the second line of a documentation string, if you
3121 have one, because that looks odd when you use @kbd{C-h f}
3122 (@code{describe-function}). The documentation string is optional, but
3123 it is so useful, it should be included in almost every function you
3126 @findex * @r{(multiplication)}
3127 The third line of the example consists of the body of the function
3128 definition. (Most functions' definitions, of course, are longer than
3129 this.) In this function, the body is the list, @code{(* 7 number)}, which
3130 says to multiply the value of @var{number} by 7. (In Emacs Lisp,
3131 @code{*} is the function for multiplication, just as @code{+} is the
3132 function for addition.)
3134 When you use the @code{multiply-by-seven} function, the argument
3135 @code{number} evaluates to the actual number you want used. Here is an
3136 example that shows how @code{multiply-by-seven} is used; but don't try
3137 to evaluate this yet!
3140 (multiply-by-seven 3)
3144 The symbol @code{number}, specified in the function definition in the
3145 next section, is given or ``bound to'' the value 3 in the actual use of
3146 the function. Note that although @code{number} was inside parentheses
3147 in the function definition, the argument passed to the
3148 @code{multiply-by-seven} function is not in parentheses. The
3149 parentheses are written in the function definition so the computer can
3150 figure out where the argument list ends and the rest of the function
3153 If you evaluate this example, you are likely to get an error message.
3154 (Go ahead, try it!) This is because we have written the function
3155 definition, but not yet told the computer about the definition---we have
3156 not yet installed (or `loaded') the function definition in Emacs.
3157 Installing a function is the process that tells the Lisp interpreter the
3158 definition of the function. Installation is described in the next
3162 @section Install a Function Definition
3163 @cindex Install a Function Definition
3164 @cindex Definition installation
3165 @cindex Function definition installation
3167 If you are reading this inside of Info in Emacs, you can try out the
3168 @code{multiply-by-seven} function by first evaluating the function
3169 definition and then evaluating @code{(multiply-by-seven 3)}. A copy of
3170 the function definition follows. Place the cursor after the last
3171 parenthesis of the function definition and type @kbd{C-x C-e}. When you
3172 do this, @code{multiply-by-seven} will appear in the echo area. (What
3173 this means is that when a function definition is evaluated, the value it
3174 returns is the name of the defined function.) At the same time, this
3175 action installs the function definition.
3179 (defun multiply-by-seven (number)
3180 "Multiply NUMBER by seven."
3186 By evaluating this @code{defun}, you have just installed
3187 @code{multiply-by-seven} in Emacs. The function is now just as much a
3188 part of Emacs as @code{forward-word} or any other editing function you
3189 use. (@code{multiply-by-seven} will stay installed until you quit
3190 Emacs. To reload code automatically whenever you start Emacs, see
3191 @ref{Permanent Installation, , Installing Code Permanently}.)
3194 * Effect of installation::
3195 * Change a defun:: How to change a function definition.
3199 @node Effect of installation
3200 @unnumberedsubsec The effect of installation
3203 You can see the effect of installing @code{multiply-by-seven} by
3204 evaluating the following sample. Place the cursor after the following
3205 expression and type @kbd{C-x C-e}. The number 21 will appear in the
3209 (multiply-by-seven 3)
3212 If you wish, you can read the documentation for the function by typing
3213 @kbd{C-h f} (@code{describe-function}) and then the name of the
3214 function, @code{multiply-by-seven}. When you do this, a
3215 @file{*Help*} window will appear on your screen that says:
3219 multiply-by-seven is a Lisp function.
3220 (multiply-by-seven NUMBER)
3222 Multiply NUMBER by seven.
3227 (To return to a single window on your screen, type @kbd{C-x 1}.)
3229 @node Change a defun
3230 @subsection Change a Function Definition
3231 @cindex Changing a function definition
3232 @cindex Function definition, how to change
3233 @cindex Definition, how to change
3235 If you want to change the code in @code{multiply-by-seven}, just rewrite
3236 it. To install the new version in place of the old one, evaluate the
3237 function definition again. This is how you modify code in Emacs. It is
3240 As an example, you can change the @code{multiply-by-seven} function to
3241 add the number to itself seven times instead of multiplying the number
3242 by seven. It produces the same answer, but by a different path. At
3243 the same time, we will add a comment to the code; a comment is text
3244 that the Lisp interpreter ignores, but that a human reader may find
3245 useful or enlightening. The comment is that this is the ``second
3250 (defun multiply-by-seven (number) ; @r{Second version.}
3251 "Multiply NUMBER by seven."
3252 (+ number number number number number number number))
3256 @cindex Comments in Lisp code
3257 The comment follows a semicolon, @samp{;}. In Lisp, everything on a
3258 line that follows a semicolon is a comment. The end of the line is the
3259 end of the comment. To stretch a comment over two or more lines, begin
3260 each line with a semicolon.
3262 @xref{Beginning a .emacs File, , Beginning a @file{.emacs}
3263 File}, and @ref{Comments, , Comments, elisp, The GNU Emacs Lisp
3264 Reference Manual}, for more about comments.
3266 You can install this version of the @code{multiply-by-seven} function by
3267 evaluating it in the same way you evaluated the first function: place
3268 the cursor after the last parenthesis and type @kbd{C-x C-e}.
3270 In summary, this is how you write code in Emacs Lisp: you write a
3271 function; install it; test it; and then make fixes or enhancements and
3275 @section Make a Function Interactive
3276 @cindex Interactive functions
3279 You make a function interactive by placing a list that begins with
3280 the special form @code{interactive} immediately after the
3281 documentation. A user can invoke an interactive function by typing
3282 @kbd{M-x} and then the name of the function; or by typing the keys to
3283 which it is bound, for example, by typing @kbd{C-n} for
3284 @code{next-line} or @kbd{C-x h} for @code{mark-whole-buffer}.
3286 Interestingly, when you call an interactive function interactively,
3287 the value returned is not automatically displayed in the echo area.
3288 This is because you often call an interactive function for its side
3289 effects, such as moving forward by a word or line, and not for the
3290 value returned. If the returned value were displayed in the echo area
3291 each time you typed a key, it would be very distracting.
3294 * Interactive multiply-by-seven:: An overview.
3295 * multiply-by-seven in detail:: The interactive version.
3299 @node Interactive multiply-by-seven
3300 @unnumberedsubsec An Interactive @code{multiply-by-seven}, An Overview
3303 Both the use of the special form @code{interactive} and one way to
3304 display a value in the echo area can be illustrated by creating an
3305 interactive version of @code{multiply-by-seven}.
3312 (defun multiply-by-seven (number) ; @r{Interactive version.}
3313 "Multiply NUMBER by seven."
3315 (message "The result is %d" (* 7 number)))
3320 You can install this code by placing your cursor after it and typing
3321 @kbd{C-x C-e}. The name of the function will appear in your echo area.
3322 Then, you can use this code by typing @kbd{C-u} and a number and then
3323 typing @kbd{M-x multiply-by-seven} and pressing @key{RET}. The phrase
3324 @samp{The result is @dots{}} followed by the product will appear in the
3327 Speaking more generally, you invoke a function like this in either of two
3332 By typing a prefix argument that contains the number to be passed, and
3333 then typing @kbd{M-x} and the name of the function, as with
3334 @kbd{C-u 3 M-x forward-sentence}; or,
3337 By typing whatever key or keychord the function is bound to, as with
3342 Both the examples just mentioned work identically to move point forward
3343 three sentences. (Since @code{multiply-by-seven} is not bound to a key,
3344 it could not be used as an example of key binding.)
3346 (@xref{Keybindings, , Some Keybindings}, to learn how to bind a command
3349 A prefix argument is passed to an interactive function by typing the
3350 @key{META} key followed by a number, for example, @kbd{M-3 M-e}, or by
3351 typing @kbd{C-u} and then a number, for example, @kbd{C-u 3 M-e} (if you
3352 type @kbd{C-u} without a number, it defaults to 4).
3354 @node multiply-by-seven in detail
3355 @subsection An Interactive @code{multiply-by-seven}
3357 Let's look at the use of the special form @code{interactive} and then at
3358 the function @code{message} in the interactive version of
3359 @code{multiply-by-seven}. You will recall that the function definition
3364 (defun multiply-by-seven (number) ; @r{Interactive version.}
3365 "Multiply NUMBER by seven."
3367 (message "The result is %d" (* 7 number)))
3371 In this function, the expression, @code{(interactive "p")}, is a list of
3372 two elements. The @code{"p"} tells Emacs to pass the prefix argument to
3373 the function and use its value for the argument of the function.
3376 The argument will be a number. This means that the symbol
3377 @code{number} will be bound to a number in the line:
3380 (message "The result is %d" (* 7 number))
3385 For example, if your prefix argument is 5, the Lisp interpreter will
3386 evaluate the line as if it were:
3389 (message "The result is %d" (* 7 5))
3393 (If you are reading this in GNU Emacs, you can evaluate this expression
3394 yourself.) First, the interpreter will evaluate the inner list, which
3395 is @code{(* 7 5)}. This returns a value of 35. Next, it
3396 will evaluate the outer list, passing the values of the second and
3397 subsequent elements of the list to the function @code{message}.
3399 As we have seen, @code{message} is an Emacs Lisp function especially
3400 designed for sending a one line message to a user. (@xref{message, ,
3401 The @code{message} function}.) In summary, the @code{message}
3402 function prints its first argument in the echo area as is, except for
3403 occurrences of @samp{%d} or @samp{%s} (and various other %-sequences
3404 which we have not mentioned). When it sees a control sequence, the
3405 function looks to the second or subsequent arguments and prints the
3406 value of the argument in the location in the string where the control
3407 sequence is located.
3409 In the interactive @code{multiply-by-seven} function, the control string
3410 is @samp{%d}, which requires a number, and the value returned by
3411 evaluating @code{(* 7 5)} is the number 35. Consequently, the number 35
3412 is printed in place of the @samp{%d} and the message is @samp{The result
3415 (Note that when you call the function @code{multiply-by-seven}, the
3416 message is printed without quotes, but when you call @code{message}, the
3417 text is printed in double quotes. This is because the value returned by
3418 @code{message} is what appears in the echo area when you evaluate an
3419 expression whose first element is @code{message}; but when embedded in a
3420 function, @code{message} prints the text as a side effect without
3423 @node Interactive Options
3424 @section Different Options for @code{interactive}
3425 @cindex Options for @code{interactive}
3426 @cindex Interactive options
3428 In the example, @code{multiply-by-seven} used @code{"p"} as the
3429 argument to @code{interactive}. This argument told Emacs to interpret
3430 your typing either @kbd{C-u} followed by a number or @key{META}
3431 followed by a number as a command to pass that number to the function
3432 as its argument. Emacs has more than twenty characters predefined for
3433 use with @code{interactive}. In almost every case, one of these
3434 options will enable you to pass the right information interactively to
3435 a function. (@xref{Interactive Codes, , Code Characters for
3436 @code{interactive}, elisp, The GNU Emacs Lisp Reference Manual}.)
3439 Consider the function @code{zap-to-char}. Its interactive expression
3443 (interactive "p\ncZap to char: ")
3446 The first part of the argument to @code{interactive} is @samp{p}, with
3447 which you are already familiar. This argument tells Emacs to
3448 interpret a `prefix', as a number to be passed to the function. You
3449 can specify a prefix either by typing @kbd{C-u} followed by a number
3450 or by typing @key{META} followed by a number. The prefix is the
3451 number of specified characters. Thus, if your prefix is three and the
3452 specified character is @samp{x}, then you will delete all the text up
3453 to and including the third next @samp{x}. If you do not set a prefix,
3454 then you delete all the text up to and including the specified
3455 character, but no more.
3457 The @samp{c} tells the function the name of the character to which to delete.
3459 More formally, a function with two or more arguments can have
3460 information passed to each argument by adding parts to the string that
3461 follows @code{interactive}. When you do this, the information is
3462 passed to each argument in the same order it is specified in the
3463 @code{interactive} list. In the string, each part is separated from
3464 the next part by a @samp{\n}, which is a newline. For example, you
3465 can follow @samp{p} with a @samp{\n} and an @samp{cZap to char:@: }.
3466 This causes Emacs to pass the value of the prefix argument (if there
3467 is one) and the character.
3469 In this case, the function definition looks like the following, where
3470 @code{arg} and @code{char} are the symbols to which @code{interactive}
3471 binds the prefix argument and the specified character:
3475 (defun @var{name-of-function} (arg char)
3476 "@var{documentation}@dots{}"
3477 (interactive "p\ncZap to char: ")
3478 @var{body-of-function}@dots{})
3483 (The space after the colon in the prompt makes it look better when you
3484 are prompted. @xref{copy-to-buffer, , The Definition of
3485 @code{copy-to-buffer}}, for an example.)
3487 When a function does not take arguments, @code{interactive} does not
3488 require any. Such a function contains the simple expression
3489 @code{(interactive)}. The @code{mark-whole-buffer} function is like
3492 Alternatively, if the special letter-codes are not right for your
3493 application, you can pass your own arguments to @code{interactive} as
3496 @xref{append-to-buffer, , The Definition of @code{append-to-buffer}},
3497 for an example. @xref{Using Interactive, , Using @code{Interactive},
3498 elisp, The GNU Emacs Lisp Reference Manual}, for a more complete
3499 explanation about this technique.
3501 @node Permanent Installation
3502 @section Install Code Permanently
3503 @cindex Install code permanently
3504 @cindex Permanent code installation
3505 @cindex Code installation
3507 When you install a function definition by evaluating it, it will stay
3508 installed until you quit Emacs. The next time you start a new session
3509 of Emacs, the function will not be installed unless you evaluate the
3510 function definition again.
3512 At some point, you may want to have code installed automatically
3513 whenever you start a new session of Emacs. There are several ways of
3518 If you have code that is just for yourself, you can put the code for the
3519 function definition in your @file{.emacs} initialization file. When you
3520 start Emacs, your @file{.emacs} file is automatically evaluated and all
3521 the function definitions within it are installed.
3522 @xref{Emacs Initialization, , Your @file{.emacs} File}.
3525 Alternatively, you can put the function definitions that you want
3526 installed in one or more files of their own and use the @code{load}
3527 function to cause Emacs to evaluate and thereby install each of the
3528 functions in the files.
3529 @xref{Loading Files, , Loading Files}.
3532 Thirdly, if you have code that your whole site will use, it is usual
3533 to put it in a file called @file{site-init.el} that is loaded when
3534 Emacs is built. This makes the code available to everyone who uses
3535 your machine. (See the @file{INSTALL} file that is part of the Emacs
3539 Finally, if you have code that everyone who uses Emacs may want, you
3540 can post it on a computer network or send a copy to the Free Software
3541 Foundation. (When you do this, please license the code and its
3542 documentation under a license that permits other people to run, copy,
3543 study, modify, and redistribute the code and which protects you from
3544 having your work taken from you.) If you send a copy of your code to
3545 the Free Software Foundation, and properly protect yourself and
3546 others, it may be included in the next release of Emacs. In large
3547 part, this is how Emacs has grown over the past years, by donations.
3553 The @code{let} expression is a special form in Lisp that you will need
3554 to use in most function definitions.
3556 @code{let} is used to attach or bind a symbol to a value in such a way
3557 that the Lisp interpreter will not confuse the variable with a
3558 variable of the same name that is not part of the function.
3560 To understand why the @code{let} special form is necessary, consider
3561 the situation in which you own a home that you generally refer to as
3562 `the house', as in the sentence, ``The house needs painting.'' If you
3563 are visiting a friend and your host refers to `the house', he is
3564 likely to be referring to @emph{his} house, not yours, that is, to a
3567 If your friend is referring to his house and you think he is referring
3568 to your house, you may be in for some confusion. The same thing could
3569 happen in Lisp if a variable that is used inside of one function has
3570 the same name as a variable that is used inside of another function,
3571 and the two are not intended to refer to the same value. The
3572 @code{let} special form prevents this kind of confusion.
3575 * Prevent confusion::
3576 * Parts of let Expression::
3577 * Sample let Expression::
3578 * Uninitialized let Variables::
3582 @node Prevent confusion
3583 @unnumberedsubsec @code{let} Prevents Confusion
3586 @cindex @samp{local variable} defined
3587 @cindex @samp{variable, local}, defined
3588 The @code{let} special form prevents confusion. @code{let} creates a
3589 name for a @dfn{local variable} that overshadows any use of the same
3590 name outside the @code{let} expression. This is like understanding
3591 that whenever your host refers to `the house', he means his house, not
3592 yours. (Symbols used in argument lists work the same way.
3593 @xref{defun, , The @code{defun} Macro}.)
3595 Local variables created by a @code{let} expression retain their value
3596 @emph{only} within the @code{let} expression itself (and within
3597 expressions called within the @code{let} expression); the local
3598 variables have no effect outside the @code{let} expression.
3600 Another way to think about @code{let} is that it is like a @code{setq}
3601 that is temporary and local. The values set by @code{let} are
3602 automatically undone when the @code{let} is finished. The setting
3603 only affects expressions that are inside the bounds of the @code{let}
3604 expression. In computer science jargon, we would say ``the binding of
3605 a symbol is visible only in functions called in the @code{let} form;
3606 in Emacs Lisp, scoping is dynamic, not lexical.''
3608 @code{let} can create more than one variable at once. Also,
3609 @code{let} gives each variable it creates an initial value, either a
3610 value specified by you, or @code{nil}. (In the jargon, this is called
3611 `binding the variable to the value'.) After @code{let} has created
3612 and bound the variables, it executes the code in the body of the
3613 @code{let}, and returns the value of the last expression in the body,
3614 as the value of the whole @code{let} expression. (`Execute' is a jargon
3615 term that means to evaluate a list; it comes from the use of the word
3616 meaning `to give practical effect to' (@cite{Oxford English
3617 Dictionary}). Since you evaluate an expression to perform an action,
3618 `execute' has evolved as a synonym to `evaluate'.)
3620 @node Parts of let Expression
3621 @subsection The Parts of a @code{let} Expression
3622 @cindex @code{let} expression, parts of
3623 @cindex Parts of @code{let} expression
3625 @cindex @samp{varlist} defined
3626 A @code{let} expression is a list of three parts. The first part is
3627 the symbol @code{let}. The second part is a list, called a
3628 @dfn{varlist}, each element of which is either a symbol by itself or a
3629 two-element list, the first element of which is a symbol. The third
3630 part of the @code{let} expression is the body of the @code{let}. The
3631 body usually consists of one or more lists.
3634 A template for a @code{let} expression looks like this:
3637 (let @var{varlist} @var{body}@dots{})
3641 The symbols in the varlist are the variables that are given initial
3642 values by the @code{let} special form. Symbols by themselves are given
3643 the initial value of @code{nil}; and each symbol that is the first
3644 element of a two-element list is bound to the value that is returned
3645 when the Lisp interpreter evaluates the second element.
3647 Thus, a varlist might look like this: @code{(thread (needles 3))}. In
3648 this case, in a @code{let} expression, Emacs binds the symbol
3649 @code{thread} to an initial value of @code{nil}, and binds the symbol
3650 @code{needles} to an initial value of 3.
3652 When you write a @code{let} expression, what you do is put the
3653 appropriate expressions in the slots of the @code{let} expression
3656 If the varlist is composed of two-element lists, as is often the case,
3657 the template for the @code{let} expression looks like this:
3661 (let ((@var{variable} @var{value})
3662 (@var{variable} @var{value})
3668 @node Sample let Expression
3669 @subsection Sample @code{let} Expression
3670 @cindex Sample @code{let} expression
3671 @cindex @code{let} expression sample
3673 The following expression creates and gives initial values
3674 to the two variables @code{zebra} and @code{tiger}. The body of the
3675 @code{let} expression is a list which calls the @code{message} function.
3679 (let ((zebra 'stripes)
3681 (message "One kind of animal has %s and another is %s."
3686 Here, the varlist is @code{((zebra 'stripes) (tiger 'fierce))}.
3688 The two variables are @code{zebra} and @code{tiger}. Each variable is
3689 the first element of a two-element list and each value is the second
3690 element of its two-element list. In the varlist, Emacs binds the
3691 variable @code{zebra} to the value @code{stripes}@footnote{According
3692 to Jared Diamond in @cite{Guns, Germs, and Steel}, ``@dots{} zebras
3693 become impossibly dangerous as they grow older'' but the claim here is
3694 that they do not become fierce like a tiger. (1997, W. W. Norton and
3695 Co., ISBN 0-393-03894-2, page 171)}, and binds the
3696 variable @code{tiger} to the value @code{fierce}. In this example,
3697 both values are symbols preceded by a quote. The values could just as
3698 well have been another list or a string. The body of the @code{let}
3699 follows after the list holding the variables. In this example, the
3700 body is a list that uses the @code{message} function to print a string
3704 You may evaluate the example in the usual fashion, by placing the
3705 cursor after the last parenthesis and typing @kbd{C-x C-e}. When you do
3706 this, the following will appear in the echo area:
3709 "One kind of animal has stripes and another is fierce."
3712 As we have seen before, the @code{message} function prints its first
3713 argument, except for @samp{%s}. In this example, the value of the variable
3714 @code{zebra} is printed at the location of the first @samp{%s} and the
3715 value of the variable @code{tiger} is printed at the location of the
3718 @node Uninitialized let Variables
3719 @subsection Uninitialized Variables in a @code{let} Statement
3720 @cindex Uninitialized @code{let} variables
3721 @cindex @code{let} variables uninitialized
3723 If you do not bind the variables in a @code{let} statement to specific
3724 initial values, they will automatically be bound to an initial value of
3725 @code{nil}, as in the following expression:
3734 "Here are %d variables with %s, %s, and %s value."
3735 birch pine fir oak))
3740 Here, the varlist is @code{((birch 3) pine fir (oak 'some))}.
3743 If you evaluate this expression in the usual way, the following will
3744 appear in your echo area:
3747 "Here are 3 variables with nil, nil, and some value."
3751 In this example, Emacs binds the symbol @code{birch} to the number 3,
3752 binds the symbols @code{pine} and @code{fir} to @code{nil}, and binds
3753 the symbol @code{oak} to the value @code{some}.
3755 Note that in the first part of the @code{let}, the variables @code{pine}
3756 and @code{fir} stand alone as atoms that are not surrounded by
3757 parentheses; this is because they are being bound to @code{nil}, the
3758 empty list. But @code{oak} is bound to @code{some} and so is a part of
3759 the list @code{(oak 'some)}. Similarly, @code{birch} is bound to the
3760 number 3 and so is in a list with that number. (Since a number
3761 evaluates to itself, the number does not need to be quoted. Also, the
3762 number is printed in the message using a @samp{%d} rather than a
3763 @samp{%s}.) The four variables as a group are put into a list to
3764 delimit them from the body of the @code{let}.
3767 @section The @code{if} Special Form
3769 @cindex Conditional with @code{if}
3771 A third special form, in addition to @code{defun} and @code{let}, is the
3772 conditional @code{if}. This form is used to instruct the computer to
3773 make decisions. You can write function definitions without using
3774 @code{if}, but it is used often enough, and is important enough, to be
3775 included here. It is used, for example, in the code for the
3776 function @code{beginning-of-buffer}.
3778 The basic idea behind an @code{if}, is that ``@emph{if} a test is true,
3779 @emph{then} an expression is evaluated.'' If the test is not true, the
3780 expression is not evaluated. For example, you might make a decision
3781 such as, ``if it is warm and sunny, then go to the beach!''
3784 * if in more detail::
3785 * type-of-animal in detail:: An example of an @code{if} expression.
3789 @node if in more detail
3790 @unnumberedsubsec @code{if} in more detail
3793 @cindex @samp{if-part} defined
3794 @cindex @samp{then-part} defined
3795 An @code{if} expression written in Lisp does not use the word `then';
3796 the test and the action are the second and third elements of the list
3797 whose first element is @code{if}. Nonetheless, the test part of an
3798 @code{if} expression is often called the @dfn{if-part} and the second
3799 argument is often called the @dfn{then-part}.
3801 Also, when an @code{if} expression is written, the true-or-false-test
3802 is usually written on the same line as the symbol @code{if}, but the
3803 action to carry out if the test is true, the ``then-part'', is written
3804 on the second and subsequent lines. This makes the @code{if}
3805 expression easier to read.
3809 (if @var{true-or-false-test}
3810 @var{action-to-carry-out-if-test-is-true})
3815 The true-or-false-test will be an expression that
3816 is evaluated by the Lisp interpreter.
3818 Here is an example that you can evaluate in the usual manner. The test
3819 is whether the number 5 is greater than the number 4. Since it is, the
3820 message @samp{5 is greater than 4!} will be printed.
3824 (if (> 5 4) ; @r{if-part}
3825 (message "5 is greater than 4!")) ; @r{then-part}
3830 (The function @code{>} tests whether its first argument is greater than
3831 its second argument and returns true if it is.)
3832 @findex > (greater than)
3834 Of course, in actual use, the test in an @code{if} expression will not
3835 be fixed for all time as it is by the expression @code{(> 5 4)}.
3836 Instead, at least one of the variables used in the test will be bound to
3837 a value that is not known ahead of time. (If the value were known ahead
3838 of time, we would not need to run the test!)
3840 For example, the value may be bound to an argument of a function
3841 definition. In the following function definition, the character of the
3842 animal is a value that is passed to the function. If the value bound to
3843 @code{characteristic} is @code{fierce}, then the message, @samp{It's a
3844 tiger!} will be printed; otherwise, @code{nil} will be returned.
3848 (defun type-of-animal (characteristic)
3849 "Print message in echo area depending on CHARACTERISTIC.
3850 If the CHARACTERISTIC is the symbol `fierce',
3851 then warn of a tiger."
3852 (if (equal characteristic 'fierce)
3853 (message "It's a tiger!")))
3859 If you are reading this inside of GNU Emacs, you can evaluate the
3860 function definition in the usual way to install it in Emacs, and then you
3861 can evaluate the following two expressions to see the results:
3865 (type-of-animal 'fierce)
3867 (type-of-animal 'zebra)
3872 @c Following sentences rewritten to prevent overfull hbox.
3874 When you evaluate @code{(type-of-animal 'fierce)}, you will see the
3875 following message printed in the echo area: @code{"It's a tiger!"}; and
3876 when you evaluate @code{(type-of-animal 'zebra)} you will see @code{nil}
3877 printed in the echo area.
3879 @node type-of-animal in detail
3880 @subsection The @code{type-of-animal} Function in Detail
3882 Let's look at the @code{type-of-animal} function in detail.
3884 The function definition for @code{type-of-animal} was written by filling
3885 the slots of two templates, one for a function definition as a whole, and
3886 a second for an @code{if} expression.
3889 The template for every function that is not interactive is:
3893 (defun @var{name-of-function} (@var{argument-list})
3894 "@var{documentation}@dots{}"
3900 The parts of the function that match this template look like this:
3904 (defun type-of-animal (characteristic)
3905 "Print message in echo area depending on CHARACTERISTIC.
3906 If the CHARACTERISTIC is the symbol `fierce',
3907 then warn of a tiger."
3908 @var{body: the} @code{if} @var{expression})
3912 The name of function is @code{type-of-animal}; it is passed the value
3913 of one argument. The argument list is followed by a multi-line
3914 documentation string. The documentation string is included in the
3915 example because it is a good habit to write documentation string for
3916 every function definition. The body of the function definition
3917 consists of the @code{if} expression.
3920 The template for an @code{if} expression looks like this:
3924 (if @var{true-or-false-test}
3925 @var{action-to-carry-out-if-the-test-returns-true})
3930 In the @code{type-of-animal} function, the code for the @code{if}
3935 (if (equal characteristic 'fierce)
3936 (message "It's a tiger!")))
3941 Here, the true-or-false-test is the expression:
3944 (equal characteristic 'fierce)
3948 In Lisp, @code{equal} is a function that determines whether its first
3949 argument is equal to its second argument. The second argument is the
3950 quoted symbol @code{'fierce} and the first argument is the value of the
3951 symbol @code{characteristic}---in other words, the argument passed to
3954 In the first exercise of @code{type-of-animal}, the argument
3955 @code{fierce} is passed to @code{type-of-animal}. Since @code{fierce}
3956 is equal to @code{fierce}, the expression, @code{(equal characteristic
3957 'fierce)}, returns a value of true. When this happens, the @code{if}
3958 evaluates the second argument or then-part of the @code{if}:
3959 @code{(message "It's tiger!")}.
3961 On the other hand, in the second exercise of @code{type-of-animal}, the
3962 argument @code{zebra} is passed to @code{type-of-animal}. @code{zebra}
3963 is not equal to @code{fierce}, so the then-part is not evaluated and
3964 @code{nil} is returned by the @code{if} expression.
3967 @section If--then--else Expressions
3970 An @code{if} expression may have an optional third argument, called
3971 the @dfn{else-part}, for the case when the true-or-false-test returns
3972 false. When this happens, the second argument or then-part of the
3973 overall @code{if} expression is @emph{not} evaluated, but the third or
3974 else-part @emph{is} evaluated. You might think of this as the cloudy
3975 day alternative for the decision ``if it is warm and sunny, then go to
3976 the beach, else read a book!''.
3978 The word ``else'' is not written in the Lisp code; the else-part of an
3979 @code{if} expression comes after the then-part. In the written Lisp, the
3980 else-part is usually written to start on a line of its own and is
3981 indented less than the then-part:
3985 (if @var{true-or-false-test}
3986 @var{action-to-carry-out-if-the-test-returns-true}
3987 @var{action-to-carry-out-if-the-test-returns-false})
3991 For example, the following @code{if} expression prints the message @samp{4
3992 is not greater than 5!} when you evaluate it in the usual way:
3996 (if (> 4 5) ; @r{if-part}
3997 (message "4 falsely greater than 5!") ; @r{then-part}
3998 (message "4 is not greater than 5!")) ; @r{else-part}
4003 Note that the different levels of indentation make it easy to
4004 distinguish the then-part from the else-part. (GNU Emacs has several
4005 commands that automatically indent @code{if} expressions correctly.
4006 @xref{Typing Lists, , GNU Emacs Helps You Type Lists}.)
4008 We can extend the @code{type-of-animal} function to include an
4009 else-part by simply incorporating an additional part to the @code{if}
4013 You can see the consequences of doing this if you evaluate the following
4014 version of the @code{type-of-animal} function definition to install it
4015 and then evaluate the two subsequent expressions to pass different
4016 arguments to the function.
4020 (defun type-of-animal (characteristic) ; @r{Second version.}
4021 "Print message in echo area depending on CHARACTERISTIC.
4022 If the CHARACTERISTIC is the symbol `fierce',
4023 then warn of a tiger;
4024 else say it's not fierce."
4025 (if (equal characteristic 'fierce)
4026 (message "It's a tiger!")
4027 (message "It's not fierce!")))
4034 (type-of-animal 'fierce)
4036 (type-of-animal 'zebra)
4041 @c Following sentence rewritten to prevent overfull hbox.
4043 When you evaluate @code{(type-of-animal 'fierce)}, you will see the
4044 following message printed in the echo area: @code{"It's a tiger!"}; but
4045 when you evaluate @code{(type-of-animal 'zebra)}, you will see
4046 @code{"It's not fierce!"}.
4048 (Of course, if the @var{characteristic} were @code{ferocious}, the
4049 message @code{"It's not fierce!"} would be printed; and it would be
4050 misleading! When you write code, you need to take into account the
4051 possibility that some such argument will be tested by the @code{if}
4052 and write your program accordingly.)
4054 @node Truth & Falsehood
4055 @section Truth and Falsehood in Emacs Lisp
4056 @cindex Truth and falsehood in Emacs Lisp
4057 @cindex Falsehood and truth in Emacs Lisp
4060 There is an important aspect to the truth test in an @code{if}
4061 expression. So far, we have spoken of `true' and `false' as values of
4062 predicates as if they were new kinds of Emacs Lisp objects. In fact,
4063 `false' is just our old friend @code{nil}. Anything else---anything
4066 The expression that tests for truth is interpreted as @dfn{true}
4067 if the result of evaluating it is a value that is not @code{nil}. In
4068 other words, the result of the test is considered true if the value
4069 returned is a number such as 47, a string such as @code{"hello"}, or a
4070 symbol (other than @code{nil}) such as @code{flowers}, or a list (so
4071 long as it is not empty), or even a buffer!
4074 * nil explained:: @code{nil} has two meanings.
4079 @unnumberedsubsec An explanation of @code{nil}
4082 Before illustrating a test for truth, we need an explanation of @code{nil}.
4084 In Emacs Lisp, the symbol @code{nil} has two meanings. First, it means the
4085 empty list. Second, it means false and is the value returned when a
4086 true-or-false-test tests false. @code{nil} can be written as an empty
4087 list, @code{()}, or as @code{nil}. As far as the Lisp interpreter is
4088 concerned, @code{()} and @code{nil} are the same. Humans, however, tend
4089 to use @code{nil} for false and @code{()} for the empty list.
4091 In Emacs Lisp, any value that is not @code{nil}---is not the empty
4092 list---is considered true. This means that if an evaluation returns
4093 something that is not an empty list, an @code{if} expression will test
4094 true. For example, if a number is put in the slot for the test, it
4095 will be evaluated and will return itself, since that is what numbers
4096 do when evaluated. In this conditional, the @code{if} expression will
4097 test true. The expression tests false only when @code{nil}, an empty
4098 list, is returned by evaluating the expression.
4100 You can see this by evaluating the two expressions in the following examples.
4102 In the first example, the number 4 is evaluated as the test in the
4103 @code{if} expression and returns itself; consequently, the then-part
4104 of the expression is evaluated and returned: @samp{true} appears in
4105 the echo area. In the second example, the @code{nil} indicates false;
4106 consequently, the else-part of the expression is evaluated and
4107 returned: @samp{false} appears in the echo area.
4124 Incidentally, if some other useful value is not available for a test that
4125 returns true, then the Lisp interpreter will return the symbol @code{t}
4126 for true. For example, the expression @code{(> 5 4)} returns @code{t}
4127 when evaluated, as you can see by evaluating it in the usual way:
4135 On the other hand, this function returns @code{nil} if the test is false.
4141 @node save-excursion
4142 @section @code{save-excursion}
4143 @findex save-excursion
4144 @cindex Region, what it is
4145 @cindex Preserving point, mark, and buffer
4146 @cindex Point, mark, buffer preservation
4150 The @code{save-excursion} function is the third and final special form
4151 that we will discuss in this chapter.
4153 In Emacs Lisp programs used for editing, the @code{save-excursion}
4154 function is very common. It saves the location of point and mark,
4155 executes the body of the function, and then restores point and mark to
4156 their previous positions if their locations were changed. Its primary
4157 purpose is to keep the user from being surprised and disturbed by
4158 unexpected movement of point or mark.
4161 * Point and mark:: A review of various locations.
4162 * Template for save-excursion::
4166 @node Point and mark
4167 @unnumberedsubsec Point and Mark
4170 Before discussing @code{save-excursion}, however, it may be useful
4171 first to review what point and mark are in GNU Emacs. @dfn{Point} is
4172 the current location of the cursor. Wherever the cursor
4173 is, that is point. More precisely, on terminals where the cursor
4174 appears to be on top of a character, point is immediately before the
4175 character. In Emacs Lisp, point is an integer. The first character in
4176 a buffer is number one, the second is number two, and so on. The
4177 function @code{point} returns the current position of the cursor as a
4178 number. Each buffer has its own value for point.
4180 The @dfn{mark} is another position in the buffer; its value can be set
4181 with a command such as @kbd{C-@key{SPC}} (@code{set-mark-command}). If
4182 a mark has been set, you can use the command @kbd{C-x C-x}
4183 (@code{exchange-point-and-mark}) to cause the cursor to jump to the mark
4184 and set the mark to be the previous position of point. In addition, if
4185 you set another mark, the position of the previous mark is saved in the
4186 mark ring. Many mark positions can be saved this way. You can jump the
4187 cursor to a saved mark by typing @kbd{C-u C-@key{SPC}} one or more
4190 The part of the buffer between point and mark is called @dfn{the
4191 region}. Numerous commands work on the region, including
4192 @code{center-region}, @code{count-lines-region}, @code{kill-region}, and
4193 @code{print-region}.
4195 The @code{save-excursion} special form saves the locations of point and
4196 mark and restores those positions after the code within the body of the
4197 special form is evaluated by the Lisp interpreter. Thus, if point were
4198 in the beginning of a piece of text and some code moved point to the end
4199 of the buffer, the @code{save-excursion} would put point back to where
4200 it was before, after the expressions in the body of the function were
4203 In Emacs, a function frequently moves point as part of its internal
4204 workings even though a user would not expect this. For example,
4205 @code{count-lines-region} moves point. To prevent the user from being
4206 bothered by jumps that are both unexpected and (from the user's point of
4207 view) unnecessary, @code{save-excursion} is often used to keep point and
4208 mark in the location expected by the user. The use of
4209 @code{save-excursion} is good housekeeping.
4211 To make sure the house stays clean, @code{save-excursion} restores the
4212 values of point and mark even if something goes wrong in the code inside
4213 of it (or, to be more precise and to use the proper jargon, ``in case of
4214 abnormal exit''). This feature is very helpful.
4216 In addition to recording the values of point and mark,
4217 @code{save-excursion} keeps track of the current buffer, and restores
4218 it, too. This means you can write code that will change the buffer and
4219 have @code{save-excursion} switch you back to the original buffer.
4220 This is how @code{save-excursion} is used in @code{append-to-buffer}.
4221 (@xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.)
4223 @node Template for save-excursion
4224 @subsection Template for a @code{save-excursion} Expression
4227 The template for code using @code{save-excursion} is simple:
4237 The body of the function is one or more expressions that will be
4238 evaluated in sequence by the Lisp interpreter. If there is more than
4239 one expression in the body, the value of the last one will be returned
4240 as the value of the @code{save-excursion} function. The other
4241 expressions in the body are evaluated only for their side effects; and
4242 @code{save-excursion} itself is used only for its side effect (which
4243 is restoring the positions of point and mark).
4246 In more detail, the template for a @code{save-excursion} expression
4252 @var{first-expression-in-body}
4253 @var{second-expression-in-body}
4254 @var{third-expression-in-body}
4256 @var{last-expression-in-body})
4261 An expression, of course, may be a symbol on its own or a list.
4263 In Emacs Lisp code, a @code{save-excursion} expression often occurs
4264 within the body of a @code{let} expression. It looks like this:
4277 In the last few chapters we have introduced a macro and a fair number
4278 of functions and special forms. Here they are described in brief,
4279 along with a few similar functions that have not been mentioned yet.
4282 @item eval-last-sexp
4283 Evaluate the last symbolic expression before the current location of
4284 point. The value is printed in the echo area unless the function is
4285 invoked with an argument; in that case, the output is printed in the
4286 current buffer. This command is normally bound to @kbd{C-x C-e}.
4289 Define function. This macro has up to five parts: the name, a
4290 template for the arguments that will be passed to the function,
4291 documentation, an optional interactive declaration, and the body of
4295 For example, in an early version of Emacs, the function definition was
4296 as follows. (It is slightly more complex now that it seeks the first
4297 non-whitespace character rather than the first visible character.)
4301 (defun back-to-indentation ()
4302 "Move point to first visible character on line."
4304 (beginning-of-line 1)
4305 (skip-chars-forward " \t"))
4312 (defun backward-to-indentation (&optional arg)
4313 "Move backward ARG lines and position at first nonblank character."
4315 (forward-line (- (or arg 1)))
4316 (skip-chars-forward " \t"))
4318 (defun back-to-indentation ()
4319 "Move point to the first non-whitespace character on this line."
4321 (beginning-of-line 1)
4322 (skip-syntax-forward " " (line-end-position))
4323 ;; Move back over chars that have whitespace syntax but have the p flag.
4324 (backward-prefix-chars))
4328 Declare to the interpreter that the function can be used
4329 interactively. This special form may be followed by a string with one
4330 or more parts that pass the information to the arguments of the
4331 function, in sequence. These parts may also tell the interpreter to
4332 prompt for information. Parts of the string are separated by
4333 newlines, @samp{\n}.
4336 Common code characters are:
4340 The name of an existing buffer.
4343 The name of an existing file.
4346 The numeric prefix argument. (Note that this `p' is lower case.)
4349 Point and the mark, as two numeric arguments, smallest first. This
4350 is the only code letter that specifies two successive arguments
4354 @xref{Interactive Codes, , Code Characters for @samp{interactive},
4355 elisp, The GNU Emacs Lisp Reference Manual}, for a complete list of
4359 Declare that a list of variables is for use within the body of the
4360 @code{let} and give them an initial value, either @code{nil} or a
4361 specified value; then evaluate the rest of the expressions in the body
4362 of the @code{let} and return the value of the last one. Inside the
4363 body of the @code{let}, the Lisp interpreter does not see the values of
4364 the variables of the same names that are bound outside of the
4372 (let ((foo (buffer-name))
4373 (bar (buffer-size)))
4375 "This buffer is %s and has %d characters."
4380 @item save-excursion
4381 Record the values of point and mark and the current buffer before
4382 evaluating the body of this special form. Restore the values of point
4383 and mark and buffer afterward.
4390 (message "We are %d characters into this buffer."
4393 (goto-char (point-min)) (point))))
4398 Evaluate the first argument to the function; if it is true, evaluate
4399 the second argument; else evaluate the third argument, if there is one.
4401 The @code{if} special form is called a @dfn{conditional}. There are
4402 other conditionals in Emacs Lisp, but @code{if} is perhaps the most
4410 (if (= 22 emacs-major-version)
4411 (message "This is version 22 Emacs")
4412 (message "This is not version 22 Emacs"))
4421 The @code{<} function tests whether its first argument is smaller than
4422 its second argument. A corresponding function, @code{>}, tests whether
4423 the first argument is greater than the second. Likewise, @code{<=}
4424 tests whether the first argument is less than or equal to the second and
4425 @code{>=} tests whether the first argument is greater than or equal to
4426 the second. In all cases, both arguments must be numbers or markers
4427 (markers indicate positions in buffers).
4431 The @code{=} function tests whether two arguments, both numbers or
4437 Test whether two objects are the same. @code{equal} uses one meaning
4438 of the word `same' and @code{eq} uses another: @code{equal} returns
4439 true if the two objects have a similar structure and contents, such as
4440 two copies of the same book. On the other hand, @code{eq}, returns
4441 true if both arguments are actually the same object.
4450 The @code{string-lessp} function tests whether its first argument is
4451 smaller than the second argument. A shorter, alternative name for the
4452 same function (a @code{defalias}) is @code{string<}.
4454 The arguments to @code{string-lessp} must be strings or symbols; the
4455 ordering is lexicographic, so case is significant. The print names of
4456 symbols are used instead of the symbols themselves.
4458 @cindex @samp{empty string} defined
4459 An empty string, @samp{""}, a string with no characters in it, is
4460 smaller than any string of characters.
4462 @code{string-equal} provides the corresponding test for equality. Its
4463 shorter, alternative name is @code{string=}. There are no string test
4464 functions that correspond to @var{>}, @code{>=}, or @code{<=}.
4467 Print a message in the echo area. The first argument is a string that
4468 can contain @samp{%s}, @samp{%d}, or @samp{%c} to print the value of
4469 arguments that follow the string. The argument used by @samp{%s} must
4470 be a string or a symbol; the argument used by @samp{%d} must be a
4471 number. The argument used by @samp{%c} must be an @sc{ascii} code
4472 number; it will be printed as the character with that @sc{ascii} code.
4473 (Various other %-sequences have not been mentioned.)
4477 The @code{setq} function sets the value of its first argument to the
4478 value of the second argument. The first argument is automatically
4479 quoted by @code{setq}. It does the same for succeeding pairs of
4480 arguments. Another function, @code{set}, takes only two arguments and
4481 evaluates both of them before setting the value returned by its first
4482 argument to the value returned by its second argument.
4485 Without an argument, return the name of the buffer, as a string.
4487 @item buffer-file-name
4488 Without an argument, return the name of the file the buffer is
4491 @item current-buffer
4492 Return the buffer in which Emacs is active; it may not be
4493 the buffer that is visible on the screen.
4496 Return the most recently selected buffer (other than the buffer passed
4497 to @code{other-buffer} as an argument and other than the current
4500 @item switch-to-buffer
4501 Select a buffer for Emacs to be active in and display it in the current
4502 window so users can look at it. Usually bound to @kbd{C-x b}.
4505 Switch Emacs's attention to a buffer on which programs will run. Don't
4506 alter what the window is showing.
4509 Return the number of characters in the current buffer.
4512 Return the value of the current position of the cursor, as an
4513 integer counting the number of characters from the beginning of the
4517 Return the minimum permissible value of point in
4518 the current buffer. This is 1, unless narrowing is in effect.
4521 Return the value of the maximum permissible value of point in the
4522 current buffer. This is the end of the buffer, unless narrowing is in
4527 @node defun Exercises
4532 Write a non-interactive function that doubles the value of its
4533 argument, a number. Make that function interactive.
4536 Write a function that tests whether the current value of
4537 @code{fill-column} is greater than the argument passed to the function,
4538 and if so, prints an appropriate message.
4541 @node Buffer Walk Through
4542 @chapter A Few Buffer--Related Functions
4544 In this chapter we study in detail several of the functions used in GNU
4545 Emacs. This is called a ``walk-through''. These functions are used as
4546 examples of Lisp code, but are not imaginary examples; with the
4547 exception of the first, simplified function definition, these functions
4548 show the actual code used in GNU Emacs. You can learn a great deal from
4549 these definitions. The functions described here are all related to
4550 buffers. Later, we will study other functions.
4553 * Finding More:: How to find more information.
4554 * simplified-beginning-of-buffer:: Shows @code{goto-char},
4555 @code{point-min}, and @code{push-mark}.
4556 * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}.
4557 * append-to-buffer:: Uses @code{save-excursion} and
4558 @code{insert-buffer-substring}.
4559 * Buffer Related Review:: Review.
4560 * Buffer Exercises::
4564 @section Finding More Information
4566 @findex describe-function, @r{introduced}
4567 @cindex Find function documentation
4568 In this walk-through, I will describe each new function as we come to
4569 it, sometimes in detail and sometimes briefly. If you are interested,
4570 you can get the full documentation of any Emacs Lisp function at any
4571 time by typing @kbd{C-h f} and then the name of the function (and then
4572 @key{RET}). Similarly, you can get the full documentation for a
4573 variable by typing @kbd{C-h v} and then the name of the variable (and
4576 @cindex Find source of function
4577 @c In version 22, tells location both of C and of Emacs Lisp
4578 Also, @code{describe-function} will tell you the location of the
4579 function definition.
4581 Put point into the name of the file that contains the function and
4582 press the @key{RET} key. In this case, @key{RET} means
4583 @code{push-button} rather than `return' or `enter'. Emacs will take
4584 you directly to the function definition.
4589 If you move point over the file name and press
4590 the @key{RET} key, which in this case means @code{help-follow} rather
4591 than `return' or `enter', Emacs will take you directly to the function
4595 More generally, if you want to see a function in its original source
4596 file, you can use the @code{find-tag} function to jump to it.
4597 @code{find-tag} works with a wide variety of languages, not just
4598 Lisp, and C, and it works with non-programming text as well. For
4599 example, @code{find-tag} will jump to the various nodes in the
4600 Texinfo source file of this document.
4601 The @code{find-tag} function depends on `tags tables' that record
4602 the locations of the functions, variables, and other items to which
4603 @code{find-tag} jumps.
4605 To use the @code{find-tag} command, type @kbd{M-.} (i.e., press the
4606 period key while holding down the @key{META} key, or else type the
4607 @key{ESC} key and then type the period key), and then, at the prompt,
4608 type in the name of the function whose source code you want to see,
4609 such as @code{mark-whole-buffer}, and then type @key{RET}. Emacs will
4610 switch buffers and display the source code for the function on your
4611 screen. To switch back to your current buffer, type @kbd{C-x b
4612 @key{RET}}. (On some keyboards, the @key{META} key is labeled
4615 @c !!! 22.1.1 tags table location in this paragraph
4616 @cindex TAGS table, specifying
4618 Depending on how the initial default values of your copy of Emacs are
4619 set, you may also need to specify the location of your `tags table',
4620 which is a file called @file{TAGS}. For example, if you are
4621 interested in Emacs sources, the tags table you will most likely want,
4622 if it has already been created for you, will be in a subdirectory of
4623 the @file{/usr/local/share/emacs/} directory; thus you would use the
4624 @code{M-x visit-tags-table} command and specify a pathname such as
4625 @file{/usr/local/share/emacs/22.1.1/lisp/TAGS}. If the tags table
4626 has not already been created, you will have to create it yourself. It
4627 will be in a file such as @file{/usr/local/src/emacs/src/TAGS}.
4630 To create a @file{TAGS} file in a specific directory, switch to that
4631 directory in Emacs using @kbd{M-x cd} command, or list the directory
4632 with @kbd{C-x d} (@code{dired}). Then run the compile command, with
4633 @w{@code{etags *.el}} as the command to execute:
4636 M-x compile RET etags *.el RET
4639 For more information, see @ref{etags, , Create Your Own @file{TAGS} File}.
4641 After you become more familiar with Emacs Lisp, you will find that you will
4642 frequently use @code{find-tag} to navigate your way around source code;
4643 and you will create your own @file{TAGS} tables.
4645 @cindex Library, as term for `file'
4646 Incidentally, the files that contain Lisp code are conventionally
4647 called @dfn{libraries}. The metaphor is derived from that of a
4648 specialized library, such as a law library or an engineering library,
4649 rather than a general library. Each library, or file, contains
4650 functions that relate to a particular topic or activity, such as
4651 @file{abbrev.el} for handling abbreviations and other typing
4652 shortcuts, and @file{help.el} for on-line help. (Sometimes several
4653 libraries provide code for a single activity, as the various
4654 @file{rmail@dots{}} files provide code for reading electronic mail.)
4655 In @cite{The GNU Emacs Manual}, you will see sentences such as ``The
4656 @kbd{C-h p} command lets you search the standard Emacs Lisp libraries
4657 by topic keywords.''
4659 @node simplified-beginning-of-buffer
4660 @section A Simplified @code{beginning-of-buffer} Definition
4661 @findex simplified-beginning-of-buffer
4663 The @code{beginning-of-buffer} command is a good function to start with
4664 since you are likely to be familiar with it and it is easy to
4665 understand. Used as an interactive command, @code{beginning-of-buffer}
4666 moves the cursor to the beginning of the buffer, leaving the mark at the
4667 previous position. It is generally bound to @kbd{M-<}.
4669 In this section, we will discuss a shortened version of the function
4670 that shows how it is most frequently used. This shortened function
4671 works as written, but it does not contain the code for a complex option.
4672 In another section, we will describe the entire function.
4673 (@xref{beginning-of-buffer, , Complete Definition of
4674 @code{beginning-of-buffer}}.)
4676 Before looking at the code, let's consider what the function
4677 definition has to contain: it must include an expression that makes
4678 the function interactive so it can be called by typing @kbd{M-x
4679 beginning-of-buffer} or by typing a keychord such as @kbd{M-<}; it
4680 must include code to leave a mark at the original position in the
4681 buffer; and it must include code to move the cursor to the beginning
4685 Here is the complete text of the shortened version of the function:
4689 (defun simplified-beginning-of-buffer ()
4690 "Move point to the beginning of the buffer;
4691 leave mark at previous position."
4694 (goto-char (point-min)))
4698 Like all function definitions, this definition has five parts following
4699 the macro @code{defun}:
4703 The name: in this example, @code{simplified-beginning-of-buffer}.
4706 A list of the arguments: in this example, an empty list, @code{()},
4709 The documentation string.
4712 The interactive expression.
4719 In this function definition, the argument list is empty; this means that
4720 this function does not require any arguments. (When we look at the
4721 definition for the complete function, we will see that it may be passed
4722 an optional argument.)
4724 The interactive expression tells Emacs that the function is intended to
4725 be used interactively. In this example, @code{interactive} does not have
4726 an argument because @code{simplified-beginning-of-buffer} does not
4730 The body of the function consists of the two lines:
4735 (goto-char (point-min))
4739 The first of these lines is the expression, @code{(push-mark)}. When
4740 this expression is evaluated by the Lisp interpreter, it sets a mark at
4741 the current position of the cursor, wherever that may be. The position
4742 of this mark is saved in the mark ring.
4744 The next line is @code{(goto-char (point-min))}. This expression
4745 jumps the cursor to the minimum point in the buffer, that is, to the
4746 beginning of the buffer (or to the beginning of the accessible portion
4747 of the buffer if it is narrowed. @xref{Narrowing & Widening, ,
4748 Narrowing and Widening}.)
4750 The @code{push-mark} command sets a mark at the place where the cursor
4751 was located before it was moved to the beginning of the buffer by the
4752 @code{(goto-char (point-min))} expression. Consequently, you can, if
4753 you wish, go back to where you were originally by typing @kbd{C-x C-x}.
4755 That is all there is to the function definition!
4757 @findex describe-function
4758 When you are reading code such as this and come upon an unfamiliar
4759 function, such as @code{goto-char}, you can find out what it does by
4760 using the @code{describe-function} command. To use this command, type
4761 @kbd{C-h f} and then type in the name of the function and press
4762 @key{RET}. The @code{describe-function} command will print the
4763 function's documentation string in a @file{*Help*} window. For
4764 example, the documentation for @code{goto-char} is:
4768 Set point to POSITION, a number or marker.
4769 Beginning of buffer is position (point-min), end is (point-max).
4774 The function's one argument is the desired position.
4777 (The prompt for @code{describe-function} will offer you the symbol
4778 under or preceding the cursor, so you can save typing by positioning
4779 the cursor right over or after the function and then typing @kbd{C-h f
4782 The @code{end-of-buffer} function definition is written in the same way as
4783 the @code{beginning-of-buffer} definition except that the body of the
4784 function contains the expression @code{(goto-char (point-max))} in place
4785 of @code{(goto-char (point-min))}.
4787 @node mark-whole-buffer
4788 @section The Definition of @code{mark-whole-buffer}
4789 @findex mark-whole-buffer
4791 The @code{mark-whole-buffer} function is no harder to understand than the
4792 @code{simplified-beginning-of-buffer} function. In this case, however,
4793 we will look at the complete function, not a shortened version.
4795 The @code{mark-whole-buffer} function is not as commonly used as the
4796 @code{beginning-of-buffer} function, but is useful nonetheless: it
4797 marks a whole buffer as a region by putting point at the beginning and
4798 a mark at the end of the buffer. It is generally bound to @kbd{C-x
4802 * mark-whole-buffer overview::
4803 * Body of mark-whole-buffer:: Only three lines of code.
4807 @node mark-whole-buffer overview
4808 @unnumberedsubsec An overview of @code{mark-whole-buffer}
4812 In GNU Emacs 22, the code for the complete function looks like this:
4816 (defun mark-whole-buffer ()
4817 "Put point at beginning and mark at end of buffer.
4818 You probably should not use this function in Lisp programs;
4819 it is usually a mistake for a Lisp function to use any subroutine
4820 that uses or sets the mark."
4823 (push-mark (point-max) nil t)
4824 (goto-char (point-min)))
4829 Like all other functions, the @code{mark-whole-buffer} function fits
4830 into the template for a function definition. The template looks like
4835 (defun @var{name-of-function} (@var{argument-list})
4836 "@var{documentation}@dots{}"
4837 (@var{interactive-expression}@dots{})
4842 Here is how the function works: the name of the function is
4843 @code{mark-whole-buffer}; it is followed by an empty argument list,
4844 @samp{()}, which means that the function does not require arguments.
4845 The documentation comes next.
4847 The next line is an @code{(interactive)} expression that tells Emacs
4848 that the function will be used interactively. These details are similar
4849 to the @code{simplified-beginning-of-buffer} function described in the
4853 @node Body of mark-whole-buffer
4854 @subsection Body of @code{mark-whole-buffer}
4856 The body of the @code{mark-whole-buffer} function consists of three
4863 (push-mark (point-max) nil t)
4864 (goto-char (point-min))
4868 The first of these lines is the expression, @code{(push-mark (point))}.
4870 This line does exactly the same job as the first line of the body of
4871 the @code{simplified-beginning-of-buffer} function, which is written
4872 @code{(push-mark)}. In both cases, the Lisp interpreter sets a mark
4873 at the current position of the cursor.
4875 I don't know why the expression in @code{mark-whole-buffer} is written
4876 @code{(push-mark (point))} and the expression in
4877 @code{beginning-of-buffer} is written @code{(push-mark)}. Perhaps
4878 whoever wrote the code did not know that the arguments for
4879 @code{push-mark} are optional and that if @code{push-mark} is not
4880 passed an argument, the function automatically sets mark at the
4881 location of point by default. Or perhaps the expression was written
4882 so as to parallel the structure of the next line. In any case, the
4883 line causes Emacs to determine the position of point and set a mark
4886 In earlier versions of GNU Emacs, the next line of
4887 @code{mark-whole-buffer} was @code{(push-mark (point-max))}. This
4888 expression sets a mark at the point in the buffer that has the highest
4889 number. This will be the end of the buffer (or, if the buffer is
4890 narrowed, the end of the accessible portion of the buffer.
4891 @xref{Narrowing & Widening, , Narrowing and Widening}, for more about
4892 narrowing.) After this mark has been set, the previous mark, the one
4893 set at point, is no longer set, but Emacs remembers its position, just
4894 as all other recent marks are always remembered. This means that you
4895 can, if you wish, go back to that position by typing @kbd{C-u
4899 In GNU Emacs 22, the @code{(point-max)} is slightly more complicated.
4903 (push-mark (point-max) nil t)
4907 The expression works nearly the same as before. It sets a mark at the
4908 highest numbered place in the buffer that it can. However, in this
4909 version, @code{push-mark} has two additional arguments. The second
4910 argument to @code{push-mark} is @code{nil}. This tells the function
4911 it @emph{should} display a message that says `Mark set' when it pushes
4912 the mark. The third argument is @code{t}. This tells
4913 @code{push-mark} to activate the mark when Transient Mark mode is
4914 turned on. Transient Mark mode highlights the currently active
4915 region. It is often turned off.
4917 Finally, the last line of the function is @code{(goto-char
4918 (point-min)))}. This is written exactly the same way as it is written
4919 in @code{beginning-of-buffer}. The expression moves the cursor to
4920 the minimum point in the buffer, that is, to the beginning of the buffer
4921 (or to the beginning of the accessible portion of the buffer). As a
4922 result of this, point is placed at the beginning of the buffer and mark
4923 is set at the end of the buffer. The whole buffer is, therefore, the
4926 @node append-to-buffer
4927 @section The Definition of @code{append-to-buffer}
4928 @findex append-to-buffer
4930 The @code{append-to-buffer} command is more complex than the
4931 @code{mark-whole-buffer} command. What it does is copy the region
4932 (that is, the part of the buffer between point and mark) from the
4933 current buffer to a specified buffer.
4936 * append-to-buffer overview::
4937 * append interactive:: A two part interactive expression.
4938 * append-to-buffer body:: Incorporates a @code{let} expression.
4939 * append save-excursion:: How the @code{save-excursion} works.
4943 @node append-to-buffer overview
4944 @unnumberedsubsec An Overview of @code{append-to-buffer}
4947 @findex insert-buffer-substring
4948 The @code{append-to-buffer} command uses the
4949 @code{insert-buffer-substring} function to copy the region.
4950 @code{insert-buffer-substring} is described by its name: it takes a
4951 string of characters from part of a buffer, a ``substring'', and
4952 inserts them into another buffer.
4954 Most of @code{append-to-buffer} is
4955 concerned with setting up the conditions for
4956 @code{insert-buffer-substring} to work: the code must specify both the
4957 buffer to which the text will go, the window it comes from and goes
4958 to, and the region that will be copied.
4961 Here is the complete text of the function:
4965 (defun append-to-buffer (buffer start end)
4966 "Append to specified buffer the text of the region.
4967 It is inserted into that buffer before its point.
4971 When calling from a program, give three arguments:
4972 BUFFER (or buffer name), START and END.
4973 START and END specify the portion of the current buffer to be copied."
4975 (list (read-buffer "Append to buffer: " (other-buffer
4976 (current-buffer) t))
4977 (region-beginning) (region-end)))
4980 (let ((oldbuf (current-buffer)))
4982 (let* ((append-to (get-buffer-create buffer))
4983 (windows (get-buffer-window-list append-to t t))
4985 (set-buffer append-to)
4986 (setq point (point))
4987 (barf-if-buffer-read-only)
4988 (insert-buffer-substring oldbuf start end)
4989 (dolist (window windows)
4990 (when (= (window-point window) point)
4991 (set-window-point window (point))))))))
4995 The function can be understood by looking at it as a series of
4996 filled-in templates.
4998 The outermost template is for the function definition. In this
4999 function, it looks like this (with several slots filled in):
5003 (defun append-to-buffer (buffer start end)
5004 "@var{documentation}@dots{}"
5005 (interactive @dots{})
5010 The first line of the function includes its name and three arguments.
5011 The arguments are the @code{buffer} to which the text will be copied, and
5012 the @code{start} and @code{end} of the region in the current buffer that
5015 The next part of the function is the documentation, which is clear and
5016 complete. As is conventional, the three arguments are written in
5017 upper case so you will notice them easily. Even better, they are
5018 described in the same order as in the argument list.
5020 Note that the documentation distinguishes between a buffer and its
5021 name. (The function can handle either.)
5023 @node append interactive
5024 @subsection The @code{append-to-buffer} Interactive Expression
5026 Since the @code{append-to-buffer} function will be used interactively,
5027 the function must have an @code{interactive} expression. (For a
5028 review of @code{interactive}, see @ref{Interactive, , Making a
5029 Function Interactive}.) The expression reads as follows:
5035 "Append to buffer: "
5036 (other-buffer (current-buffer) t))
5043 This expression is not one with letters standing for parts, as
5044 described earlier. Instead, it starts a list with these parts:
5046 The first part of the list is an expression to read the name of a
5047 buffer and return it as a string. That is @code{read-buffer}. The
5048 function requires a prompt as its first argument, @samp{"Append to
5049 buffer: "}. Its second argument tells the command what value to
5050 provide if you don't specify anything.
5052 In this case that second argument is an expression containing the
5053 function @code{other-buffer}, an exception, and a @samp{t}, standing
5056 The first argument to @code{other-buffer}, the exception, is yet
5057 another function, @code{current-buffer}. That is not going to be
5058 returned. The second argument is the symbol for true, @code{t}. that
5059 tells @code{other-buffer} that it may show visible buffers (except in
5060 this case, it will not show the current buffer, which makes sense).
5063 The expression looks like this:
5066 (other-buffer (current-buffer) t)
5069 The second and third arguments to the @code{list} expression are
5070 @code{(region-beginning)} and @code{(region-end)}. These two
5071 functions specify the beginning and end of the text to be appended.
5074 Originally, the command used the letters @samp{B} and @samp{r}.
5075 The whole @code{interactive} expression looked like this:
5078 (interactive "BAppend to buffer:@: \nr")
5082 But when that was done, the default value of the buffer switched to
5083 was invisible. That was not wanted.
5085 (The prompt was separated from the second argument with a newline,
5086 @samp{\n}. It was followed by an @samp{r} that told Emacs to bind the
5087 two arguments that follow the symbol @code{buffer} in the function's
5088 argument list (that is, @code{start} and @code{end}) to the values of
5089 point and mark. That argument worked fine.)
5091 @node append-to-buffer body
5092 @subsection The Body of @code{append-to-buffer}
5095 in GNU Emacs 22 in /usr/local/src/emacs/lisp/simple.el
5097 (defun append-to-buffer (buffer start end)
5098 "Append to specified buffer the text of the region.
5099 It is inserted into that buffer before its point.
5101 When calling from a program, give three arguments:
5102 BUFFER (or buffer name), START and END.
5103 START and END specify the portion of the current buffer to be copied."
5105 (list (read-buffer "Append to buffer: " (other-buffer (current-buffer) t))
5106 (region-beginning) (region-end)))
5107 (let ((oldbuf (current-buffer)))
5109 (let* ((append-to (get-buffer-create buffer))
5110 (windows (get-buffer-window-list append-to t t))
5112 (set-buffer append-to)
5113 (setq point (point))
5114 (barf-if-buffer-read-only)
5115 (insert-buffer-substring oldbuf start end)
5116 (dolist (window windows)
5117 (when (= (window-point window) point)
5118 (set-window-point window (point))))))))
5121 The body of the @code{append-to-buffer} function begins with @code{let}.
5123 As we have seen before (@pxref{let, , @code{let}}), the purpose of a
5124 @code{let} expression is to create and give initial values to one or
5125 more variables that will only be used within the body of the
5126 @code{let}. This means that such a variable will not be confused with
5127 any variable of the same name outside the @code{let} expression.
5129 We can see how the @code{let} expression fits into the function as a
5130 whole by showing a template for @code{append-to-buffer} with the
5131 @code{let} expression in outline:
5135 (defun append-to-buffer (buffer start end)
5136 "@var{documentation}@dots{}"
5137 (interactive @dots{})
5138 (let ((@var{variable} @var{value}))
5143 The @code{let} expression has three elements:
5147 The symbol @code{let};
5150 A varlist containing, in this case, a single two-element list,
5151 @code{(@var{variable} @var{value})};
5154 The body of the @code{let} expression.
5158 In the @code{append-to-buffer} function, the varlist looks like this:
5161 (oldbuf (current-buffer))
5165 In this part of the @code{let} expression, the one variable,
5166 @code{oldbuf}, is bound to the value returned by the
5167 @code{(current-buffer)} expression. The variable, @code{oldbuf}, is
5168 used to keep track of the buffer in which you are working and from
5169 which you will copy.
5171 The element or elements of a varlist are surrounded by a set of
5172 parentheses so the Lisp interpreter can distinguish the varlist from
5173 the body of the @code{let}. As a consequence, the two-element list
5174 within the varlist is surrounded by a circumscribing set of parentheses.
5175 The line looks like this:
5179 (let ((oldbuf (current-buffer)))
5185 The two parentheses before @code{oldbuf} might surprise you if you did
5186 not realize that the first parenthesis before @code{oldbuf} marks the
5187 boundary of the varlist and the second parenthesis marks the beginning
5188 of the two-element list, @code{(oldbuf (current-buffer))}.
5190 @node append save-excursion
5191 @subsection @code{save-excursion} in @code{append-to-buffer}
5193 The body of the @code{let} expression in @code{append-to-buffer}
5194 consists of a @code{save-excursion} expression.
5196 The @code{save-excursion} function saves the locations of point and
5197 mark, and restores them to those positions after the expressions in the
5198 body of the @code{save-excursion} complete execution. In addition,
5199 @code{save-excursion} keeps track of the original buffer, and
5200 restores it. This is how @code{save-excursion} is used in
5201 @code{append-to-buffer}.
5204 @cindex Indentation for formatting
5205 @cindex Formatting convention
5206 Incidentally, it is worth noting here that a Lisp function is normally
5207 formatted so that everything that is enclosed in a multi-line spread is
5208 indented more to the right than the first symbol. In this function
5209 definition, the @code{let} is indented more than the @code{defun}, and
5210 the @code{save-excursion} is indented more than the @code{let}, like
5226 This formatting convention makes it easy to see that the lines in
5227 the body of the @code{save-excursion} are enclosed by the parentheses
5228 associated with @code{save-excursion}, just as the
5229 @code{save-excursion} itself is enclosed by the parentheses associated
5230 with the @code{let}:
5234 (let ((oldbuf (current-buffer)))
5237 (set-buffer @dots{})
5238 (insert-buffer-substring oldbuf start end)
5244 The use of the @code{save-excursion} function can be viewed as a process
5245 of filling in the slots of a template:
5250 @var{first-expression-in-body}
5251 @var{second-expression-in-body}
5253 @var{last-expression-in-body})
5259 In this function, the body of the @code{save-excursion} contains only
5260 one expression, the @code{let*} expression. You know about a
5261 @code{let} function. The @code{let*} function is different. It has a
5262 @samp{*} in its name. It enables Emacs to set each variable in its
5263 varlist in sequence, one after another.
5265 Its critical feature is that variables later in the varlist can make
5266 use of the values to which Emacs set variables earlier in the varlist.
5267 @xref{fwd-para let, , The @code{let*} expression}.
5269 We will skip functions like @code{let*} and focus on two: the
5270 @code{set-buffer} function and the @code{insert-buffer-substring}
5274 In the old days, the @code{set-buffer} expression was simply
5277 (set-buffer (get-buffer-create buffer))
5285 (set-buffer append-to)
5289 @code{append-to} is bound to @code{(get-buffer-create buffer)} earlier
5290 on in the @code{let*} expression. That extra binding would not be
5291 necessary except for that @code{append-to} is used later in the
5292 varlist as an argument to @code{get-buffer-window-list}.
5297 (let ((oldbuf (current-buffer)))
5299 (let* ((append-to (get-buffer-create buffer))
5300 (windows (get-buffer-window-list append-to t t))
5302 (set-buffer append-to)
5303 (setq point (point))
5304 (barf-if-buffer-read-only)
5305 (insert-buffer-substring oldbuf start end)
5306 (dolist (window windows)
5307 (when (= (window-point window) point)
5308 (set-window-point window (point))))))))
5311 The @code{append-to-buffer} function definition inserts text from the
5312 buffer in which you are currently to a named buffer. It happens that
5313 @code{insert-buffer-substring} copies text from another buffer to the
5314 current buffer, just the reverse---that is why the
5315 @code{append-to-buffer} definition starts out with a @code{let} that
5316 binds the local symbol @code{oldbuf} to the value returned by
5317 @code{current-buffer}.
5320 The @code{insert-buffer-substring} expression looks like this:
5323 (insert-buffer-substring oldbuf start end)
5327 The @code{insert-buffer-substring} function copies a string
5328 @emph{from} the buffer specified as its first argument and inserts the
5329 string into the present buffer. In this case, the argument to
5330 @code{insert-buffer-substring} is the value of the variable created
5331 and bound by the @code{let}, namely the value of @code{oldbuf}, which
5332 was the current buffer when you gave the @code{append-to-buffer}
5335 After @code{insert-buffer-substring} has done its work,
5336 @code{save-excursion} will restore the action to the original buffer
5337 and @code{append-to-buffer} will have done its job.
5340 Written in skeletal form, the workings of the body look like this:
5344 (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer})
5345 (save-excursion ; @r{Keep track of buffer.}
5347 @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer})
5349 @var{change-back-to-original-buffer-when-finished}
5350 @var{let-the-local-meaning-of-}@code{oldbuf}@var{-disappear-when-finished}
5354 In summary, @code{append-to-buffer} works as follows: it saves the
5355 value of the current buffer in the variable called @code{oldbuf}. It
5356 gets the new buffer (creating one if need be) and switches Emacs's
5357 attention to it. Using the value of @code{oldbuf}, it inserts the
5358 region of text from the old buffer into the new buffer; and then using
5359 @code{save-excursion}, it brings you back to your original buffer.
5361 In looking at @code{append-to-buffer}, you have explored a fairly
5362 complex function. It shows how to use @code{let} and
5363 @code{save-excursion}, and how to change to and come back from another
5364 buffer. Many function definitions use @code{let},
5365 @code{save-excursion}, and @code{set-buffer} this way.
5367 @node Buffer Related Review
5370 Here is a brief summary of the various functions discussed in this chapter.
5373 @item describe-function
5374 @itemx describe-variable
5375 Print the documentation for a function or variable.
5376 Conventionally bound to @kbd{C-h f} and @kbd{C-h v}.
5379 Find the file containing the source for a function or variable and
5380 switch buffers to it, positioning point at the beginning of the item.
5381 Conventionally bound to @kbd{M-.} (that's a period following the
5384 @item save-excursion
5385 Save the location of point and mark and restore their values after the
5386 arguments to @code{save-excursion} have been evaluated. Also, remember
5387 the current buffer and return to it.
5390 Set mark at a location and record the value of the previous mark on the
5391 mark ring. The mark is a location in the buffer that will keep its
5392 relative position even if text is added to or removed from the buffer.
5395 Set point to the location specified by the value of the argument, which
5396 can be a number, a marker, or an expression that returns the number of
5397 a position, such as @code{(point-min)}.
5399 @item insert-buffer-substring
5400 Copy a region of text from a buffer that is passed to the function as
5401 an argument and insert the region into the current buffer.
5403 @item mark-whole-buffer
5404 Mark the whole buffer as a region. Normally bound to @kbd{C-x h}.
5407 Switch the attention of Emacs to another buffer, but do not change the
5408 window being displayed. Used when the program rather than a human is
5409 to work on a different buffer.
5411 @item get-buffer-create
5413 Find a named buffer or create one if a buffer of that name does not
5414 exist. The @code{get-buffer} function returns @code{nil} if the named
5415 buffer does not exist.
5419 @node Buffer Exercises
5424 Write your own @code{simplified-end-of-buffer} function definition;
5425 then test it to see whether it works.
5428 Use @code{if} and @code{get-buffer} to write a function that prints a
5429 message telling you whether a buffer exists.
5432 Using @code{find-tag}, find the source for the @code{copy-to-buffer}
5437 @chapter A Few More Complex Functions
5439 In this chapter, we build on what we have learned in previous chapters
5440 by looking at more complex functions. The @code{copy-to-buffer}
5441 function illustrates use of two @code{save-excursion} expressions in
5442 one definition, while the @code{insert-buffer} function illustrates
5443 use of an asterisk in an @code{interactive} expression, use of
5444 @code{or}, and the important distinction between a name and the object
5445 to which the name refers.
5448 * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}.
5449 * insert-buffer:: Read-only, and with @code{or}.
5450 * beginning-of-buffer:: Shows @code{goto-char},
5451 @code{point-min}, and @code{push-mark}.
5452 * Second Buffer Related Review::
5453 * optional Exercise::
5456 @node copy-to-buffer
5457 @section The Definition of @code{copy-to-buffer}
5458 @findex copy-to-buffer
5460 After understanding how @code{append-to-buffer} works, it is easy to
5461 understand @code{copy-to-buffer}. This function copies text into a
5462 buffer, but instead of adding to the second buffer, it replaces all the
5463 previous text in the second buffer.
5466 The body of @code{copy-to-buffer} looks like this,
5471 (interactive "BCopy to buffer: \nr")
5472 (let ((oldbuf (current-buffer)))
5473 (with-current-buffer (get-buffer-create buffer)
5474 (barf-if-buffer-read-only)
5477 (insert-buffer-substring oldbuf start end)))))
5481 The @code{copy-to-buffer} function has a simpler @code{interactive}
5482 expression than @code{append-to-buffer}.
5485 The definition then says
5488 (with-current-buffer (get-buffer-create buffer) @dots{}
5491 First, look at the earliest inner expression; that is evaluated first.
5492 That expression starts with @code{get-buffer-create buffer}. The
5493 function tells the computer to use the buffer with the name specified
5494 as the one to which you are copying, or if such a buffer does not
5495 exist, to create it. Then, the @code{with-current-buffer} function
5496 evaluates its body with that buffer temporarily current.
5498 (This demonstrates another way to shift the computer's attention but
5499 not the user's. The @code{append-to-buffer} function showed how to do
5500 the same with @code{save-excursion} and @code{set-buffer}.
5501 @code{with-current-buffer} is a newer, and arguably easier,
5504 The @code{barf-if-buffer-read-only} function sends you an error
5505 message saying the buffer is read-only if you cannot modify it.
5507 The next line has the @code{erase-buffer} function as its sole
5508 contents. That function erases the buffer.
5510 Finally, the last two lines contain the @code{save-excursion}
5511 expression with @code{insert-buffer-substring} as its body.
5512 The @code{insert-buffer-substring} expression copies the text from
5513 the buffer you are in (and you have not seen the computer shift its
5514 attention, so you don't know that that buffer is now called
5517 Incidentally, this is what is meant by `replacement'. To replace text,
5518 Emacs erases the previous text and then inserts new text.
5521 In outline, the body of @code{copy-to-buffer} looks like this:
5525 (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer})
5526 (@var{with-the-buffer-you-are-copying-to}
5527 (@var{but-do-not-erase-or-copy-to-a-read-only-buffer})
5530 @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer})))
5535 @section The Definition of @code{insert-buffer}
5536 @findex insert-buffer
5538 @code{insert-buffer} is yet another buffer-related function. This
5539 command copies another buffer @emph{into} the current buffer. It is the
5540 reverse of @code{append-to-buffer} or @code{copy-to-buffer}, since they
5541 copy a region of text @emph{from} the current buffer to another buffer.
5543 Here is a discussion based on the original code. The code was
5544 simplified in 2003 and is harder to understand.
5546 (@xref{New insert-buffer, , New Body for @code{insert-buffer}}, to see
5547 a discussion of the new body.)
5549 In addition, this code illustrates the use of @code{interactive} with a
5550 buffer that might be @dfn{read-only} and the important distinction
5551 between the name of an object and the object actually referred to.
5554 * insert-buffer code::
5555 * insert-buffer interactive:: When you can read, but not write.
5556 * insert-buffer body:: The body has an @code{or} and a @code{let}.
5557 * if & or:: Using an @code{if} instead of an @code{or}.
5558 * Insert or:: How the @code{or} expression works.
5559 * Insert let:: Two @code{save-excursion} expressions.
5560 * New insert-buffer::
5564 @node insert-buffer code
5565 @unnumberedsubsec The Code for @code{insert-buffer}
5569 Here is the earlier code:
5573 (defun insert-buffer (buffer)
5574 "Insert after point the contents of BUFFER.
5575 Puts mark after the inserted text.
5576 BUFFER may be a buffer or a buffer name."
5577 (interactive "*bInsert buffer:@: ")
5580 (or (bufferp buffer)
5581 (setq buffer (get-buffer buffer)))
5582 (let (start end newmark)
5586 (setq start (point-min) end (point-max)))
5589 (insert-buffer-substring buffer start end)
5590 (setq newmark (point)))
5591 (push-mark newmark)))
5596 As with other function definitions, you can use a template to see an
5597 outline of the function:
5601 (defun insert-buffer (buffer)
5602 "@var{documentation}@dots{}"
5603 (interactive "*bInsert buffer:@: ")
5608 @node insert-buffer interactive
5609 @subsection The Interactive Expression in @code{insert-buffer}
5610 @findex interactive, @r{example use of}
5612 In @code{insert-buffer}, the argument to the @code{interactive}
5613 declaration has two parts, an asterisk, @samp{*}, and @samp{bInsert
5617 * Read-only buffer:: When a buffer cannot be modified.
5618 * b for interactive:: An existing buffer or else its name.
5621 @node Read-only buffer
5622 @unnumberedsubsubsec A Read-only Buffer
5623 @cindex Read-only buffer
5624 @cindex Asterisk for read-only buffer
5625 @findex * @r{for read-only buffer}
5627 The asterisk is for the situation when the current buffer is a
5628 read-only buffer---a buffer that cannot be modified. If
5629 @code{insert-buffer} is called when the current buffer is read-only, a
5630 message to this effect is printed in the echo area and the terminal
5631 may beep or blink at you; you will not be permitted to insert anything
5632 into current buffer. The asterisk does not need to be followed by a
5633 newline to separate it from the next argument.
5635 @node b for interactive
5636 @unnumberedsubsubsec @samp{b} in an Interactive Expression
5638 The next argument in the interactive expression starts with a lower
5639 case @samp{b}. (This is different from the code for
5640 @code{append-to-buffer}, which uses an upper-case @samp{B}.
5641 @xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.)
5642 The lower-case @samp{b} tells the Lisp interpreter that the argument
5643 for @code{insert-buffer} should be an existing buffer or else its
5644 name. (The upper-case @samp{B} option provides for the possibility
5645 that the buffer does not exist.) Emacs will prompt you for the name
5646 of the buffer, offering you a default buffer, with name completion
5647 enabled. If the buffer does not exist, you receive a message that
5648 says ``No match''; your terminal may beep at you as well.
5650 The new and simplified code generates a list for @code{interactive}.
5651 It uses the @code{barf-if-buffer-read-only} and @code{read-buffer}
5652 functions with which we are already familiar and the @code{progn}
5653 special form with which we are not. (It will be described later.)
5655 @node insert-buffer body
5656 @subsection The Body of the @code{insert-buffer} Function
5658 The body of the @code{insert-buffer} function has two major parts: an
5659 @code{or} expression and a @code{let} expression. The purpose of the
5660 @code{or} expression is to ensure that the argument @code{buffer} is
5661 bound to a buffer and not just the name of a buffer. The body of the
5662 @code{let} expression contains the code which copies the other buffer
5663 into the current buffer.
5666 In outline, the two expressions fit into the @code{insert-buffer}
5671 (defun insert-buffer (buffer)
5672 "@var{documentation}@dots{}"
5673 (interactive "*bInsert buffer:@: ")
5678 (let (@var{varlist})
5679 @var{body-of-}@code{let}@dots{} )
5683 To understand how the @code{or} expression ensures that the argument
5684 @code{buffer} is bound to a buffer and not to the name of a buffer, it
5685 is first necessary to understand the @code{or} function.
5687 Before doing this, let me rewrite this part of the function using
5688 @code{if} so that you can see what is done in a manner that will be familiar.
5691 @subsection @code{insert-buffer} With an @code{if} Instead of an @code{or}
5693 The job to be done is to make sure the value of @code{buffer} is a
5694 buffer itself and not the name of a buffer. If the value is the name,
5695 then the buffer itself must be got.
5697 You can imagine yourself at a conference where an usher is wandering
5698 around holding a list with your name on it and looking for you: the
5699 usher is ``bound'' to your name, not to you; but when the usher finds
5700 you and takes your arm, the usher becomes ``bound'' to you.
5703 In Lisp, you might describe this situation like this:
5707 (if (not (holding-on-to-guest))
5708 (find-and-take-arm-of-guest))
5712 We want to do the same thing with a buffer---if we do not have the
5713 buffer itself, we want to get it.
5716 Using a predicate called @code{bufferp} that tells us whether we have a
5717 buffer (rather than its name), we can write the code like this:
5721 (if (not (bufferp buffer)) ; @r{if-part}
5722 (setq buffer (get-buffer buffer))) ; @r{then-part}
5727 Here, the true-or-false-test of the @code{if} expression is
5728 @w{@code{(not (bufferp buffer))}}; and the then-part is the expression
5729 @w{@code{(setq buffer (get-buffer buffer))}}.
5731 In the test, the function @code{bufferp} returns true if its argument is
5732 a buffer---but false if its argument is the name of the buffer. (The
5733 last character of the function name @code{bufferp} is the character
5734 @samp{p}; as we saw earlier, such use of @samp{p} is a convention that
5735 indicates that the function is a predicate, which is a term that means
5736 that the function will determine whether some property is true or false.
5737 @xref{Wrong Type of Argument, , Using the Wrong Type Object as an
5741 The function @code{not} precedes the expression @code{(bufferp buffer)},
5742 so the true-or-false-test looks like this:
5745 (not (bufferp buffer))
5749 @code{not} is a function that returns true if its argument is false
5750 and false if its argument is true. So if @code{(bufferp buffer)}
5751 returns true, the @code{not} expression returns false and vice-verse:
5752 what is ``not true'' is false and what is ``not false'' is true.
5754 Using this test, the @code{if} expression works as follows: when the
5755 value of the variable @code{buffer} is actually a buffer rather than
5756 its name, the true-or-false-test returns false and the @code{if}
5757 expression does not evaluate the then-part. This is fine, since we do
5758 not need to do anything to the variable @code{buffer} if it really is
5761 On the other hand, when the value of @code{buffer} is not a buffer
5762 itself, but the name of a buffer, the true-or-false-test returns true
5763 and the then-part of the expression is evaluated. In this case, the
5764 then-part is @code{(setq buffer (get-buffer buffer))}. This
5765 expression uses the @code{get-buffer} function to return an actual
5766 buffer itself, given its name. The @code{setq} then sets the variable
5767 @code{buffer} to the value of the buffer itself, replacing its previous
5768 value (which was the name of the buffer).
5771 @subsection The @code{or} in the Body
5773 The purpose of the @code{or} expression in the @code{insert-buffer}
5774 function is to ensure that the argument @code{buffer} is bound to a
5775 buffer and not just to the name of a buffer. The previous section shows
5776 how the job could have been done using an @code{if} expression.
5777 However, the @code{insert-buffer} function actually uses @code{or}.
5778 To understand this, it is necessary to understand how @code{or} works.
5781 An @code{or} function can have any number of arguments. It evaluates
5782 each argument in turn and returns the value of the first of its
5783 arguments that is not @code{nil}. Also, and this is a crucial feature
5784 of @code{or}, it does not evaluate any subsequent arguments after
5785 returning the first non-@code{nil} value.
5788 The @code{or} expression looks like this:
5792 (or (bufferp buffer)
5793 (setq buffer (get-buffer buffer)))
5798 The first argument to @code{or} is the expression @code{(bufferp buffer)}.
5799 This expression returns true (a non-@code{nil} value) if the buffer is
5800 actually a buffer, and not just the name of a buffer. In the @code{or}
5801 expression, if this is the case, the @code{or} expression returns this
5802 true value and does not evaluate the next expression---and this is fine
5803 with us, since we do not want to do anything to the value of
5804 @code{buffer} if it really is a buffer.
5806 On the other hand, if the value of @code{(bufferp buffer)} is @code{nil},
5807 which it will be if the value of @code{buffer} is the name of a buffer,
5808 the Lisp interpreter evaluates the next element of the @code{or}
5809 expression. This is the expression @code{(setq buffer (get-buffer
5810 buffer))}. This expression returns a non-@code{nil} value, which
5811 is the value to which it sets the variable @code{buffer}---and this
5812 value is a buffer itself, not the name of a buffer.
5814 The result of all this is that the symbol @code{buffer} is always
5815 bound to a buffer itself rather than to the name of a buffer. All
5816 this is necessary because the @code{set-buffer} function in a
5817 following line only works with a buffer itself, not with the name to a
5821 Incidentally, using @code{or}, the situation with the usher would be
5825 (or (holding-on-to-guest) (find-and-take-arm-of-guest))
5829 @subsection The @code{let} Expression in @code{insert-buffer}
5831 After ensuring that the variable @code{buffer} refers to a buffer itself
5832 and not just to the name of a buffer, the @code{insert-buffer function}
5833 continues with a @code{let} expression. This specifies three local
5834 variables, @code{start}, @code{end}, and @code{newmark} and binds them
5835 to the initial value @code{nil}. These variables are used inside the
5836 remainder of the @code{let} and temporarily hide any other occurrence of
5837 variables of the same name in Emacs until the end of the @code{let}.
5840 The body of the @code{let} contains two @code{save-excursion}
5841 expressions. First, we will look at the inner @code{save-excursion}
5842 expression in detail. The expression looks like this:
5848 (setq start (point-min) end (point-max)))
5853 The expression @code{(set-buffer buffer)} changes Emacs's attention
5854 from the current buffer to the one from which the text will copied.
5855 In that buffer, the variables @code{start} and @code{end} are set to
5856 the beginning and end of the buffer, using the commands
5857 @code{point-min} and @code{point-max}. Note that we have here an
5858 illustration of how @code{setq} is able to set two variables in the
5859 same expression. The first argument of @code{setq} is set to the
5860 value of its second, and its third argument is set to the value of its
5863 After the body of the inner @code{save-excursion} is evaluated, the
5864 @code{save-excursion} restores the original buffer, but @code{start} and
5865 @code{end} remain set to the values of the beginning and end of the
5866 buffer from which the text will be copied.
5869 The outer @code{save-excursion} expression looks like this:
5874 (@var{inner-}@code{save-excursion}@var{-expression}
5875 (@var{go-to-new-buffer-and-set-}@code{start}@var{-and-}@code{end})
5876 (insert-buffer-substring buffer start end)
5877 (setq newmark (point)))
5882 The @code{insert-buffer-substring} function copies the text
5883 @emph{into} the current buffer @emph{from} the region indicated by
5884 @code{start} and @code{end} in @code{buffer}. Since the whole of the
5885 second buffer lies between @code{start} and @code{end}, the whole of
5886 the second buffer is copied into the buffer you are editing. Next,
5887 the value of point, which will be at the end of the inserted text, is
5888 recorded in the variable @code{newmark}.
5890 After the body of the outer @code{save-excursion} is evaluated, point
5891 and mark are relocated to their original places.
5893 However, it is convenient to locate a mark at the end of the newly
5894 inserted text and locate point at its beginning. The @code{newmark}
5895 variable records the end of the inserted text. In the last line of
5896 the @code{let} expression, the @code{(push-mark newmark)} expression
5897 function sets a mark to this location. (The previous location of the
5898 mark is still accessible; it is recorded on the mark ring and you can
5899 go back to it with @kbd{C-u C-@key{SPC}}.) Meanwhile, point is
5900 located at the beginning of the inserted text, which is where it was
5901 before you called the insert function, the position of which was saved
5902 by the first @code{save-excursion}.
5905 The whole @code{let} expression looks like this:
5909 (let (start end newmark)
5913 (setq start (point-min) end (point-max)))
5914 (insert-buffer-substring buffer start end)
5915 (setq newmark (point)))
5916 (push-mark newmark))
5920 Like the @code{append-to-buffer} function, the @code{insert-buffer}
5921 function uses @code{let}, @code{save-excursion}, and
5922 @code{set-buffer}. In addition, the function illustrates one way to
5923 use @code{or}. All these functions are building blocks that we will
5924 find and use again and again.
5926 @node New insert-buffer
5927 @subsection New Body for @code{insert-buffer}
5928 @findex insert-buffer, new version body
5929 @findex new version body for insert-buffer
5931 The body in the GNU Emacs 22 version is more confusing than the original.
5934 It consists of two expressions,
5940 (insert-buffer-substring (get-buffer buffer))
5948 except, and this is what confuses novices, very important work is done
5949 inside the @code{push-mark} expression.
5951 The @code{get-buffer} function returns a buffer with the name
5952 provided. You will note that the function is @emph{not} called
5953 @code{get-buffer-create}; it does not create a buffer if one does not
5954 already exist. The buffer returned by @code{get-buffer}, an existing
5955 buffer, is passed to @code{insert-buffer-substring}, which inserts the
5956 whole of the buffer (since you did not specify anything else).
5958 The location into which the buffer is inserted is recorded by
5959 @code{push-mark}. Then the function returns @code{nil}, the value of
5960 its last command. Put another way, the @code{insert-buffer} function
5961 exists only to produce a side effect, inserting another buffer, not to
5964 @node beginning-of-buffer
5965 @section Complete Definition of @code{beginning-of-buffer}
5966 @findex beginning-of-buffer
5968 The basic structure of the @code{beginning-of-buffer} function has
5969 already been discussed. (@xref{simplified-beginning-of-buffer, , A
5970 Simplified @code{beginning-of-buffer} Definition}.)
5971 This section describes the complex part of the definition.
5973 As previously described, when invoked without an argument,
5974 @code{beginning-of-buffer} moves the cursor to the beginning of the
5975 buffer (in truth, the beginning of the accessible portion of the
5976 buffer), leaving the mark at the previous position. However, when the
5977 command is invoked with a number between one and ten, the function
5978 considers that number to be a fraction of the length of the buffer,
5979 measured in tenths, and Emacs moves the cursor that fraction of the
5980 way from the beginning of the buffer. Thus, you can either call this
5981 function with the key command @kbd{M-<}, which will move the cursor to
5982 the beginning of the buffer, or with a key command such as @kbd{C-u 7
5983 M-<} which will move the cursor to a point 70% of the way through the
5984 buffer. If a number bigger than ten is used for the argument, it
5985 moves to the end of the buffer.
5987 The @code{beginning-of-buffer} function can be called with or without an
5988 argument. The use of the argument is optional.
5991 * Optional Arguments::
5992 * beginning-of-buffer opt arg:: Example with optional argument.
5993 * beginning-of-buffer complete::
5996 @node Optional Arguments
5997 @subsection Optional Arguments
5999 Unless told otherwise, Lisp expects that a function with an argument in
6000 its function definition will be called with a value for that argument.
6001 If that does not happen, you get an error and a message that says
6002 @samp{Wrong number of arguments}.
6004 @cindex Optional arguments
6007 However, optional arguments are a feature of Lisp: a particular
6008 @dfn{keyword} is used to tell the Lisp interpreter that an argument is
6009 optional. The keyword is @code{&optional}. (The @samp{&} in front of
6010 @samp{optional} is part of the keyword.) In a function definition, if
6011 an argument follows the keyword @code{&optional}, no value need be
6012 passed to that argument when the function is called.
6015 The first line of the function definition of @code{beginning-of-buffer}
6016 therefore looks like this:
6019 (defun beginning-of-buffer (&optional arg)
6023 In outline, the whole function looks like this:
6027 (defun beginning-of-buffer (&optional arg)
6028 "@var{documentation}@dots{}"
6030 (or (@var{is-the-argument-a-cons-cell} arg)
6031 (and @var{are-both-transient-mark-mode-and-mark-active-true})
6033 (let (@var{determine-size-and-set-it})
6035 (@var{if-there-is-an-argument}
6036 @var{figure-out-where-to-go}
6043 The function is similar to the @code{simplified-beginning-of-buffer}
6044 function except that the @code{interactive} expression has @code{"P"}
6045 as an argument and the @code{goto-char} function is followed by an
6046 if-then-else expression that figures out where to put the cursor if
6047 there is an argument that is not a cons cell.
6049 (Since I do not explain a cons cell for many more chapters, please
6050 consider ignoring the function @code{consp}. @xref{List
6051 Implementation, , How Lists are Implemented}, and @ref{Cons Cell Type,
6052 , Cons Cell and List Types, elisp, The GNU Emacs Lisp Reference
6055 The @code{"P"} in the @code{interactive} expression tells Emacs to
6056 pass a prefix argument, if there is one, to the function in raw form.
6057 A prefix argument is made by typing the @key{META} key followed by a
6058 number, or by typing @kbd{C-u} and then a number. (If you don't type
6059 a number, @kbd{C-u} defaults to a cons cell with a 4. A lowercase
6060 @code{"p"} in the @code{interactive} expression causes the function to
6061 convert a prefix arg to a number.)
6063 The true-or-false-test of the @code{if} expression looks complex, but
6064 it is not: it checks whether @code{arg} has a value that is not
6065 @code{nil} and whether it is a cons cell. (That is what @code{consp}
6066 does; it checks whether its argument is a cons cell.) If @code{arg}
6067 has a value that is not @code{nil} (and is not a cons cell), which
6068 will be the case if @code{beginning-of-buffer} is called with a
6069 numeric argument, then this true-or-false-test will return true and
6070 the then-part of the @code{if} expression will be evaluated. On the
6071 other hand, if @code{beginning-of-buffer} is not called with an
6072 argument, the value of @code{arg} will be @code{nil} and the else-part
6073 of the @code{if} expression will be evaluated. The else-part is
6074 simply @code{point-min}, and when this is the outcome, the whole
6075 @code{goto-char} expression is @code{(goto-char (point-min))}, which
6076 is how we saw the @code{beginning-of-buffer} function in its
6079 @node beginning-of-buffer opt arg
6080 @subsection @code{beginning-of-buffer} with an Argument
6082 When @code{beginning-of-buffer} is called with an argument, an
6083 expression is evaluated which calculates what value to pass to
6084 @code{goto-char}. This expression is rather complicated at first sight.
6085 It includes an inner @code{if} expression and much arithmetic. It looks
6090 (if (> (buffer-size) 10000)
6091 ;; @r{Avoid overflow for large buffer sizes!}
6092 (* (prefix-numeric-value arg)
6097 size (prefix-numeric-value arg))) 10)))
6102 * Disentangle beginning-of-buffer::
6103 * Large buffer case::
6104 * Small buffer case::
6108 @node Disentangle beginning-of-buffer
6109 @unnumberedsubsubsec Disentangle @code{beginning-of-buffer}
6112 Like other complex-looking expressions, the conditional expression
6113 within @code{beginning-of-buffer} can be disentangled by looking at it
6114 as parts of a template, in this case, the template for an if-then-else
6115 expression. In skeletal form, the expression looks like this:
6119 (if (@var{buffer-is-large}
6120 @var{divide-buffer-size-by-10-and-multiply-by-arg}
6121 @var{else-use-alternate-calculation}
6125 The true-or-false-test of this inner @code{if} expression checks the
6126 size of the buffer. The reason for this is that the old version 18
6127 Emacs used numbers that are no bigger than eight million or so and in
6128 the computation that followed, the programmer feared that Emacs might
6129 try to use over-large numbers if the buffer were large. The term
6130 `overflow', mentioned in the comment, means numbers that are over
6131 large. More recent versions of Emacs use larger numbers, but this
6132 code has not been touched, if only because people now look at buffers
6133 that are far, far larger than ever before.
6135 There are two cases: if the buffer is large and if it is not.
6137 @node Large buffer case
6138 @unnumberedsubsubsec What happens in a large buffer
6140 In @code{beginning-of-buffer}, the inner @code{if} expression tests
6141 whether the size of the buffer is greater than 10,000 characters. To do
6142 this, it uses the @code{>} function and the computation of @code{size}
6143 that comes from the let expression.
6145 In the old days, the function @code{buffer-size} was used. Not only
6146 was that function called several times, it gave the size of the whole
6147 buffer, not the accessible part. The computation makes much more
6148 sense when it handles just the accessible part. (@xref{Narrowing &
6149 Widening, , Narrowing and Widening}, for more information on focusing
6150 attention to an `accessible' part.)
6153 The line looks like this:
6161 When the buffer is large, the then-part of the @code{if} expression is
6162 evaluated. It reads like this (after formatting for easy reading):
6167 (prefix-numeric-value arg)
6173 This expression is a multiplication, with two arguments to the function
6176 The first argument is @code{(prefix-numeric-value arg)}. When
6177 @code{"P"} is used as the argument for @code{interactive}, the value
6178 passed to the function as its argument is passed a ``raw prefix
6179 argument'', and not a number. (It is a number in a list.) To perform
6180 the arithmetic, a conversion is necessary, and
6181 @code{prefix-numeric-value} does the job.
6183 @findex / @r{(division)}
6185 The second argument is @code{(/ size 10)}. This expression divides
6186 the numeric value by ten---the numeric value of the size of the
6187 accessible portion of the buffer. This produces a number that tells
6188 how many characters make up one tenth of the buffer size. (In Lisp,
6189 @code{/} is used for division, just as @code{*} is used for
6193 In the multiplication expression as a whole, this amount is multiplied
6194 by the value of the prefix argument---the multiplication looks like this:
6198 (* @var{numeric-value-of-prefix-arg}
6199 @var{number-of-characters-in-one-tenth-of-the-accessible-buffer})
6204 If, for example, the prefix argument is @samp{7}, the one-tenth value
6205 will be multiplied by 7 to give a position 70% of the way through.
6208 The result of all this is that if the accessible portion of the buffer
6209 is large, the @code{goto-char} expression reads like this:
6213 (goto-char (* (prefix-numeric-value arg)
6218 This puts the cursor where we want it.
6220 @node Small buffer case
6221 @unnumberedsubsubsec What happens in a small buffer
6223 If the buffer contains fewer than 10,000 characters, a slightly
6224 different computation is performed. You might think this is not
6225 necessary, since the first computation could do the job. However, in
6226 a small buffer, the first method may not put the cursor on exactly the
6227 desired line; the second method does a better job.
6230 The code looks like this:
6232 @c Keep this on one line.
6234 (/ (+ 10 (* size (prefix-numeric-value arg))) 10))
6239 This is code in which you figure out what happens by discovering how the
6240 functions are embedded in parentheses. It is easier to read if you
6241 reformat it with each expression indented more deeply than its
6242 enclosing expression:
6250 (prefix-numeric-value arg)))
6257 Looking at parentheses, we see that the innermost operation is
6258 @code{(prefix-numeric-value arg)}, which converts the raw argument to
6259 a number. In the following expression, this number is multiplied by
6260 the size of the accessible portion of the buffer:
6263 (* size (prefix-numeric-value arg))
6267 This multiplication creates a number that may be larger than the size of
6268 the buffer---seven times larger if the argument is 7, for example. Ten
6269 is then added to this number and finally the large number is divided by
6270 ten to provide a value that is one character larger than the percentage
6271 position in the buffer.
6273 The number that results from all this is passed to @code{goto-char} and
6274 the cursor is moved to that point.
6277 @node beginning-of-buffer complete
6278 @subsection The Complete @code{beginning-of-buffer}
6281 Here is the complete text of the @code{beginning-of-buffer} function:
6287 (defun beginning-of-buffer (&optional arg)
6288 "Move point to the beginning of the buffer;
6289 leave mark at previous position.
6290 With \\[universal-argument] prefix,
6291 do not set mark at previous position.
6293 put point N/10 of the way from the beginning.
6295 If the buffer is narrowed,
6296 this command uses the beginning and size
6297 of the accessible part of the buffer.
6301 Don't use this command in Lisp programs!
6302 \(goto-char (point-min)) is faster
6303 and avoids clobbering the mark."
6306 (and transient-mark-mode mark-active)
6310 (let ((size (- (point-max) (point-min))))
6311 (goto-char (if (and arg (not (consp arg)))
6314 ;; Avoid overflow for large buffer sizes!
6315 (* (prefix-numeric-value arg)
6317 (/ (+ 10 (* size (prefix-numeric-value arg)))
6320 (if arg (forward-line 1)))
6325 From before GNU Emacs 22
6328 (defun beginning-of-buffer (&optional arg)
6329 "Move point to the beginning of the buffer;
6330 leave mark at previous position.
6331 With arg N, put point N/10 of the way
6332 from the true beginning.
6335 Don't use this in Lisp programs!
6336 \(goto-char (point-min)) is faster
6337 and does not set the mark."
6344 (if (> (buffer-size) 10000)
6345 ;; @r{Avoid overflow for large buffer sizes!}
6346 (* (prefix-numeric-value arg)
6347 (/ (buffer-size) 10))
6350 (/ (+ 10 (* (buffer-size)
6351 (prefix-numeric-value arg)))
6354 (if arg (forward-line 1)))
6360 Except for two small points, the previous discussion shows how this
6361 function works. The first point deals with a detail in the
6362 documentation string, and the second point concerns the last line of
6366 In the documentation string, there is reference to an expression:
6369 \\[universal-argument]
6373 A @samp{\\} is used before the first square bracket of this
6374 expression. This @samp{\\} tells the Lisp interpreter to substitute
6375 whatever key is currently bound to the @samp{[@dots{}]}. In the case
6376 of @code{universal-argument}, that is usually @kbd{C-u}, but it might
6377 be different. (@xref{Documentation Tips, , Tips for Documentation
6378 Strings, elisp, The GNU Emacs Lisp Reference Manual}, for more
6382 Finally, the last line of the @code{beginning-of-buffer} command says
6383 to move point to the beginning of the next line if the command is
6384 invoked with an argument:
6387 (if arg (forward-line 1)))
6391 This puts the cursor at the beginning of the first line after the
6392 appropriate tenths position in the buffer. This is a flourish that
6393 means that the cursor is always located @emph{at least} the requested
6394 tenths of the way through the buffer, which is a nicety that is,
6395 perhaps, not necessary, but which, if it did not occur, would be sure
6398 On the other hand, it also means that if you specify the command with
6399 a @kbd{C-u}, but without a number, that is to say, if the `raw prefix
6400 argument' is simply a cons cell, then the command puts you at the
6401 beginning of the second line @dots{} I don't know whether this is
6402 intended or whether no one has dealt with the code to avoid this
6405 @node Second Buffer Related Review
6408 Here is a brief summary of some of the topics covered in this chapter.
6412 Evaluate each argument in sequence, and return the value of the first
6413 argument that is not @code{nil}; if none return a value that is not
6414 @code{nil}, return @code{nil}. In brief, return the first true value
6415 of the arguments; return a true value if one @emph{or} any of the
6419 Evaluate each argument in sequence, and if any are @code{nil}, return
6420 @code{nil}; if none are @code{nil}, return the value of the last
6421 argument. In brief, return a true value only if all the arguments are
6422 true; return a true value if one @emph{and} each of the others is
6426 A keyword used to indicate that an argument to a function definition
6427 is optional; this means that the function can be evaluated without the
6428 argument, if desired.
6430 @item prefix-numeric-value
6431 Convert the `raw prefix argument' produced by @code{(interactive
6432 "P")} to a numeric value.
6435 Move point forward to the beginning of the next line, or if the argument
6436 is greater than one, forward that many lines. If it can't move as far
6437 forward as it is supposed to, @code{forward-line} goes forward as far as
6438 it can and then returns a count of the number of additional lines it was
6439 supposed to move but couldn't.
6442 Delete the entire contents of the current buffer.
6445 Return @code{t} if its argument is a buffer; otherwise return @code{nil}.
6448 @node optional Exercise
6449 @section @code{optional} Argument Exercise
6451 Write an interactive function with an optional argument that tests
6452 whether its argument, a number, is greater than or equal to, or else,
6453 less than the value of @code{fill-column}, and tells you which, in a
6454 message. However, if you do not pass an argument to the function, use
6455 56 as a default value.
6457 @node Narrowing & Widening
6458 @chapter Narrowing and Widening
6459 @cindex Focusing attention (narrowing)
6463 Narrowing is a feature of Emacs that makes it possible for you to focus
6464 on a specific part of a buffer, and work without accidentally changing
6465 other parts. Narrowing is normally disabled since it can confuse
6469 * Narrowing advantages:: The advantages of narrowing
6470 * save-restriction:: The @code{save-restriction} special form.
6471 * what-line:: The number of the line that point is on.
6476 @node Narrowing advantages
6477 @unnumberedsec The Advantages of Narrowing
6480 With narrowing, the rest of a buffer is made invisible, as if it weren't
6481 there. This is an advantage if, for example, you want to replace a word
6482 in one part of a buffer but not in another: you narrow to the part you want
6483 and the replacement is carried out only in that section, not in the rest
6484 of the buffer. Searches will only work within a narrowed region, not
6485 outside of one, so if you are fixing a part of a document, you can keep
6486 yourself from accidentally finding parts you do not need to fix by
6487 narrowing just to the region you want.
6488 (The key binding for @code{narrow-to-region} is @kbd{C-x n n}.)
6490 However, narrowing does make the rest of the buffer invisible, which
6491 can scare people who inadvertently invoke narrowing and think they
6492 have deleted a part of their file. Moreover, the @code{undo} command
6493 (which is usually bound to @kbd{C-x u}) does not turn off narrowing
6494 (nor should it), so people can become quite desperate if they do not
6495 know that they can return the rest of a buffer to visibility with the
6496 @code{widen} command.
6497 (The key binding for @code{widen} is @kbd{C-x n w}.)
6499 Narrowing is just as useful to the Lisp interpreter as to a human.
6500 Often, an Emacs Lisp function is designed to work on just part of a
6501 buffer; or conversely, an Emacs Lisp function needs to work on all of a
6502 buffer that has been narrowed. The @code{what-line} function, for
6503 example, removes the narrowing from a buffer, if it has any narrowing
6504 and when it has finished its job, restores the narrowing to what it was.
6505 On the other hand, the @code{count-lines} function
6506 uses narrowing to restrict itself to just that portion
6507 of the buffer in which it is interested and then restores the previous
6510 @node save-restriction
6511 @section The @code{save-restriction} Special Form
6512 @findex save-restriction
6514 In Emacs Lisp, you can use the @code{save-restriction} special form to
6515 keep track of whatever narrowing is in effect, if any. When the Lisp
6516 interpreter meets with @code{save-restriction}, it executes the code
6517 in the body of the @code{save-restriction} expression, and then undoes
6518 any changes to narrowing that the code caused. If, for example, the
6519 buffer is narrowed and the code that follows @code{save-restriction}
6520 gets rid of the narrowing, @code{save-restriction} returns the buffer
6521 to its narrowed region afterwards. In the @code{what-line} command,
6522 any narrowing the buffer may have is undone by the @code{widen}
6523 command that immediately follows the @code{save-restriction} command.
6524 Any original narrowing is restored just before the completion of the
6528 The template for a @code{save-restriction} expression is simple:
6538 The body of the @code{save-restriction} is one or more expressions that
6539 will be evaluated in sequence by the Lisp interpreter.
6541 Finally, a point to note: when you use both @code{save-excursion} and
6542 @code{save-restriction}, one right after the other, you should use
6543 @code{save-excursion} outermost. If you write them in reverse order,
6544 you may fail to record narrowing in the buffer to which Emacs switches
6545 after calling @code{save-excursion}. Thus, when written together,
6546 @code{save-excursion} and @code{save-restriction} should be written
6557 In other circumstances, when not written together, the
6558 @code{save-excursion} and @code{save-restriction} special forms must
6559 be written in the order appropriate to the function.
6575 /usr/local/src/emacs/lisp/simple.el
6578 "Print the current buffer line number and narrowed line number of point."
6580 (let ((start (point-min))
6581 (n (line-number-at-pos)))
6583 (message "Line %d" n)
6587 (message "line %d (narrowed line %d)"
6588 (+ n (line-number-at-pos start) -1) n))))))
6590 (defun line-number-at-pos (&optional pos)
6591 "Return (narrowed) buffer line number at position POS.
6592 If POS is nil, use current buffer location.
6593 Counting starts at (point-min), so the value refers
6594 to the contents of the accessible portion of the buffer."
6595 (let ((opoint (or pos (point))) start)
6597 (goto-char (point-min))
6598 (setq start (point))
6601 (1+ (count-lines start (point))))))
6603 (defun count-lines (start end)
6604 "Return number of lines between START and END.
6605 This is usually the number of newlines between them,
6606 but can be one more if START is not equal to END
6607 and the greater of them is not at the start of a line."
6610 (narrow-to-region start end)
6611 (goto-char (point-min))
6612 (if (eq selective-display t)
6615 (while (re-search-forward "[\n\C-m]" nil t 40)
6616 (setq done (+ 40 done)))
6617 (while (re-search-forward "[\n\C-m]" nil t 1)
6618 (setq done (+ 1 done)))
6619 (goto-char (point-max))
6620 (if (and (/= start end)
6624 (- (buffer-size) (forward-line (buffer-size)))))))
6628 @section @code{what-line}
6630 @cindex Widening, example of
6632 The @code{what-line} command tells you the number of the line in which
6633 the cursor is located. The function illustrates the use of the
6634 @code{save-restriction} and @code{save-excursion} commands. Here is the
6635 original text of the function:
6640 "Print the current line number (in the buffer) of point."
6647 (1+ (count-lines 1 (point)))))))
6651 (In recent versions of GNU Emacs, the @code{what-line} function has
6652 been expanded to tell you your line number in a narrowed buffer as
6653 well as your line number in a widened buffer. The recent version is
6654 more complex than the version shown here. If you feel adventurous,
6655 you might want to look at it after figuring out how this version
6656 works. You will probably need to use @kbd{C-h f}
6657 (@code{describe-function}). The newer version uses a conditional to
6658 determine whether the buffer has been narrowed.
6660 (Also, it uses @code{line-number-at-pos}, which among other simple
6661 expressions, such as @code{(goto-char (point-min))}, moves point to
6662 the beginning of the current line with @code{(forward-line 0)} rather
6663 than @code{beginning-of-line}.)
6665 The @code{what-line} function as shown here has a documentation line
6666 and is interactive, as you would expect. The next two lines use the
6667 functions @code{save-restriction} and @code{widen}.
6669 The @code{save-restriction} special form notes whatever narrowing is in
6670 effect, if any, in the current buffer and restores that narrowing after
6671 the code in the body of the @code{save-restriction} has been evaluated.
6673 The @code{save-restriction} special form is followed by @code{widen}.
6674 This function undoes any narrowing the current buffer may have had
6675 when @code{what-line} was called. (The narrowing that was there is
6676 the narrowing that @code{save-restriction} remembers.) This widening
6677 makes it possible for the line counting commands to count from the
6678 beginning of the buffer. Otherwise, they would have been limited to
6679 counting within the accessible region. Any original narrowing is
6680 restored just before the completion of the function by the
6681 @code{save-restriction} special form.
6683 The call to @code{widen} is followed by @code{save-excursion}, which
6684 saves the location of the cursor (i.e., of point) and of the mark, and
6685 restores them after the code in the body of the @code{save-excursion}
6686 uses the @code{beginning-of-line} function to move point.
6688 (Note that the @code{(widen)} expression comes between the
6689 @code{save-restriction} and @code{save-excursion} special forms. When
6690 you write the two @code{save- @dots{}} expressions in sequence, write
6691 @code{save-excursion} outermost.)
6694 The last two lines of the @code{what-line} function are functions to
6695 count the number of lines in the buffer and then print the number in the
6701 (1+ (count-lines 1 (point)))))))
6705 The @code{message} function prints a one-line message at the bottom of
6706 the Emacs screen. The first argument is inside of quotation marks and
6707 is printed as a string of characters. However, it may contain a
6708 @samp{%d} expression to print a following argument. @samp{%d} prints
6709 the argument as a decimal, so the message will say something such as
6713 The number that is printed in place of the @samp{%d} is computed by the
6714 last line of the function:
6717 (1+ (count-lines 1 (point)))
6723 (defun count-lines (start end)
6724 "Return number of lines between START and END.
6725 This is usually the number of newlines between them,
6726 but can be one more if START is not equal to END
6727 and the greater of them is not at the start of a line."
6730 (narrow-to-region start end)
6731 (goto-char (point-min))
6732 (if (eq selective-display t)
6735 (while (re-search-forward "[\n\C-m]" nil t 40)
6736 (setq done (+ 40 done)))
6737 (while (re-search-forward "[\n\C-m]" nil t 1)
6738 (setq done (+ 1 done)))
6739 (goto-char (point-max))
6740 (if (and (/= start end)
6744 (- (buffer-size) (forward-line (buffer-size)))))))
6748 What this does is count the lines from the first position of the
6749 buffer, indicated by the @code{1}, up to @code{(point)}, and then add
6750 one to that number. (The @code{1+} function adds one to its
6751 argument.) We add one to it because line 2 has only one line before
6752 it, and @code{count-lines} counts only the lines @emph{before} the
6755 After @code{count-lines} has done its job, and the message has been
6756 printed in the echo area, the @code{save-excursion} restores point and
6757 mark to their original positions; and @code{save-restriction} restores
6758 the original narrowing, if any.
6760 @node narrow Exercise
6761 @section Exercise with Narrowing
6763 Write a function that will display the first 60 characters of the
6764 current buffer, even if you have narrowed the buffer to its latter
6765 half so that the first line is inaccessible. Restore point, mark, and
6766 narrowing. For this exercise, you need to use a whole potpourri of
6767 functions, including @code{save-restriction}, @code{widen},
6768 @code{goto-char}, @code{point-min}, @code{message}, and
6769 @code{buffer-substring}.
6771 @cindex Properties, mention of @code{buffer-substring-no-properties}
6772 (@code{buffer-substring} is a previously unmentioned function you will
6773 have to investigate yourself; or perhaps you will have to use
6774 @code{buffer-substring-no-properties} or
6775 @code{filter-buffer-substring} @dots{}, yet other functions. Text
6776 properties are a feature otherwise not discussed here. @xref{Text
6777 Properties, , Text Properties, elisp, The GNU Emacs Lisp Reference
6780 Additionally, do you really need @code{goto-char} or @code{point-min}?
6781 Or can you write the function without them?
6783 @node car cdr & cons
6784 @chapter @code{car}, @code{cdr}, @code{cons}: Fundamental Functions
6785 @findex car, @r{introduced}
6786 @findex cdr, @r{introduced}
6788 In Lisp, @code{car}, @code{cdr}, and @code{cons} are fundamental
6789 functions. The @code{cons} function is used to construct lists, and
6790 the @code{car} and @code{cdr} functions are used to take them apart.
6792 In the walk through of the @code{copy-region-as-kill} function, we
6793 will see @code{cons} as well as two variants on @code{cdr},
6794 namely, @code{setcdr} and @code{nthcdr}. (@xref{copy-region-as-kill}.)
6797 * Strange Names:: An historical aside: why the strange names?
6798 * car & cdr:: Functions for extracting part of a list.
6799 * cons:: Constructing a list.
6800 * nthcdr:: Calling @code{cdr} repeatedly.
6802 * setcar:: Changing the first element of a list.
6803 * setcdr:: Changing the rest of a list.
6809 @unnumberedsec Strange Names
6812 The name of the @code{cons} function is not unreasonable: it is an
6813 abbreviation of the word `construct'. The origins of the names for
6814 @code{car} and @code{cdr}, on the other hand, are esoteric: @code{car}
6815 is an acronym from the phrase `Contents of the Address part of the
6816 Register'; and @code{cdr} (pronounced `could-er') is an acronym from
6817 the phrase `Contents of the Decrement part of the Register'. These
6818 phrases refer to specific pieces of hardware on the very early
6819 computer on which the original Lisp was developed. Besides being
6820 obsolete, the phrases have been completely irrelevant for more than 25
6821 years to anyone thinking about Lisp. Nonetheless, although a few
6822 brave scholars have begun to use more reasonable names for these
6823 functions, the old terms are still in use. In particular, since the
6824 terms are used in the Emacs Lisp source code, we will use them in this
6828 @section @code{car} and @code{cdr}
6830 The @sc{car} of a list is, quite simply, the first item in the list.
6831 Thus the @sc{car} of the list @code{(rose violet daisy buttercup)} is
6835 If you are reading this in Info in GNU Emacs, you can see this by
6836 evaluating the following:
6839 (car '(rose violet daisy buttercup))
6843 After evaluating the expression, @code{rose} will appear in the echo
6846 Clearly, a more reasonable name for the @code{car} function would be
6847 @code{first} and this is often suggested.
6849 @code{car} does not remove the first item from the list; it only reports
6850 what it is. After @code{car} has been applied to a list, the list is
6851 still the same as it was. In the jargon, @code{car} is
6852 `non-destructive'. This feature turns out to be important.
6854 The @sc{cdr} of a list is the rest of the list, that is, the
6855 @code{cdr} function returns the part of the list that follows the
6856 first item. Thus, while the @sc{car} of the list @code{'(rose violet
6857 daisy buttercup)} is @code{rose}, the rest of the list, the value
6858 returned by the @code{cdr} function, is @code{(violet daisy
6862 You can see this by evaluating the following in the usual way:
6865 (cdr '(rose violet daisy buttercup))
6869 When you evaluate this, @code{(violet daisy buttercup)} will appear in
6872 Like @code{car}, @code{cdr} does not remove any elements from the
6873 list---it just returns a report of what the second and subsequent
6876 Incidentally, in the example, the list of flowers is quoted. If it were
6877 not, the Lisp interpreter would try to evaluate the list by calling
6878 @code{rose} as a function. In this example, we do not want to do that.
6880 Clearly, a more reasonable name for @code{cdr} would be @code{rest}.
6882 (There is a lesson here: when you name new functions, consider very
6883 carefully what you are doing, since you may be stuck with the names
6884 for far longer than you expect. The reason this document perpetuates
6885 these names is that the Emacs Lisp source code uses them, and if I did
6886 not use them, you would have a hard time reading the code; but do,
6887 please, try to avoid using these terms yourself. The people who come
6888 after you will be grateful to you.)
6890 When @code{car} and @code{cdr} are applied to a list made up of symbols,
6891 such as the list @code{(pine fir oak maple)}, the element of the list
6892 returned by the function @code{car} is the symbol @code{pine} without
6893 any parentheses around it. @code{pine} is the first element in the
6894 list. However, the @sc{cdr} of the list is a list itself, @code{(fir
6895 oak maple)}, as you can see by evaluating the following expressions in
6900 (car '(pine fir oak maple))
6902 (cdr '(pine fir oak maple))
6906 On the other hand, in a list of lists, the first element is itself a
6907 list. @code{car} returns this first element as a list. For example,
6908 the following list contains three sub-lists, a list of carnivores, a
6909 list of herbivores and a list of sea mammals:
6913 (car '((lion tiger cheetah)
6914 (gazelle antelope zebra)
6915 (whale dolphin seal)))
6920 In this example, the first element or @sc{car} of the list is the list of
6921 carnivores, @code{(lion tiger cheetah)}, and the rest of the list is
6922 @code{((gazelle antelope zebra) (whale dolphin seal))}.
6926 (cdr '((lion tiger cheetah)
6927 (gazelle antelope zebra)
6928 (whale dolphin seal)))
6932 It is worth saying again that @code{car} and @code{cdr} are
6933 non-destructive---that is, they do not modify or change lists to which
6934 they are applied. This is very important for how they are used.
6936 Also, in the first chapter, in the discussion about atoms, I said that
6937 in Lisp, ``certain kinds of atom, such as an array, can be separated
6938 into parts; but the mechanism for doing this is different from the
6939 mechanism for splitting a list. As far as Lisp is concerned, the
6940 atoms of a list are unsplittable.'' (@xref{Lisp Atoms}.) The
6941 @code{car} and @code{cdr} functions are used for splitting lists and
6942 are considered fundamental to Lisp. Since they cannot split or gain
6943 access to the parts of an array, an array is considered an atom.
6944 Conversely, the other fundamental function, @code{cons}, can put
6945 together or construct a list, but not an array. (Arrays are handled
6946 by array-specific functions. @xref{Arrays, , Arrays, elisp, The GNU
6947 Emacs Lisp Reference Manual}.)
6950 @section @code{cons}
6951 @findex cons, @r{introduced}
6953 The @code{cons} function constructs lists; it is the inverse of
6954 @code{car} and @code{cdr}. For example, @code{cons} can be used to make
6955 a four element list from the three element list, @code{(fir oak maple)}:
6958 (cons 'pine '(fir oak maple))
6963 After evaluating this list, you will see
6966 (pine fir oak maple)
6970 appear in the echo area. @code{cons} causes the creation of a new
6971 list in which the element is followed by the elements of the original
6974 We often say that `@code{cons} puts a new element at the beginning of
6975 a list; it attaches or pushes elements onto the list', but this
6976 phrasing can be misleading, since @code{cons} does not change an
6977 existing list, but creates a new one.
6979 Like @code{car} and @code{cdr}, @code{cons} is non-destructive.
6983 * length:: How to find the length of a list.
6988 @unnumberedsubsec Build a list
6991 @code{cons} must have a list to attach to.@footnote{Actually, you can
6992 @code{cons} an element to an atom to produce a dotted pair. Dotted
6993 pairs are not discussed here; see @ref{Dotted Pair Notation, , Dotted
6994 Pair Notation, elisp, The GNU Emacs Lisp Reference Manual}.} You
6995 cannot start from absolutely nothing. If you are building a list, you
6996 need to provide at least an empty list at the beginning. Here is a
6997 series of @code{cons} expressions that build up a list of flowers. If
6998 you are reading this in Info in GNU Emacs, you can evaluate each of
6999 the expressions in the usual way; the value is printed in this text
7000 after @samp{@result{}}, which you may read as `evaluates to'.
7004 (cons 'buttercup ())
7005 @result{} (buttercup)
7009 (cons 'daisy '(buttercup))
7010 @result{} (daisy buttercup)
7014 (cons 'violet '(daisy buttercup))
7015 @result{} (violet daisy buttercup)
7019 (cons 'rose '(violet daisy buttercup))
7020 @result{} (rose violet daisy buttercup)
7025 In the first example, the empty list is shown as @code{()} and a list
7026 made up of @code{buttercup} followed by the empty list is constructed.
7027 As you can see, the empty list is not shown in the list that was
7028 constructed. All that you see is @code{(buttercup)}. The empty list is
7029 not counted as an element of a list because there is nothing in an empty
7030 list. Generally speaking, an empty list is invisible.
7032 The second example, @code{(cons 'daisy '(buttercup))} constructs a new,
7033 two element list by putting @code{daisy} in front of @code{buttercup};
7034 and the third example constructs a three element list by putting
7035 @code{violet} in front of @code{daisy} and @code{buttercup}.
7038 @subsection Find the Length of a List: @code{length}
7041 You can find out how many elements there are in a list by using the Lisp
7042 function @code{length}, as in the following examples:
7046 (length '(buttercup))
7051 (length '(daisy buttercup))
7056 (length (cons 'violet '(daisy buttercup)))
7062 In the third example, the @code{cons} function is used to construct a
7063 three element list which is then passed to the @code{length} function as
7067 We can also use @code{length} to count the number of elements in an
7078 As you would expect, the number of elements in an empty list is zero.
7080 An interesting experiment is to find out what happens if you try to find
7081 the length of no list at all; that is, if you try to call @code{length}
7082 without giving it an argument, not even an empty list:
7090 What you see, if you evaluate this, is the error message
7093 Lisp error: (wrong-number-of-arguments length 0)
7097 This means that the function receives the wrong number of
7098 arguments, zero, when it expects some other number of arguments. In
7099 this case, one argument is expected, the argument being a list whose
7100 length the function is measuring. (Note that @emph{one} list is
7101 @emph{one} argument, even if the list has many elements inside it.)
7103 The part of the error message that says @samp{length} is the name of
7107 @code{length} is still a subroutine, but you need C-h f to discover that.
7109 In an earlier version:
7110 This is written with a special notation, @samp{#<subr},
7111 that indicates that the function @code{length} is one of the primitive
7112 functions written in C rather than in Emacs Lisp. (@samp{subr} is an
7113 abbreviation for `subroutine'.) @xref{What Is a Function, , What Is a
7114 Function?, elisp , The GNU Emacs Lisp Reference Manual}, for more
7119 @section @code{nthcdr}
7122 The @code{nthcdr} function is associated with the @code{cdr} function.
7123 What it does is take the @sc{cdr} of a list repeatedly.
7125 If you take the @sc{cdr} of the list @code{(pine fir
7126 oak maple)}, you will be returned the list @code{(fir oak maple)}. If you
7127 repeat this on what was returned, you will be returned the list
7128 @code{(oak maple)}. (Of course, repeated @sc{cdr}ing on the original
7129 list will just give you the original @sc{cdr} since the function does
7130 not change the list. You need to evaluate the @sc{cdr} of the
7131 @sc{cdr} and so on.) If you continue this, eventually you will be
7132 returned an empty list, which in this case, instead of being shown as
7133 @code{()} is shown as @code{nil}.
7136 For review, here is a series of repeated @sc{cdr}s, the text following
7137 the @samp{@result{}} shows what is returned.
7141 (cdr '(pine fir oak maple))
7142 @result{}(fir oak maple)
7146 (cdr '(fir oak maple))
7147 @result{} (oak maple)
7172 You can also do several @sc{cdr}s without printing the values in
7177 (cdr (cdr '(pine fir oak maple)))
7178 @result{} (oak maple)
7183 In this example, the Lisp interpreter evaluates the innermost list first.
7184 The innermost list is quoted, so it just passes the list as it is to the
7185 innermost @code{cdr}. This @code{cdr} passes a list made up of the
7186 second and subsequent elements of the list to the outermost @code{cdr},
7187 which produces a list composed of the third and subsequent elements of
7188 the original list. In this example, the @code{cdr} function is repeated
7189 and returns a list that consists of the original list without its
7192 The @code{nthcdr} function does the same as repeating the call to
7193 @code{cdr}. In the following example, the argument 2 is passed to the
7194 function @code{nthcdr}, along with the list, and the value returned is
7195 the list without its first two items, which is exactly the same
7196 as repeating @code{cdr} twice on the list:
7200 (nthcdr 2 '(pine fir oak maple))
7201 @result{} (oak maple)
7206 Using the original four element list, we can see what happens when
7207 various numeric arguments are passed to @code{nthcdr}, including 0, 1,
7212 ;; @r{Leave the list as it was.}
7213 (nthcdr 0 '(pine fir oak maple))
7214 @result{} (pine fir oak maple)
7218 ;; @r{Return a copy without the first element.}
7219 (nthcdr 1 '(pine fir oak maple))
7220 @result{} (fir oak maple)
7224 ;; @r{Return a copy of the list without three elements.}
7225 (nthcdr 3 '(pine fir oak maple))
7230 ;; @r{Return a copy lacking all four elements.}
7231 (nthcdr 4 '(pine fir oak maple))
7236 ;; @r{Return a copy lacking all elements.}
7237 (nthcdr 5 '(pine fir oak maple))
7246 The @code{nthcdr} function takes the @sc{cdr} of a list repeatedly.
7247 The @code{nth} function takes the @sc{car} of the result returned by
7248 @code{nthcdr}. It returns the Nth element of the list.
7251 Thus, if it were not defined in C for speed, the definition of
7252 @code{nth} would be:
7257 "Returns the Nth element of LIST.
7258 N counts from zero. If LIST is not that long, nil is returned."
7259 (car (nthcdr n list)))
7264 (Originally, @code{nth} was defined in Emacs Lisp in @file{subr.el},
7265 but its definition was redone in C in the 1980s.)
7267 The @code{nth} function returns a single element of a list.
7268 This can be very convenient.
7270 Note that the elements are numbered from zero, not one. That is to
7271 say, the first element of a list, its @sc{car} is the zeroth element.
7272 This is called `zero-based' counting and often bothers people who
7273 are accustomed to the first element in a list being number one, which
7281 (nth 0 '("one" "two" "three"))
7284 (nth 1 '("one" "two" "three"))
7289 It is worth mentioning that @code{nth}, like @code{nthcdr} and
7290 @code{cdr}, does not change the original list---the function is
7291 non-destructive. This is in sharp contrast to the @code{setcar} and
7292 @code{setcdr} functions.
7295 @section @code{setcar}
7298 As you might guess from their names, the @code{setcar} and @code{setcdr}
7299 functions set the @sc{car} or the @sc{cdr} of a list to a new value.
7300 They actually change the original list, unlike @code{car} and @code{cdr}
7301 which leave the original list as it was. One way to find out how this
7302 works is to experiment. We will start with the @code{setcar} function.
7305 First, we can make a list and then set the value of a variable to the
7306 list, using the @code{setq} function. Here is a list of animals:
7309 (setq animals '(antelope giraffe lion tiger))
7313 If you are reading this in Info inside of GNU Emacs, you can evaluate
7314 this expression in the usual fashion, by positioning the cursor after
7315 the expression and typing @kbd{C-x C-e}. (I'm doing this right here
7316 as I write this. This is one of the advantages of having the
7317 interpreter built into the computing environment. Incidentally, when
7318 there is nothing on the line after the final parentheses, such as a
7319 comment, point can be on the next line. Thus, if your cursor is in
7320 the first column of the next line, you do not need to move it.
7321 Indeed, Emacs permits any amount of white space after the final
7325 When we evaluate the variable @code{animals}, we see that it is bound to
7326 the list @code{(antelope giraffe lion tiger)}:
7331 @result{} (antelope giraffe lion tiger)
7336 Put another way, the variable @code{animals} points to the list
7337 @code{(antelope giraffe lion tiger)}.
7339 Next, evaluate the function @code{setcar} while passing it two
7340 arguments, the variable @code{animals} and the quoted symbol
7341 @code{hippopotamus}; this is done by writing the three element list
7342 @code{(setcar animals 'hippopotamus)} and then evaluating it in the
7346 (setcar animals 'hippopotamus)
7351 After evaluating this expression, evaluate the variable @code{animals}
7352 again. You will see that the list of animals has changed:
7357 @result{} (hippopotamus giraffe lion tiger)
7362 The first element on the list, @code{antelope} is replaced by
7363 @code{hippopotamus}.
7365 So we can see that @code{setcar} did not add a new element to the list
7366 as @code{cons} would have; it replaced @code{antelope} with
7367 @code{hippopotamus}; it @emph{changed} the list.
7370 @section @code{setcdr}
7373 The @code{setcdr} function is similar to the @code{setcar} function,
7374 except that the function replaces the second and subsequent elements of
7375 a list rather than the first element.
7377 (To see how to change the last element of a list, look ahead to
7378 @ref{kill-new function, , The @code{kill-new} function}, which uses
7379 the @code{nthcdr} and @code{setcdr} functions.)
7382 To see how this works, set the value of the variable to a list of
7383 domesticated animals by evaluating the following expression:
7386 (setq domesticated-animals '(horse cow sheep goat))
7391 If you now evaluate the list, you will be returned the list
7392 @code{(horse cow sheep goat)}:
7396 domesticated-animals
7397 @result{} (horse cow sheep goat)
7402 Next, evaluate @code{setcdr} with two arguments, the name of the
7403 variable which has a list as its value, and the list to which the
7404 @sc{cdr} of the first list will be set;
7407 (setcdr domesticated-animals '(cat dog))
7411 If you evaluate this expression, the list @code{(cat dog)} will appear
7412 in the echo area. This is the value returned by the function. The
7413 result we are interested in is the ``side effect'', which we can see by
7414 evaluating the variable @code{domesticated-animals}:
7418 domesticated-animals
7419 @result{} (horse cat dog)
7424 Indeed, the list is changed from @code{(horse cow sheep goat)} to
7425 @code{(horse cat dog)}. The @sc{cdr} of the list is changed from
7426 @code{(cow sheep goat)} to @code{(cat dog)}.
7431 Construct a list of four birds by evaluating several expressions with
7432 @code{cons}. Find out what happens when you @code{cons} a list onto
7433 itself. Replace the first element of the list of four birds with a
7434 fish. Replace the rest of that list with a list of other fish.
7436 @node Cutting & Storing Text
7437 @chapter Cutting and Storing Text
7438 @cindex Cutting and storing text
7439 @cindex Storing and cutting text
7440 @cindex Killing text
7441 @cindex Clipping text
7442 @cindex Erasing text
7443 @cindex Deleting text
7445 Whenever you cut or clip text out of a buffer with a `kill' command in
7446 GNU Emacs, it is stored in a list and you can bring it back with a
7449 (The use of the word `kill' in Emacs for processes which specifically
7450 @emph{do not} destroy the values of the entities is an unfortunate
7451 historical accident. A much more appropriate word would be `clip' since
7452 that is what the kill commands do; they clip text out of a buffer and
7453 put it into storage from which it can be brought back. I have often
7454 been tempted to replace globally all occurrences of `kill' in the Emacs
7455 sources with `clip' and all occurrences of `killed' with `clipped'.)
7458 * Storing Text:: Text is stored in a list.
7459 * zap-to-char:: Cutting out text up to a character.
7460 * kill-region:: Cutting text out of a region.
7461 * copy-region-as-kill:: A definition for copying text.
7462 * Digression into C:: Minor note on C programming language macros.
7463 * defvar:: How to give a variable an initial value.
7464 * cons & search-fwd Review::
7465 * search Exercises::
7470 @unnumberedsec Storing Text in a List
7473 When text is cut out of a buffer, it is stored on a list. Successive
7474 pieces of text are stored on the list successively, so the list might
7478 ("a piece of text" "previous piece")
7483 The function @code{cons} can be used to create a new list from a piece
7484 of text (an `atom', to use the jargon) and an existing list, like
7489 (cons "another piece"
7490 '("a piece of text" "previous piece"))
7496 If you evaluate this expression, a list of three elements will appear in
7500 ("another piece" "a piece of text" "previous piece")
7503 With the @code{car} and @code{nthcdr} functions, you can retrieve
7504 whichever piece of text you want. For example, in the following code,
7505 @code{nthcdr 1 @dots{}} returns the list with the first item removed;
7506 and the @code{car} returns the first element of that remainder---the
7507 second element of the original list:
7511 (car (nthcdr 1 '("another piece"
7514 @result{} "a piece of text"
7518 The actual functions in Emacs are more complex than this, of course.
7519 The code for cutting and retrieving text has to be written so that
7520 Emacs can figure out which element in the list you want---the first,
7521 second, third, or whatever. In addition, when you get to the end of
7522 the list, Emacs should give you the first element of the list, rather
7523 than nothing at all.
7525 The list that holds the pieces of text is called the @dfn{kill ring}.
7526 This chapter leads up to a description of the kill ring and how it is
7527 used by first tracing how the @code{zap-to-char} function works. This
7528 function uses (or `calls') a function that invokes a function that
7529 manipulates the kill ring. Thus, before reaching the mountains, we
7530 climb the foothills.
7532 A subsequent chapter describes how text that is cut from the buffer is
7533 retrieved. @xref{Yanking, , Yanking Text Back}.
7536 @section @code{zap-to-char}
7539 @c FIXME remove obsolete stuff
7540 The @code{zap-to-char} function changed little between GNU Emacs
7541 version 19 and GNU Emacs version 22. However, @code{zap-to-char}
7542 calls another function, @code{kill-region}, which enjoyed a major
7545 The @code{kill-region} function in Emacs 19 is complex, but does not
7546 use code that is important at this time. We will skip it.
7548 The @code{kill-region} function in Emacs 22 is easier to read than the
7549 same function in Emacs 19 and introduces a very important concept,
7550 that of error handling. We will walk through the function.
7552 But first, let us look at the interactive @code{zap-to-char} function.
7555 * Complete zap-to-char:: The complete implementation.
7556 * zap-to-char interactive:: A three part interactive expression.
7557 * zap-to-char body:: A short overview.
7558 * search-forward:: How to search for a string.
7559 * progn:: The @code{progn} special form.
7560 * Summing up zap-to-char:: Using @code{point} and @code{search-forward}.
7564 @node Complete zap-to-char
7565 @unnumberedsubsec The Complete @code{zap-to-char} Implementation
7568 The @code{zap-to-char} function removes the text in the region between
7569 the location of the cursor (i.e., of point) up to and including the
7570 next occurrence of a specified character. The text that
7571 @code{zap-to-char} removes is put in the kill ring; and it can be
7572 retrieved from the kill ring by typing @kbd{C-y} (@code{yank}). If
7573 the command is given an argument, it removes text through that number
7574 of occurrences. Thus, if the cursor were at the beginning of this
7575 sentence and the character were @samp{s}, @samp{Thus} would be
7576 removed. If the argument were two, @samp{Thus, if the curs} would be
7577 removed, up to and including the @samp{s} in @samp{cursor}.
7579 If the specified character is not found, @code{zap-to-char} will say
7580 ``Search failed'', tell you the character you typed, and not remove
7583 In order to determine how much text to remove, @code{zap-to-char} uses
7584 a search function. Searches are used extensively in code that
7585 manipulates text, and we will focus attention on them as well as on the
7589 @c GNU Emacs version 19
7590 (defun zap-to-char (arg char) ; version 19 implementation
7591 "Kill up to and including ARG'th occurrence of CHAR.
7592 Goes backward if ARG is negative; error if CHAR not found."
7593 (interactive "*p\ncZap to char: ")
7594 (kill-region (point)
7597 (char-to-string char) nil nil arg)
7602 Here is the complete text of the version 22 implementation of the function:
7607 (defun zap-to-char (arg char)
7608 "Kill up to and including ARG'th occurrence of CHAR.
7609 Case is ignored if `case-fold-search' is non-nil in the current buffer.
7610 Goes backward if ARG is negative; error if CHAR not found."
7611 (interactive "p\ncZap to char: ")
7612 (if (char-table-p translation-table-for-input)
7613 (setq char (or (aref translation-table-for-input char) char)))
7614 (kill-region (point) (progn
7615 (search-forward (char-to-string char)
7621 The documentation is thorough. You do need to know the jargon meaning
7624 @node zap-to-char interactive
7625 @subsection The @code{interactive} Expression
7628 The interactive expression in the @code{zap-to-char} command looks like
7632 (interactive "p\ncZap to char: ")
7635 The part within quotation marks, @code{"p\ncZap to char:@: "}, specifies
7636 two different things. First, and most simply, is the @samp{p}.
7637 This part is separated from the next part by a newline, @samp{\n}.
7638 The @samp{p} means that the first argument to the function will be
7639 passed the value of a `processed prefix'. The prefix argument is
7640 passed by typing @kbd{C-u} and a number, or @kbd{M-} and a number. If
7641 the function is called interactively without a prefix, 1 is passed to
7644 The second part of @code{"p\ncZap to char:@: "} is
7645 @samp{cZap to char:@: }. In this part, the lower case @samp{c}
7646 indicates that @code{interactive} expects a prompt and that the
7647 argument will be a character. The prompt follows the @samp{c} and is
7648 the string @samp{Zap to char:@: } (with a space after the colon to
7651 What all this does is prepare the arguments to @code{zap-to-char} so they
7652 are of the right type, and give the user a prompt.
7654 In a read-only buffer, the @code{zap-to-char} function copies the text
7655 to the kill ring, but does not remove it. The echo area displays a
7656 message saying that the buffer is read-only. Also, the terminal may
7657 beep or blink at you.
7659 @node zap-to-char body
7660 @subsection The Body of @code{zap-to-char}
7662 The body of the @code{zap-to-char} function contains the code that
7663 kills (that is, removes) the text in the region from the current
7664 position of the cursor up to and including the specified character.
7666 The first part of the code looks like this:
7669 (if (char-table-p translation-table-for-input)
7670 (setq char (or (aref translation-table-for-input char) char)))
7671 (kill-region (point) (progn
7672 (search-forward (char-to-string char) nil nil arg)
7677 @code{char-table-p} is an hitherto unseen function. It determines
7678 whether its argument is a character table. When it is, it sets the
7679 character passed to @code{zap-to-char} to one of them, if that
7680 character exists, or to the character itself. (This becomes important
7681 for certain characters in non-European languages. The @code{aref}
7682 function extracts an element from an array. It is an array-specific
7683 function that is not described in this document. @xref{Arrays, ,
7684 Arrays, elisp, The GNU Emacs Lisp Reference Manual}.)
7687 @code{(point)} is the current position of the cursor.
7689 The next part of the code is an expression using @code{progn}. The body
7690 of the @code{progn} consists of calls to @code{search-forward} and
7693 It is easier to understand how @code{progn} works after learning about
7694 @code{search-forward}, so we will look at @code{search-forward} and
7695 then at @code{progn}.
7697 @node search-forward
7698 @subsection The @code{search-forward} Function
7699 @findex search-forward
7701 The @code{search-forward} function is used to locate the
7702 zapped-for-character in @code{zap-to-char}. If the search is
7703 successful, @code{search-forward} leaves point immediately after the
7704 last character in the target string. (In @code{zap-to-char}, the
7705 target string is just one character long. @code{zap-to-char} uses the
7706 function @code{char-to-string} to ensure that the computer treats that
7707 character as a string.) If the search is backwards,
7708 @code{search-forward} leaves point just before the first character in
7709 the target. Also, @code{search-forward} returns @code{t} for true.
7710 (Moving point is therefore a `side effect'.)
7713 In @code{zap-to-char}, the @code{search-forward} function looks like this:
7716 (search-forward (char-to-string char) nil nil arg)
7719 The @code{search-forward} function takes four arguments:
7723 The first argument is the target, what is searched for. This must be a
7724 string, such as @samp{"z"}.
7726 As it happens, the argument passed to @code{zap-to-char} is a single
7727 character. Because of the way computers are built, the Lisp
7728 interpreter may treat a single character as being different from a
7729 string of characters. Inside the computer, a single character has a
7730 different electronic format than a string of one character. (A single
7731 character can often be recorded in the computer using exactly one
7732 byte; but a string may be longer, and the computer needs to be ready
7733 for this.) Since the @code{search-forward} function searches for a
7734 string, the character that the @code{zap-to-char} function receives as
7735 its argument must be converted inside the computer from one format to
7736 the other; otherwise the @code{search-forward} function will fail.
7737 The @code{char-to-string} function is used to make this conversion.
7740 The second argument bounds the search; it is specified as a position in
7741 the buffer. In this case, the search can go to the end of the buffer,
7742 so no bound is set and the second argument is @code{nil}.
7745 The third argument tells the function what it should do if the search
7746 fails---it can signal an error (and print a message) or it can return
7747 @code{nil}. A @code{nil} as the third argument causes the function to
7748 signal an error when the search fails.
7751 The fourth argument to @code{search-forward} is the repeat count---how
7752 many occurrences of the string to look for. This argument is optional
7753 and if the function is called without a repeat count, this argument is
7754 passed the value 1. If this argument is negative, the search goes
7759 In template form, a @code{search-forward} expression looks like this:
7763 (search-forward "@var{target-string}"
7764 @var{limit-of-search}
7765 @var{what-to-do-if-search-fails}
7770 We will look at @code{progn} next.
7773 @subsection The @code{progn} Special Form
7776 @code{progn} is a special form that causes each of its arguments to be
7777 evaluated in sequence and then returns the value of the last one. The
7778 preceding expressions are evaluated only for the side effects they
7779 perform. The values produced by them are discarded.
7782 The template for a @code{progn} expression is very simple:
7791 In @code{zap-to-char}, the @code{progn} expression has to do two things:
7792 put point in exactly the right position; and return the location of
7793 point so that @code{kill-region} will know how far to kill to.
7795 The first argument to the @code{progn} is @code{search-forward}. When
7796 @code{search-forward} finds the string, the function leaves point
7797 immediately after the last character in the target string. (In this
7798 case the target string is just one character long.) If the search is
7799 backwards, @code{search-forward} leaves point just before the first
7800 character in the target. The movement of point is a side effect.
7802 The second and last argument to @code{progn} is the expression
7803 @code{(point)}. This expression returns the value of point, which in
7804 this case will be the location to which it has been moved by
7805 @code{search-forward}. (In the source, a line that tells the function
7806 to go to the previous character, if it is going forward, was commented
7807 out in 1999; I don't remember whether that feature or mis-feature was
7808 ever a part of the distributed source.) The value of @code{point} is
7809 returned by the @code{progn} expression and is passed to
7810 @code{kill-region} as @code{kill-region}'s second argument.
7812 @node Summing up zap-to-char
7813 @subsection Summing up @code{zap-to-char}
7815 Now that we have seen how @code{search-forward} and @code{progn} work,
7816 we can see how the @code{zap-to-char} function works as a whole.
7818 The first argument to @code{kill-region} is the position of the cursor
7819 when the @code{zap-to-char} command is given---the value of point at
7820 that time. Within the @code{progn}, the search function then moves
7821 point to just after the zapped-to-character and @code{point} returns the
7822 value of this location. The @code{kill-region} function puts together
7823 these two values of point, the first one as the beginning of the region
7824 and the second one as the end of the region, and removes the region.
7826 The @code{progn} special form is necessary because the
7827 @code{kill-region} command takes two arguments; and it would fail if
7828 @code{search-forward} and @code{point} expressions were written in
7829 sequence as two additional arguments. The @code{progn} expression is
7830 a single argument to @code{kill-region} and returns the one value that
7831 @code{kill-region} needs for its second argument.
7834 @section @code{kill-region}
7837 The @code{zap-to-char} function uses the @code{kill-region} function.
7838 This function clips text from a region and copies that text to
7839 the kill ring, from which it may be retrieved.
7844 (defun kill-region (beg end &optional yank-handler)
7845 "Kill (\"cut\") text between point and mark.
7846 This deletes the text from the buffer and saves it in the kill ring.
7847 The command \\[yank] can retrieve it from there.
7848 \(If you want to kill and then yank immediately, use \\[kill-ring-save].)
7850 If you want to append the killed region to the last killed text,
7851 use \\[append-next-kill] before \\[kill-region].
7853 If the buffer is read-only, Emacs will beep and refrain from deleting
7854 the text, but put the text in the kill ring anyway. This means that
7855 you can use the killing commands to copy text from a read-only buffer.
7857 This is the primitive for programs to kill text (as opposed to deleting it).
7858 Supply two arguments, character positions indicating the stretch of text
7860 Any command that calls this function is a \"kill command\".
7861 If the previous command was also a kill command,
7862 the text killed this time appends to the text killed last time
7863 to make one entry in the kill ring.
7865 In Lisp code, optional third arg YANK-HANDLER, if non-nil,
7866 specifies the yank-handler text property to be set on the killed
7867 text. See `insert-for-yank'."
7868 ;; Pass point first, then mark, because the order matters
7869 ;; when calling kill-append.
7870 (interactive (list (point) (mark)))
7871 (unless (and beg end)
7872 (error "The mark is not set now, so there is no region"))
7874 (let ((string (filter-buffer-substring beg end t)))
7875 (when string ;STRING is nil if BEG = END
7876 ;; Add that string to the kill ring, one way or another.
7877 (if (eq last-command 'kill-region)
7878 (kill-append string (< end beg) yank-handler)
7879 (kill-new string nil yank-handler)))
7880 (when (or string (eq last-command 'kill-region))
7881 (setq this-command 'kill-region))
7883 ((buffer-read-only text-read-only)
7884 ;; The code above failed because the buffer, or some of the characters
7885 ;; in the region, are read-only.
7886 ;; We should beep, in case the user just isn't aware of this.
7887 ;; However, there's no harm in putting
7888 ;; the region's text in the kill ring, anyway.
7889 (copy-region-as-kill beg end)
7890 ;; Set this-command now, so it will be set even if we get an error.
7891 (setq this-command 'kill-region)
7892 ;; This should barf, if appropriate, and give us the correct error.
7893 (if kill-read-only-ok
7894 (progn (message "Read only text copied to kill ring") nil)
7895 ;; Signal an error if the buffer is read-only.
7896 (barf-if-buffer-read-only)
7897 ;; If the buffer isn't read-only, the text is.
7898 (signal 'text-read-only (list (current-buffer)))))))
7901 The Emacs 22 version of that function uses @code{condition-case} and
7902 @code{copy-region-as-kill}, both of which we will explain.
7903 @code{condition-case} is an important special form.
7905 In essence, the @code{kill-region} function calls
7906 @code{condition-case}, which takes three arguments. In this function,
7907 the first argument does nothing. The second argument contains the
7908 code that does the work when all goes well. The third argument
7909 contains the code that is called in the event of an error.
7912 * Complete kill-region:: The function definition.
7913 * condition-case:: Dealing with a problem.
7918 @node Complete kill-region
7919 @unnumberedsubsec The Complete @code{kill-region} Definition
7923 We will go through the @code{condition-case} code in a moment. First,
7924 let us look at the definition of @code{kill-region}, with comments
7930 (defun kill-region (beg end)
7931 "Kill (\"cut\") text between point and mark.
7932 This deletes the text from the buffer and saves it in the kill ring.
7933 The command \\[yank] can retrieve it from there. @dots{} "
7937 ;; @bullet{} Since order matters, pass point first.
7938 (interactive (list (point) (mark)))
7939 ;; @bullet{} And tell us if we cannot cut the text.
7940 ;; `unless' is an `if' without a then-part.
7941 (unless (and beg end)
7942 (error "The mark is not set now, so there is no region"))
7946 ;; @bullet{} `condition-case' takes three arguments.
7947 ;; If the first argument is nil, as it is here,
7948 ;; information about the error signal is not
7949 ;; stored for use by another function.
7954 ;; @bullet{} The second argument to `condition-case' tells the
7955 ;; Lisp interpreter what to do when all goes well.
7959 ;; It starts with a `let' function that extracts the string
7960 ;; and tests whether it exists. If so (that is what the
7961 ;; `when' checks), it calls an `if' function that determines
7962 ;; whether the previous command was another call to
7963 ;; `kill-region'; if it was, then the new text is appended to
7964 ;; the previous text; if not, then a different function,
7965 ;; `kill-new', is called.
7969 ;; The `kill-append' function concatenates the new string and
7970 ;; the old. The `kill-new' function inserts text into a new
7971 ;; item in the kill ring.
7975 ;; `when' is an `if' without an else-part. The second `when'
7976 ;; again checks whether the current string exists; in
7977 ;; addition, it checks whether the previous command was
7978 ;; another call to `kill-region'. If one or the other
7979 ;; condition is true, then it sets the current command to
7980 ;; be `kill-region'.
7983 (let ((string (filter-buffer-substring beg end t)))
7984 (when string ;STRING is nil if BEG = END
7985 ;; Add that string to the kill ring, one way or another.
7986 (if (eq last-command 'kill-region)
7989 ;; @minus{} `yank-handler' is an optional argument to
7990 ;; `kill-region' that tells the `kill-append' and
7991 ;; `kill-new' functions how deal with properties
7992 ;; added to the text, such as `bold' or `italics'.
7993 (kill-append string (< end beg) yank-handler)
7994 (kill-new string nil yank-handler)))
7995 (when (or string (eq last-command 'kill-region))
7996 (setq this-command 'kill-region))
8001 ;; @bullet{} The third argument to `condition-case' tells the interpreter
8002 ;; what to do with an error.
8005 ;; The third argument has a conditions part and a body part.
8006 ;; If the conditions are met (in this case,
8007 ;; if text or buffer are read-only)
8008 ;; then the body is executed.
8011 ;; The first part of the third argument is the following:
8012 ((buffer-read-only text-read-only) ;; the if-part
8013 ;; @dots{} the then-part
8014 (copy-region-as-kill beg end)
8017 ;; Next, also as part of the then-part, set this-command, so
8018 ;; it will be set in an error
8019 (setq this-command 'kill-region)
8020 ;; Finally, in the then-part, send a message if you may copy
8021 ;; the text to the kill ring without signaling an error, but
8022 ;; don't if you may not.
8025 (if kill-read-only-ok
8026 (progn (message "Read only text copied to kill ring") nil)
8027 (barf-if-buffer-read-only)
8028 ;; If the buffer isn't read-only, the text is.
8029 (signal 'text-read-only (list (current-buffer)))))
8037 (defun kill-region (beg end)
8038 "Kill between point and mark.
8039 The text is deleted but saved in the kill ring."
8044 ;; 1. `condition-case' takes three arguments.
8045 ;; If the first argument is nil, as it is here,
8046 ;; information about the error signal is not
8047 ;; stored for use by another function.
8052 ;; 2. The second argument to `condition-case'
8053 ;; tells the Lisp interpreter what to do when all goes well.
8057 ;; The `delete-and-extract-region' function usually does the
8058 ;; work. If the beginning and ending of the region are both
8059 ;; the same, then the variable `string' will be empty, or nil
8060 (let ((string (delete-and-extract-region beg end)))
8064 ;; `when' is an `if' clause that cannot take an `else-part'.
8065 ;; Emacs normally sets the value of `last-command' to the
8066 ;; previous command.
8069 ;; `kill-append' concatenates the new string and the old.
8070 ;; `kill-new' inserts text into a new item in the kill ring.
8072 (if (eq last-command 'kill-region)
8073 ;; if true, prepend string
8074 (kill-append string (< end beg))
8076 (setq this-command 'kill-region))
8080 ;; 3. The third argument to `condition-case' tells the interpreter
8081 ;; what to do with an error.
8084 ;; The third argument has a conditions part and a body part.
8085 ;; If the conditions are met (in this case,
8086 ;; if text or buffer are read-only)
8087 ;; then the body is executed.
8090 ((buffer-read-only text-read-only) ;; this is the if-part
8092 (copy-region-as-kill beg end)
8095 (if kill-read-only-ok ;; usually this variable is nil
8096 (message "Read only text copied to kill ring")
8097 ;; or else, signal an error if the buffer is read-only;
8098 (barf-if-buffer-read-only)
8099 ;; and, in any case, signal that the text is read-only.
8100 (signal 'text-read-only (list (current-buffer)))))))
8105 @node condition-case
8106 @subsection @code{condition-case}
8107 @findex condition-case
8109 As we have seen earlier (@pxref{Making Errors, , Generate an Error
8110 Message}), when the Emacs Lisp interpreter has trouble evaluating an
8111 expression, it provides you with help; in the jargon, this is called
8112 ``signaling an error''. Usually, the computer stops the program and
8113 shows you a message.
8115 However, some programs undertake complicated actions. They should not
8116 simply stop on an error. In the @code{kill-region} function, the most
8117 likely error is that you will try to kill text that is read-only and
8118 cannot be removed. So the @code{kill-region} function contains code
8119 to handle this circumstance. This code, which makes up the body of
8120 the @code{kill-region} function, is inside of a @code{condition-case}
8124 The template for @code{condition-case} looks like this:
8131 @var{error-handler}@dots{})
8135 The second argument, @var{bodyform}, is straightforward. The
8136 @code{condition-case} special form causes the Lisp interpreter to
8137 evaluate the code in @var{bodyform}. If no error occurs, the special
8138 form returns the code's value and produces the side-effects, if any.
8140 In short, the @var{bodyform} part of a @code{condition-case}
8141 expression determines what should happen when everything works
8144 However, if an error occurs, among its other actions, the function
8145 generating the error signal will define one or more error condition
8148 An error handler is the third argument to @code{condition case}.
8149 An error handler has two parts, a @var{condition-name} and a
8150 @var{body}. If the @var{condition-name} part of an error handler
8151 matches a condition name generated by an error, then the @var{body}
8152 part of the error handler is run.
8154 As you will expect, the @var{condition-name} part of an error handler
8155 may be either a single condition name or a list of condition names.
8157 Also, a complete @code{condition-case} expression may contain more
8158 than one error handler. When an error occurs, the first applicable
8161 Lastly, the first argument to the @code{condition-case} expression,
8162 the @var{var} argument, is sometimes bound to a variable that
8163 contains information about the error. However, if that argument is
8164 nil, as is the case in @code{kill-region}, that information is
8168 In brief, in the @code{kill-region} function, the code
8169 @code{condition-case} works like this:
8173 @var{If no errors}, @var{run only this code}
8174 @var{but}, @var{if errors}, @var{run this other code}.
8181 copy-region-as-kill is short, 12 lines, and uses
8182 filter-buffer-substring, which is longer, 39 lines
8183 and has delete-and-extract-region in it.
8184 delete-and-extract-region is written in C.
8186 see Initializing a Variable with @code{defvar}
8188 Initializing a Variable with @code{defvar} includes line 8350
8192 @subsection Lisp macro
8196 The part of the @code{condition-case} expression that is evaluated in
8197 the expectation that all goes well has a @code{when}. The code uses
8198 @code{when} to determine whether the @code{string} variable points to
8201 A @code{when} expression is simply a programmers' convenience. It is
8202 an @code{if} without the possibility of an else clause. In your mind,
8203 you can replace @code{when} with @code{if} and understand what goes
8204 on. That is what the Lisp interpreter does.
8206 Technically speaking, @code{when} is a Lisp macro. A Lisp macro
8207 enables you to define new control constructs and other language
8208 features. It tells the interpreter how to compute another Lisp
8209 expression which will in turn compute the value. In this case, the
8210 `other expression' is an @code{if} expression.
8212 The @code{kill-region} function definition also has an @code{unless}
8213 macro; it is the converse of @code{when}. The @code{unless} macro is
8214 an @code{if} without a then clause
8216 For more about Lisp macros, see @ref{Macros, , Macros, elisp, The GNU
8217 Emacs Lisp Reference Manual}. The C programming language also
8218 provides macros. These are different, but also useful.
8221 We will briefly look at C macros in
8222 @ref{Digression into C}.
8226 Regarding the @code{when} macro, in the @code{condition-case}
8227 expression, when the string has content, then another conditional
8228 expression is executed. This is an @code{if} with both a then-part
8233 (if (eq last-command 'kill-region)
8234 (kill-append string (< end beg) yank-handler)
8235 (kill-new string nil yank-handler))
8239 The then-part is evaluated if the previous command was another call to
8240 @code{kill-region}; if not, the else-part is evaluated.
8242 @code{yank-handler} is an optional argument to @code{kill-region} that
8243 tells the @code{kill-append} and @code{kill-new} functions how deal
8244 with properties added to the text, such as `bold' or `italics'.
8246 @code{last-command} is a variable that comes with Emacs that we have
8247 not seen before. Normally, whenever a function is executed, Emacs
8248 sets the value of @code{last-command} to the previous command.
8251 In this segment of the definition, the @code{if} expression checks
8252 whether the previous command was @code{kill-region}. If it was,
8255 (kill-append string (< end beg) yank-handler)
8259 concatenates a copy of the newly clipped text to the just previously
8260 clipped text in the kill ring.
8262 @node copy-region-as-kill
8263 @section @code{copy-region-as-kill}
8264 @findex copy-region-as-kill
8267 The @code{copy-region-as-kill} function copies a region of text from a
8268 buffer and (via either @code{kill-append} or @code{kill-new}) saves it
8269 in the @code{kill-ring}.
8271 If you call @code{copy-region-as-kill} immediately after a
8272 @code{kill-region} command, Emacs appends the newly copied text to the
8273 previously copied text. This means that if you yank back the text, you
8274 get it all, from both this and the previous operation. On the other
8275 hand, if some other command precedes the @code{copy-region-as-kill},
8276 the function copies the text into a separate entry in the kill ring.
8279 * Complete copy-region-as-kill:: The complete function definition.
8280 * copy-region-as-kill body:: The body of @code{copy-region-as-kill}.
8284 @node Complete copy-region-as-kill
8285 @unnumberedsubsec The complete @code{copy-region-as-kill} function definition
8289 Here is the complete text of the version 22 @code{copy-region-as-kill}
8294 (defun copy-region-as-kill (beg end)
8295 "Save the region as if killed, but don't kill it.
8296 In Transient Mark mode, deactivate the mark.
8297 If `interprogram-cut-function' is non-nil, also save the text for a window
8298 system cut and paste."
8302 (if (eq last-command 'kill-region)
8303 (kill-append (filter-buffer-substring beg end) (< end beg))
8304 (kill-new (filter-buffer-substring beg end)))
8307 (if transient-mark-mode
8308 (setq deactivate-mark t))
8314 As usual, this function can be divided into its component parts:
8318 (defun copy-region-as-kill (@var{argument-list})
8319 "@var{documentation}@dots{}"
8325 The arguments are @code{beg} and @code{end} and the function is
8326 interactive with @code{"r"}, so the two arguments must refer to the
8327 beginning and end of the region. If you have been reading though this
8328 document from the beginning, understanding these parts of a function is
8329 almost becoming routine.
8331 The documentation is somewhat confusing unless you remember that the
8332 word `kill' has a meaning different from usual. The `Transient Mark'
8333 and @code{interprogram-cut-function} comments explain certain
8336 After you once set a mark, a buffer always contains a region. If you
8337 wish, you can use Transient Mark mode to highlight the region
8338 temporarily. (No one wants to highlight the region all the time, so
8339 Transient Mark mode highlights it only at appropriate times. Many
8340 people turn off Transient Mark mode, so the region is never
8343 Also, a windowing system allows you to copy, cut, and paste among
8344 different programs. In the X windowing system, for example, the
8345 @code{interprogram-cut-function} function is @code{x-select-text},
8346 which works with the windowing system's equivalent of the Emacs kill
8349 The body of the @code{copy-region-as-kill} function starts with an
8350 @code{if} clause. What this clause does is distinguish between two
8351 different situations: whether or not this command is executed
8352 immediately after a previous @code{kill-region} command. In the first
8353 case, the new region is appended to the previously copied text.
8354 Otherwise, it is inserted into the beginning of the kill ring as a
8355 separate piece of text from the previous piece.
8357 The last two lines of the function prevent the region from lighting up
8358 if Transient Mark mode is turned on.
8360 The body of @code{copy-region-as-kill} merits discussion in detail.
8362 @node copy-region-as-kill body
8363 @subsection The Body of @code{copy-region-as-kill}
8365 The @code{copy-region-as-kill} function works in much the same way as
8366 the @code{kill-region} function. Both are written so that two or more
8367 kills in a row combine their text into a single entry. If you yank
8368 back the text from the kill ring, you get it all in one piece.
8369 Moreover, kills that kill forward from the current position of the
8370 cursor are added to the end of the previously copied text and commands
8371 that copy text backwards add it to the beginning of the previously
8372 copied text. This way, the words in the text stay in the proper
8375 Like @code{kill-region}, the @code{copy-region-as-kill} function makes
8376 use of the @code{last-command} variable that keeps track of the
8377 previous Emacs command.
8380 * last-command & this-command::
8381 * kill-append function::
8382 * kill-new function::
8386 @node last-command & this-command
8387 @unnumberedsubsubsec @code{last-command} and @code{this-command}
8390 Normally, whenever a function is executed, Emacs sets the value of
8391 @code{this-command} to the function being executed (which in this case
8392 would be @code{copy-region-as-kill}). At the same time, Emacs sets
8393 the value of @code{last-command} to the previous value of
8394 @code{this-command}.
8396 In the first part of the body of the @code{copy-region-as-kill}
8397 function, an @code{if} expression determines whether the value of
8398 @code{last-command} is @code{kill-region}. If so, the then-part of
8399 the @code{if} expression is evaluated; it uses the @code{kill-append}
8400 function to concatenate the text copied at this call to the function
8401 with the text already in the first element (the @sc{car}) of the kill
8402 ring. On the other hand, if the value of @code{last-command} is not
8403 @code{kill-region}, then the @code{copy-region-as-kill} function
8404 attaches a new element to the kill ring using the @code{kill-new}
8408 The @code{if} expression reads as follows; it uses @code{eq}:
8412 (if (eq last-command 'kill-region)
8414 (kill-append (filter-buffer-substring beg end) (< end beg))
8416 (kill-new (filter-buffer-substring beg end)))
8420 @findex filter-buffer-substring
8421 (The @code{filter-buffer-substring} function returns a filtered
8422 substring of the buffer, if any. Optionally---the arguments are not
8423 here, so neither is done---the function may delete the initial text or
8424 return the text without its properties; this function is a replacement
8425 for the older @code{buffer-substring} function, which came before text
8426 properties were implemented.)
8428 @findex eq @r{(example of use)}
8430 The @code{eq} function tests whether its first argument is the same Lisp
8431 object as its second argument. The @code{eq} function is similar to the
8432 @code{equal} function in that it is used to test for equality, but
8433 differs in that it determines whether two representations are actually
8434 the same object inside the computer, but with different names.
8435 @code{equal} determines whether the structure and contents of two
8436 expressions are the same.
8438 If the previous command was @code{kill-region}, then the Emacs Lisp
8439 interpreter calls the @code{kill-append} function
8441 @node kill-append function
8442 @unnumberedsubsubsec The @code{kill-append} function
8446 The @code{kill-append} function looks like this:
8451 (defun kill-append (string before-p &optional yank-handler)
8452 "Append STRING to the end of the latest kill in the kill ring.
8453 If BEFORE-P is non-nil, prepend STRING to the kill.
8455 (let* ((cur (car kill-ring)))
8456 (kill-new (if before-p (concat string cur) (concat cur string))
8457 (or (= (length cur) 0)
8459 (get-text-property 0 'yank-handler cur)))
8466 (defun kill-append (string before-p)
8467 "Append STRING to the end of the latest kill in the kill ring.
8468 If BEFORE-P is non-nil, prepend STRING to the kill.
8469 If `interprogram-cut-function' is set, pass the resulting kill to
8471 (kill-new (if before-p
8472 (concat string (car kill-ring))
8473 (concat (car kill-ring) string))
8478 The @code{kill-append} function is fairly straightforward. It uses
8479 the @code{kill-new} function, which we will discuss in more detail in
8482 (Also, the function provides an optional argument called
8483 @code{yank-handler}; when invoked, this argument tells the function
8484 how to deal with properties added to the text, such as `bold' or
8487 @c !!! bug in GNU Emacs 22 version of kill-append ?
8488 It has a @code{let*} function to set the value of the first element of
8489 the kill ring to @code{cur}. (I do not know why the function does not
8490 use @code{let} instead; only one value is set in the expression.
8491 Perhaps this is a bug that produces no problems?)
8493 Consider the conditional that is one of the two arguments to
8494 @code{kill-new}. It uses @code{concat} to concatenate the new text to
8495 the @sc{car} of the kill ring. Whether it prepends or appends the
8496 text depends on the results of an @code{if} expression:
8500 (if before-p ; @r{if-part}
8501 (concat string cur) ; @r{then-part}
8502 (concat cur string)) ; @r{else-part}
8507 If the region being killed is before the region that was killed in the
8508 last command, then it should be prepended before the material that was
8509 saved in the previous kill; and conversely, if the killed text follows
8510 what was just killed, it should be appended after the previous text.
8511 The @code{if} expression depends on the predicate @code{before-p} to
8512 decide whether the newly saved text should be put before or after the
8513 previously saved text.
8515 The symbol @code{before-p} is the name of one of the arguments to
8516 @code{kill-append}. When the @code{kill-append} function is
8517 evaluated, it is bound to the value returned by evaluating the actual
8518 argument. In this case, this is the expression @code{(< end beg)}.
8519 This expression does not directly determine whether the killed text in
8520 this command is located before or after the kill text of the last
8521 command; what it does is determine whether the value of the variable
8522 @code{end} is less than the value of the variable @code{beg}. If it
8523 is, it means that the user is most likely heading towards the
8524 beginning of the buffer. Also, the result of evaluating the predicate
8525 expression, @code{(< end beg)}, will be true and the text will be
8526 prepended before the previous text. On the other hand, if the value of
8527 the variable @code{end} is greater than the value of the variable
8528 @code{beg}, the text will be appended after the previous text.
8531 When the newly saved text will be prepended, then the string with the new
8532 text will be concatenated before the old text:
8540 But if the text will be appended, it will be concatenated
8544 (concat cur string))
8547 To understand how this works, we first need to review the
8548 @code{concat} function. The @code{concat} function links together or
8549 unites two strings of text. The result is a string. For example:
8553 (concat "abc" "def")
8559 (car '("first element" "second element")))
8560 @result{} "new first element"
8563 '("first element" "second element")) " modified")
8564 @result{} "first element modified"
8568 We can now make sense of @code{kill-append}: it modifies the contents
8569 of the kill ring. The kill ring is a list, each element of which is
8570 saved text. The @code{kill-append} function uses the @code{kill-new}
8571 function which in turn uses the @code{setcar} function.
8573 @node kill-new function
8574 @unnumberedsubsubsec The @code{kill-new} function
8577 @c in GNU Emacs 22, additional documentation to kill-new:
8579 Optional third arguments YANK-HANDLER controls how the STRING is later
8580 inserted into a buffer; see `insert-for-yank' for details.
8581 When a yank handler is specified, STRING must be non-empty (the yank
8582 handler, if non-nil, is stored as a `yank-handler' text property on STRING).
8584 When the yank handler has a non-nil PARAM element, the original STRING
8585 argument is not used by `insert-for-yank'. However, since Lisp code
8586 may access and use elements from the kill ring directly, the STRING
8587 argument should still be a \"useful\" string for such uses."
8590 The @code{kill-new} function looks like this:
8594 (defun kill-new (string &optional replace yank-handler)
8595 "Make STRING the latest kill in the kill ring.
8596 Set `kill-ring-yank-pointer' to point to it.
8598 If `interprogram-cut-function' is non-nil, apply it to STRING.
8599 Optional second argument REPLACE non-nil means that STRING will replace
8600 the front of the kill ring, rather than being added to the list.
8604 (if (> (length string) 0)
8606 (put-text-property 0 (length string)
8607 'yank-handler yank-handler string))
8609 (signal 'args-out-of-range
8610 (list string "yank-handler specified for empty string"))))
8613 (if (fboundp 'menu-bar-update-yank-menu)
8614 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8617 (if (and replace kill-ring)
8618 (setcar kill-ring string)
8619 (push string kill-ring)
8620 (if (> (length kill-ring) kill-ring-max)
8621 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8624 (setq kill-ring-yank-pointer kill-ring)
8625 (if interprogram-cut-function
8626 (funcall interprogram-cut-function string (not replace))))
8631 (defun kill-new (string &optional replace)
8632 "Make STRING the latest kill in the kill ring.
8633 Set the kill-ring-yank pointer to point to it.
8634 If `interprogram-cut-function' is non-nil, apply it to STRING.
8635 Optional second argument REPLACE non-nil means that STRING will replace
8636 the front of the kill ring, rather than being added to the list."
8637 (and (fboundp 'menu-bar-update-yank-menu)
8638 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8639 (if (and replace kill-ring)
8640 (setcar kill-ring string)
8641 (setq kill-ring (cons string kill-ring))
8642 (if (> (length kill-ring) kill-ring-max)
8643 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8644 (setq kill-ring-yank-pointer kill-ring)
8645 (if interprogram-cut-function
8646 (funcall interprogram-cut-function string (not replace))))
8649 (Notice that the function is not interactive.)
8651 As usual, we can look at this function in parts.
8653 The function definition has an optional @code{yank-handler} argument,
8654 which when invoked tells the function how to deal with properties
8655 added to the text, such as `bold' or `italics'. We will skip that.
8658 The first line of the documentation makes sense:
8661 Make STRING the latest kill in the kill ring.
8665 Let's skip over the rest of the documentation for the moment.
8668 Also, let's skip over the initial @code{if} expression and those lines
8669 of code involving @code{menu-bar-update-yank-menu}. We will explain
8673 The critical lines are these:
8677 (if (and replace kill-ring)
8679 (setcar kill-ring string)
8683 (push string kill-ring)
8686 (setq kill-ring (cons string kill-ring))
8687 (if (> (length kill-ring) kill-ring-max)
8688 ;; @r{avoid overly long kill ring}
8689 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8692 (setq kill-ring-yank-pointer kill-ring)
8693 (if interprogram-cut-function
8694 (funcall interprogram-cut-function string (not replace))))
8698 The conditional test is @w{@code{(and replace kill-ring)}}.
8699 This will be true when two conditions are met: the kill ring has
8700 something in it, and the @code{replace} variable is true.
8703 When the @code{kill-append} function sets @code{replace} to be true
8704 and when the kill ring has at least one item in it, the @code{setcar}
8705 expression is executed:
8708 (setcar kill-ring string)
8711 The @code{setcar} function actually changes the first element of the
8712 @code{kill-ring} list to the value of @code{string}. It replaces the
8716 On the other hand, if the kill ring is empty, or replace is false, the
8717 else-part of the condition is executed:
8720 (push string kill-ring)
8725 @code{push} puts its first argument onto the second. It is similar to
8729 (setq kill-ring (cons string kill-ring))
8737 (add-to-list kill-ring string)
8741 When it is false, the expression first constructs a new version of the
8742 kill ring by prepending @code{string} to the existing kill ring as a
8743 new element (that is what the @code{push} does). Then it executes a
8744 second @code{if} clause. This second @code{if} clause keeps the kill
8745 ring from growing too long.
8747 Let's look at these two expressions in order.
8749 The @code{push} line of the else-part sets the new value of the kill
8750 ring to what results from adding the string being killed to the old
8753 We can see how this works with an example.
8759 (setq example-list '("here is a clause" "another clause"))
8764 After evaluating this expression with @kbd{C-x C-e}, you can evaluate
8765 @code{example-list} and see what it returns:
8770 @result{} ("here is a clause" "another clause")
8776 Now, we can add a new element on to this list by evaluating the
8777 following expression:
8778 @findex push, @r{example}
8781 (push "a third clause" example-list)
8786 When we evaluate @code{example-list}, we find its value is:
8791 @result{} ("a third clause" "here is a clause" "another clause")
8796 Thus, the third clause is added to the list by @code{push}.
8799 Now for the second part of the @code{if} clause. This expression
8800 keeps the kill ring from growing too long. It looks like this:
8804 (if (> (length kill-ring) kill-ring-max)
8805 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil))
8809 The code checks whether the length of the kill ring is greater than
8810 the maximum permitted length. This is the value of
8811 @code{kill-ring-max} (which is 60, by default). If the length of the
8812 kill ring is too long, then this code sets the last element of the
8813 kill ring to @code{nil}. It does this by using two functions,
8814 @code{nthcdr} and @code{setcdr}.
8816 We looked at @code{setcdr} earlier (@pxref{setcdr, , @code{setcdr}}).
8817 It sets the @sc{cdr} of a list, just as @code{setcar} sets the
8818 @sc{car} of a list. In this case, however, @code{setcdr} will not be
8819 setting the @sc{cdr} of the whole kill ring; the @code{nthcdr}
8820 function is used to cause it to set the @sc{cdr} of the next to last
8821 element of the kill ring---this means that since the @sc{cdr} of the
8822 next to last element is the last element of the kill ring, it will set
8823 the last element of the kill ring.
8825 @findex nthcdr, @r{example}
8826 The @code{nthcdr} function works by repeatedly taking the @sc{cdr} of a
8827 list---it takes the @sc{cdr} of the @sc{cdr} of the @sc{cdr}
8828 @dots{} It does this @var{N} times and returns the results.
8829 (@xref{nthcdr, , @code{nthcdr}}.)
8831 @findex setcdr, @r{example}
8832 Thus, if we had a four element list that was supposed to be three
8833 elements long, we could set the @sc{cdr} of the next to last element
8834 to @code{nil}, and thereby shorten the list. (If you set the last
8835 element to some other value than @code{nil}, which you could do, then
8836 you would not have shortened the list. @xref{setcdr, ,
8839 You can see shortening by evaluating the following three expressions
8840 in turn. First set the value of @code{trees} to @code{(maple oak pine
8841 birch)}, then set the @sc{cdr} of its second @sc{cdr} to @code{nil}
8842 and then find the value of @code{trees}:
8846 (setq trees '(maple oak pine birch))
8847 @result{} (maple oak pine birch)
8851 (setcdr (nthcdr 2 trees) nil)
8855 @result{} (maple oak pine)
8860 (The value returned by the @code{setcdr} expression is @code{nil} since
8861 that is what the @sc{cdr} is set to.)
8863 To repeat, in @code{kill-new}, the @code{nthcdr} function takes the
8864 @sc{cdr} a number of times that is one less than the maximum permitted
8865 size of the kill ring and @code{setcdr} sets the @sc{cdr} of that
8866 element (which will be the rest of the elements in the kill ring) to
8867 @code{nil}. This prevents the kill ring from growing too long.
8870 The next to last expression in the @code{kill-new} function is
8873 (setq kill-ring-yank-pointer kill-ring)
8876 The @code{kill-ring-yank-pointer} is a global variable that is set to be
8877 the @code{kill-ring}.
8879 Even though the @code{kill-ring-yank-pointer} is called a
8880 @samp{pointer}, it is a variable just like the kill ring. However, the
8881 name has been chosen to help humans understand how the variable is used.
8884 Now, to return to an early expression in the body of the function:
8888 (if (fboundp 'menu-bar-update-yank-menu)
8889 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8894 It starts with an @code{if} expression
8896 In this case, the expression tests first to see whether
8897 @code{menu-bar-update-yank-menu} exists as a function, and if so,
8898 calls it. The @code{fboundp} function returns true if the symbol it
8899 is testing has a function definition that `is not void'. If the
8900 symbol's function definition were void, we would receive an error
8901 message, as we did when we created errors intentionally (@pxref{Making
8902 Errors, , Generate an Error Message}).
8905 The then-part contains an expression whose first element is the
8906 function @code{and}.
8909 The @code{and} special form evaluates each of its arguments until one
8910 of the arguments returns a value of @code{nil}, in which case the
8911 @code{and} expression returns @code{nil}; however, if none of the
8912 arguments returns a value of @code{nil}, the value resulting from
8913 evaluating the last argument is returned. (Since such a value is not
8914 @code{nil}, it is considered true in Emacs Lisp.) In other words, an
8915 @code{and} expression returns a true value only if all its arguments
8916 are true. (@xref{Second Buffer Related Review}.)
8918 The expression determines whether the second argument to
8919 @code{menu-bar-update-yank-menu} is true or not.
8921 ;; If we're supposed to be extending an existing string, and that
8922 ;; string really is at the front of the menu, then update it in place.
8925 @code{menu-bar-update-yank-menu} is one of the functions that make it
8926 possible to use the `Select and Paste' menu in the Edit item of a menu
8927 bar; using a mouse, you can look at the various pieces of text you
8928 have saved and select one piece to paste.
8930 The last expression in the @code{kill-new} function adds the newly
8931 copied string to whatever facility exists for copying and pasting
8932 among different programs running in a windowing system. In the X
8933 Windowing system, for example, the @code{x-select-text} function takes
8934 the string and stores it in memory operated by X@. You can paste the
8935 string in another program, such as an Xterm.
8938 The expression looks like this:
8942 (if interprogram-cut-function
8943 (funcall interprogram-cut-function string (not replace))))
8947 If an @code{interprogram-cut-function} exists, then Emacs executes
8948 @code{funcall}, which in turn calls its first argument as a function
8949 and passes the remaining arguments to it. (Incidentally, as far as I
8950 can see, this @code{if} expression could be replaced by an @code{and}
8951 expression similar to the one in the first part of the function.)
8953 We are not going to discuss windowing systems and other programs
8954 further, but merely note that this is a mechanism that enables GNU
8955 Emacs to work easily and well with other programs.
8957 This code for placing text in the kill ring, either concatenated with
8958 an existing element or as a new element, leads us to the code for
8959 bringing back text that has been cut out of the buffer---the yank
8960 commands. However, before discussing the yank commands, it is better
8961 to learn how lists are implemented in a computer. This will make
8962 clear such mysteries as the use of the term `pointer'. But before
8963 that, we will digress into C.
8966 @c is this true in Emacs 22? Does not seems to be
8968 (If the @w{@code{(< end beg))}}
8969 expression is true, @code{kill-append} prepends the string to the just
8970 previously clipped text. For a detailed discussion, see
8971 @ref{kill-append function, , The @code{kill-append} function}.)
8973 If you then yank back the text, i.e., `paste' it, you get both
8974 pieces of text at once. That way, if you delete two words in a row,
8975 and then yank them back, you get both words, in their proper order,
8976 with one yank. (The @w{@code{(< end beg))}} expression makes sure the
8979 On the other hand, if the previous command is not @code{kill-region},
8980 then the @code{kill-new} function is called, which adds the text to
8981 the kill ring as the latest item, and sets the
8982 @code{kill-ring-yank-pointer} variable to point to it.
8986 @c Evidently, changed for Emacs 22. The zap-to-char command does not
8987 @c use the delete-and-extract-region function
8989 2006 Oct 26, the Digression into C is now OK but should come after
8990 copy-region-as-kill and filter-buffer-substring
8994 copy-region-as-kill is short, 12 lines, and uses
8995 filter-buffer-substring, which is longer, 39 lines
8996 and has delete-and-extract-region in it.
8997 delete-and-extract-region is written in C.
8999 see Initializing a Variable with @code{defvar}
9002 @node Digression into C
9003 @section Digression into C
9004 @findex delete-and-extract-region
9005 @cindex C, a digression into
9006 @cindex Digression into C
9008 The @code{copy-region-as-kill} function (@pxref{copy-region-as-kill, ,
9009 @code{copy-region-as-kill}}) uses the @code{filter-buffer-substring}
9010 function, which in turn uses the @code{delete-and-extract-region}
9011 function. It removes the contents of a region and you cannot get them
9014 Unlike the other code discussed here, the
9015 @code{delete-and-extract-region} function is not written in Emacs
9016 Lisp; it is written in C and is one of the primitives of the GNU Emacs
9017 system. Since it is very simple, I will digress briefly from Lisp and
9020 @c GNU Emacs 24 in src/editfns.c
9021 @c the DEFUN for delete-and-extract-region
9024 Like many of the other Emacs primitives,
9025 @code{delete-and-extract-region} is written as an instance of a C
9026 macro, a macro being a template for code. The complete macro looks
9031 DEFUN ("delete-and-extract-region", Fdelete_and_extract_region,
9032 Sdelete_and_extract_region, 2, 2, 0,
9033 doc: /* Delete the text between START and END and return it. */)
9034 (Lisp_Object start, Lisp_Object end)
9036 validate_region (&start, &end);
9037 if (XINT (start) == XINT (end))
9038 return empty_unibyte_string;
9039 return del_range_1 (XINT (start), XINT (end), 1, 1);
9044 Without going into the details of the macro writing process, let me
9045 point out that this macro starts with the word @code{DEFUN}. The word
9046 @code{DEFUN} was chosen since the code serves the same purpose as
9047 @code{defun} does in Lisp. (The @code{DEFUN} C macro is defined in
9048 @file{emacs/src/lisp.h}.)
9050 The word @code{DEFUN} is followed by seven parts inside of
9055 The first part is the name given to the function in Lisp,
9056 @code{delete-and-extract-region}.
9059 The second part is the name of the function in C,
9060 @code{Fdelete_and_extract_region}. By convention, it starts with
9061 @samp{F}. Since C does not use hyphens in names, underscores are used
9065 The third part is the name for the C constant structure that records
9066 information on this function for internal use. It is the name of the
9067 function in C but begins with an @samp{S} instead of an @samp{F}.
9070 The fourth and fifth parts specify the minimum and maximum number of
9071 arguments the function can have. This function demands exactly 2
9075 The sixth part is nearly like the argument that follows the
9076 @code{interactive} declaration in a function written in Lisp: a letter
9077 followed, perhaps, by a prompt. The only difference from the Lisp is
9078 when the macro is called with no arguments. Then you write a @code{0}
9079 (which is a `null string'), as in this macro.
9081 If you were to specify arguments, you would place them between
9082 quotation marks. The C macro for @code{goto-char} includes
9083 @code{"NGoto char: "} in this position to indicate that the function
9084 expects a raw prefix, in this case, a numerical location in a buffer,
9085 and provides a prompt.
9088 The seventh part is a documentation string, just like the one for a
9089 function written in Emacs Lisp. This is written as a C comment. (When
9090 you build Emacs, the program @command{lib-src/make-docfile} extracts
9091 these comments and uses them to make the ``real'' documentation.)
9095 In a C macro, the formal parameters come next, with a statement of
9096 what kind of object they are, followed by what might be called the `body'
9097 of the macro. For @code{delete-and-extract-region} the `body'
9098 consists of the following four lines:
9102 validate_region (&start, &end);
9103 if (XINT (start) == XINT (end))
9104 return empty_unibyte_string;
9105 return del_range_1 (XINT (start), XINT (end), 1, 1);
9109 The @code{validate_region} function checks whether the values
9110 passed as the beginning and end of the region are the proper type and
9111 are within range. If the beginning and end positions are the same,
9112 then return an empty string.
9114 The @code{del_range_1} function actually deletes the text. It is a
9115 complex function we will not look into. It updates the buffer and
9116 does other things. However, it is worth looking at the two arguments
9117 passed to @code{del_range}. These are @w{@code{XINT (start)}} and
9118 @w{@code{XINT (end)}}.
9120 As far as the C language is concerned, @code{start} and @code{end} are
9121 two integers that mark the beginning and end of the region to be
9122 deleted@footnote{More precisely, and requiring more expert knowledge
9123 to understand, the two integers are of type `Lisp_Object', which can
9124 also be a C union instead of an integer type.}.
9126 In early versions of Emacs, these two numbers were thirty-two bits
9127 long, but the code is slowly being generalized to handle other
9128 lengths. Three of the available bits are used to specify the type of
9129 information; the remaining bits are used as `content'.
9131 @samp{XINT} is a C macro that extracts the relevant number from the
9132 longer collection of bits; the three other bits are discarded.
9135 The command in @code{delete-and-extract-region} looks like this:
9138 del_range_1 (XINT (start), XINT (end), 1, 1);
9142 It deletes the region between the beginning position, @code{start},
9143 and the ending position, @code{end}.
9145 From the point of view of the person writing Lisp, Emacs is all very
9146 simple; but hidden underneath is a great deal of complexity to make it
9150 @section Initializing a Variable with @code{defvar}
9152 @cindex Initializing a variable
9153 @cindex Variable initialization
9158 copy-region-as-kill is short, 12 lines, and uses
9159 filter-buffer-substring, which is longer, 39 lines
9160 and has delete-and-extract-region in it.
9161 delete-and-extract-region is written in C.
9163 see Initializing a Variable with @code{defvar}
9167 The @code{copy-region-as-kill} function is written in Emacs Lisp. Two
9168 functions within it, @code{kill-append} and @code{kill-new}, copy a
9169 region in a buffer and save it in a variable called the
9170 @code{kill-ring}. This section describes how the @code{kill-ring}
9171 variable is created and initialized using the @code{defvar} special
9174 (Again we note that the term @code{kill-ring} is a misnomer. The text
9175 that is clipped out of the buffer can be brought back; it is not a ring
9176 of corpses, but a ring of resurrectable text.)
9178 In Emacs Lisp, a variable such as the @code{kill-ring} is created and
9179 given an initial value by using the @code{defvar} special form. The
9180 name comes from ``define variable''.
9182 The @code{defvar} special form is similar to @code{setq} in that it sets
9183 the value of a variable. It is unlike @code{setq} in two ways: first,
9184 it only sets the value of the variable if the variable does not already
9185 have a value. If the variable already has a value, @code{defvar} does
9186 not override the existing value. Second, @code{defvar} has a
9187 documentation string.
9189 (There is a related macro, @code{defcustom}, designed for variables
9190 that people customize. It has more features than @code{defvar}.
9191 (@xref{defcustom, , Setting Variables with @code{defcustom}}.)
9194 * See variable current value::
9195 * defvar and asterisk::
9199 @node See variable current value
9200 @unnumberedsubsec Seeing the Current Value of a Variable
9203 You can see the current value of a variable, any variable, by using
9204 the @code{describe-variable} function, which is usually invoked by
9205 typing @kbd{C-h v}. If you type @kbd{C-h v} and then @code{kill-ring}
9206 (followed by @key{RET}) when prompted, you will see what is in your
9207 current kill ring---this may be quite a lot! Conversely, if you have
9208 been doing nothing this Emacs session except read this document, you
9209 may have nothing in it. Also, you will see the documentation for
9215 List of killed text sequences.
9216 Since the kill ring is supposed to interact nicely with cut-and-paste
9217 facilities offered by window systems, use of this variable should
9220 interact nicely with `interprogram-cut-function' and
9221 `interprogram-paste-function'. The functions `kill-new',
9222 `kill-append', and `current-kill' are supposed to implement this
9223 interaction; you may want to use them instead of manipulating the kill
9229 The kill ring is defined by a @code{defvar} in the following way:
9233 (defvar kill-ring nil
9234 "List of killed text sequences.
9240 In this variable definition, the variable is given an initial value of
9241 @code{nil}, which makes sense, since if you have saved nothing, you want
9242 nothing back if you give a @code{yank} command. The documentation
9243 string is written just like the documentation string of a @code{defun}.
9244 As with the documentation string of the @code{defun}, the first line of
9245 the documentation should be a complete sentence, since some commands,
9246 like @code{apropos}, print only the first line of documentation.
9247 Succeeding lines should not be indented; otherwise they look odd when
9248 you use @kbd{C-h v} (@code{describe-variable}).
9250 @node defvar and asterisk
9251 @subsection @code{defvar} and an asterisk
9252 @findex defvar @r{for a user customizable variable}
9253 @findex defvar @r{with an asterisk}
9255 In the past, Emacs used the @code{defvar} special form both for
9256 internal variables that you would not expect a user to change and for
9257 variables that you do expect a user to change. Although you can still
9258 use @code{defvar} for user customizable variables, please use
9259 @code{defcustom} instead, since it provides a path into
9260 the Customization commands. (@xref{defcustom, , Specifying Variables
9261 using @code{defcustom}}.)
9263 When you specified a variable using the @code{defvar} special form,
9264 you could distinguish a variable that a user might want to change from
9265 others by typing an asterisk, @samp{*}, in the first column of its
9266 documentation string. For example:
9270 (defvar shell-command-default-error-buffer nil
9271 "*Buffer name for `shell-command' @dots{} error output.
9276 @findex set-variable
9278 You could (and still can) use the @code{set-variable} command to
9279 change the value of @code{shell-command-default-error-buffer}
9280 temporarily. However, options set using @code{set-variable} are set
9281 only for the duration of your editing session. The new values are not
9282 saved between sessions. Each time Emacs starts, it reads the original
9283 value, unless you change the value within your @file{.emacs} file,
9284 either by setting it manually or by using @code{customize}.
9285 @xref{Emacs Initialization, , Your @file{.emacs} File}.
9287 For me, the major use of the @code{set-variable} command is to suggest
9288 variables that I might want to set in my @file{.emacs} file. There
9289 are now more than 700 such variables, far too many to remember
9290 readily. Fortunately, you can press @key{TAB} after calling the
9291 @code{M-x set-variable} command to see the list of variables.
9292 (@xref{Examining, , Examining and Setting Variables, emacs,
9293 The GNU Emacs Manual}.)
9296 @node cons & search-fwd Review
9299 Here is a brief summary of some recently introduced functions.
9304 @code{car} returns the first element of a list; @code{cdr} returns the
9305 second and subsequent elements of a list.
9312 (car '(1 2 3 4 5 6 7))
9314 (cdr '(1 2 3 4 5 6 7))
9315 @result{} (2 3 4 5 6 7)
9320 @code{cons} constructs a list by prepending its first argument to its
9334 @code{funcall} evaluates its first argument as a function. It passes
9335 its remaining arguments to its first argument.
9338 Return the result of taking @sc{cdr} `n' times on a list.
9346 The `rest of the rest', as it were.
9353 (nthcdr 3 '(1 2 3 4 5 6 7))
9360 @code{setcar} changes the first element of a list; @code{setcdr}
9361 changes the second and subsequent elements of a list.
9368 (setq triple '(1 2 3))
9375 (setcdr triple '("foo" "bar"))
9378 @result{} (37 "foo" "bar")
9383 Evaluate each argument in sequence and then return the value of the
9396 @item save-restriction
9397 Record whatever narrowing is in effect in the current buffer, if any,
9398 and restore that narrowing after evaluating the arguments.
9400 @item search-forward
9401 Search for a string, and if the string is found, move point. With a
9402 regular expression, use the similar @code{re-search-forward}.
9403 (@xref{Regexp Search, , Regular Expression Searches}, for an
9404 explanation of regular expression patterns and searches.)
9408 @code{search-forward} and @code{re-search-forward} take four
9413 The string or regular expression to search for.
9416 Optionally, the limit of the search.
9419 Optionally, what to do if the search fails, return @code{nil} or an
9423 Optionally, how many times to repeat the search; if negative, the
9424 search goes backwards.
9428 @itemx delete-and-extract-region
9429 @itemx copy-region-as-kill
9431 @code{kill-region} cuts the text between point and mark from the
9432 buffer and stores that text in the kill ring, so you can get it back
9435 @code{copy-region-as-kill} copies the text between point and mark into
9436 the kill ring, from which you can get it by yanking. The function
9437 does not cut or remove the text from the buffer.
9440 @code{delete-and-extract-region} removes the text between point and
9441 mark from the buffer and throws it away. You cannot get it back.
9442 (This is not an interactive command.)
9445 @node search Exercises
9446 @section Searching Exercises
9450 Write an interactive function that searches for a string. If the
9451 search finds the string, leave point after it and display a message
9452 that says ``Found!''. (Do not use @code{search-forward} for the name
9453 of this function; if you do, you will overwrite the existing version of
9454 @code{search-forward} that comes with Emacs. Use a name such as
9455 @code{test-search} instead.)
9458 Write a function that prints the third element of the kill ring in the
9459 echo area, if any; if the kill ring does not contain a third element,
9460 print an appropriate message.
9463 @node List Implementation
9464 @chapter How Lists are Implemented
9465 @cindex Lists in a computer
9467 In Lisp, atoms are recorded in a straightforward fashion; if the
9468 implementation is not straightforward in practice, it is, nonetheless,
9469 straightforward in theory. The atom @samp{rose}, for example, is
9470 recorded as the four contiguous letters @samp{r}, @samp{o}, @samp{s},
9471 @samp{e}. A list, on the other hand, is kept differently. The mechanism
9472 is equally simple, but it takes a moment to get used to the idea. A
9473 list is kept using a series of pairs of pointers. In the series, the
9474 first pointer in each pair points to an atom or to another list, and the
9475 second pointer in each pair points to the next pair, or to the symbol
9476 @code{nil}, which marks the end of the list.
9478 A pointer itself is quite simply the electronic address of what is
9479 pointed to. Hence, a list is kept as a series of electronic addresses.
9482 * Lists diagrammed::
9483 * Symbols as Chest:: Exploring a powerful metaphor.
9488 @node Lists diagrammed
9489 @unnumberedsec Lists diagrammed
9492 For example, the list @code{(rose violet buttercup)} has three elements,
9493 @samp{rose}, @samp{violet}, and @samp{buttercup}. In the computer, the
9494 electronic address of @samp{rose} is recorded in a segment of computer
9495 memory along with the address that gives the electronic address of where
9496 the atom @samp{violet} is located; and that address (the one that tells
9497 where @samp{violet} is located) is kept along with an address that tells
9498 where the address for the atom @samp{buttercup} is located.
9501 This sounds more complicated than it is and is easier seen in a diagram:
9503 @c clear print-postscript-figures
9504 @c !!! cons-cell-diagram #1
9508 ___ ___ ___ ___ ___ ___
9509 |___|___|--> |___|___|--> |___|___|--> nil
9512 --> rose --> violet --> buttercup
9516 @ifset print-postscript-figures
9519 @center @image{cons-1}
9523 @ifclear print-postscript-figures
9527 ___ ___ ___ ___ ___ ___
9528 |___|___|--> |___|___|--> |___|___|--> nil
9531 --> rose --> violet --> buttercup
9538 In the diagram, each box represents a word of computer memory that
9539 holds a Lisp object, usually in the form of a memory address. The boxes,
9540 i.e., the addresses, are in pairs. Each arrow points to what the address
9541 is the address of, either an atom or another pair of addresses. The
9542 first box is the electronic address of @samp{rose} and the arrow points
9543 to @samp{rose}; the second box is the address of the next pair of boxes,
9544 the first part of which is the address of @samp{violet} and the second
9545 part of which is the address of the next pair. The very last box
9546 points to the symbol @code{nil}, which marks the end of the list.
9549 When a variable is set to a list with a function such as @code{setq},
9550 it stores the address of the first box in the variable. Thus,
9551 evaluation of the expression
9554 (setq bouquet '(rose violet buttercup))
9559 creates a situation like this:
9561 @c cons-cell-diagram #2
9567 | ___ ___ ___ ___ ___ ___
9568 --> |___|___|--> |___|___|--> |___|___|--> nil
9571 --> rose --> violet --> buttercup
9575 @ifset print-postscript-figures
9578 @center @image{cons-2}
9582 @ifclear print-postscript-figures
9588 | ___ ___ ___ ___ ___ ___
9589 --> |___|___|--> |___|___|--> |___|___|--> nil
9592 --> rose --> violet --> buttercup
9599 In this example, the symbol @code{bouquet} holds the address of the first
9603 This same list can be illustrated in a different sort of box notation
9606 @c cons-cell-diagram #2a
9612 | -------------- --------------- ----------------
9613 | | car | cdr | | car | cdr | | car | cdr |
9614 -->| rose | o------->| violet | o------->| butter- | nil |
9615 | | | | | | | cup | |
9616 -------------- --------------- ----------------
9620 @ifset print-postscript-figures
9623 @center @image{cons-2a}
9627 @ifclear print-postscript-figures
9633 | -------------- --------------- ----------------
9634 | | car | cdr | | car | cdr | | car | cdr |
9635 -->| rose | o------->| violet | o------->| butter- | nil |
9636 | | | | | | | cup | |
9637 -------------- --------------- ----------------
9643 (Symbols consist of more than pairs of addresses, but the structure of
9644 a symbol is made up of addresses. Indeed, the symbol @code{bouquet}
9645 consists of a group of address-boxes, one of which is the address of
9646 the printed word @samp{bouquet}, a second of which is the address of a
9647 function definition attached to the symbol, if any, a third of which
9648 is the address of the first pair of address-boxes for the list
9649 @code{(rose violet buttercup)}, and so on. Here we are showing that
9650 the symbol's third address-box points to the first pair of
9651 address-boxes for the list.)
9653 If a symbol is set to the @sc{cdr} of a list, the list itself is not
9654 changed; the symbol simply has an address further down the list. (In
9655 the jargon, @sc{car} and @sc{cdr} are `non-destructive'.) Thus,
9656 evaluation of the following expression
9659 (setq flowers (cdr bouquet))
9666 @c cons-cell-diagram #3
9673 | ___ ___ | ___ ___ ___ ___
9674 --> | | | --> | | | | | |
9675 |___|___|----> |___|___|--> |___|___|--> nil
9678 --> rose --> violet --> buttercup
9683 @ifset print-postscript-figures
9686 @center @image{cons-3}
9690 @ifclear print-postscript-figures
9697 | ___ ___ | ___ ___ ___ ___
9698 --> | | | --> | | | | | |
9699 |___|___|----> |___|___|--> |___|___|--> nil
9702 --> rose --> violet --> buttercup
9710 The value of @code{flowers} is @code{(violet buttercup)}, which is
9711 to say, the symbol @code{flowers} holds the address of the pair of
9712 address-boxes, the first of which holds the address of @code{violet},
9713 and the second of which holds the address of @code{buttercup}.
9715 A pair of address-boxes is called a @dfn{cons cell} or @dfn{dotted
9716 pair}. @xref{Cons Cell Type, , Cons Cell and List Types, elisp, The GNU Emacs Lisp
9717 Reference Manual}, and @ref{Dotted Pair Notation, , Dotted Pair
9718 Notation, elisp, The GNU Emacs Lisp Reference Manual}, for more
9719 information about cons cells and dotted pairs.
9722 The function @code{cons} adds a new pair of addresses to the front of
9723 a series of addresses like that shown above. For example, evaluating
9727 (setq bouquet (cons 'lily bouquet))
9734 @c cons-cell-diagram #4
9741 | ___ ___ ___ ___ | ___ ___ ___ ___
9742 --> | | | | | | --> | | | | | |
9743 |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil
9746 --> lily --> rose --> violet --> buttercup
9751 @ifset print-postscript-figures
9754 @center @image{cons-4}
9758 @ifclear print-postscript-figures
9765 | ___ ___ ___ ___ | ___ ___ ___ ___
9766 --> | | | | | | --> | | | | | |
9767 |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil
9770 --> lily --> rose --> violet --> buttercup
9779 However, this does not change the value of the symbol
9780 @code{flowers}, as you can see by evaluating the following,
9783 (eq (cdr (cdr bouquet)) flowers)
9787 which returns @code{t} for true.
9789 Until it is reset, @code{flowers} still has the value
9790 @code{(violet buttercup)}; that is, it has the address of the cons
9791 cell whose first address is of @code{violet}. Also, this does not
9792 alter any of the pre-existing cons cells; they are all still there.
9794 Thus, in Lisp, to get the @sc{cdr} of a list, you just get the address
9795 of the next cons cell in the series; to get the @sc{car} of a list,
9796 you get the address of the first element of the list; to @code{cons} a
9797 new element on a list, you add a new cons cell to the front of the list.
9798 That is all there is to it! The underlying structure of Lisp is
9801 And what does the last address in a series of cons cells refer to? It
9802 is the address of the empty list, of @code{nil}.
9804 In summary, when a Lisp variable is set to a value, it is provided with
9805 the address of the list to which the variable refers.
9807 @node Symbols as Chest
9808 @section Symbols as a Chest of Drawers
9809 @cindex Symbols as a Chest of Drawers
9810 @cindex Chest of Drawers, metaphor for a symbol
9811 @cindex Drawers, Chest of, metaphor for a symbol
9813 In an earlier section, I suggested that you might imagine a symbol as
9814 being a chest of drawers. The function definition is put in one
9815 drawer, the value in another, and so on. What is put in the drawer
9816 holding the value can be changed without affecting the contents of the
9817 drawer holding the function definition, and vice-verse.
9819 Actually, what is put in each drawer is the address of the value or
9820 function definition. It is as if you found an old chest in the attic,
9821 and in one of its drawers you found a map giving you directions to
9822 where the buried treasure lies.
9824 (In addition to its name, symbol definition, and variable value, a
9825 symbol has a `drawer' for a @dfn{property list} which can be used to
9826 record other information. Property lists are not discussed here; see
9827 @ref{Property Lists, , Property Lists, elisp, The GNU Emacs Lisp
9831 Here is a fanciful representation:
9833 @c chest-of-drawers diagram
9838 Chest of Drawers Contents of Drawers
9842 ---------------------
9843 | directions to | [map to]
9844 | symbol name | bouquet
9846 +---------------------+
9848 | symbol definition | [none]
9850 +---------------------+
9851 | directions to | [map to]
9852 | variable value | (rose violet buttercup)
9854 +---------------------+
9856 | property list | [not described here]
9858 +---------------------+
9864 @ifset print-postscript-figures
9867 @center @image{drawers}
9871 @ifclear print-postscript-figures
9876 Chest of Drawers Contents of Drawers
9880 ---------------------
9881 | directions to | [map to]
9882 | symbol name | bouquet
9884 +---------------------+
9886 | symbol definition | [none]
9888 +---------------------+
9889 | directions to | [map to]
9890 | variable value | (rose violet buttercup)
9892 +---------------------+
9894 | property list | [not described here]
9896 +---------------------+
9907 Set @code{flowers} to @code{violet} and @code{buttercup}. Cons two
9908 more flowers on to this list and set this new list to
9909 @code{more-flowers}. Set the @sc{car} of @code{flowers} to a fish.
9910 What does the @code{more-flowers} list now contain?
9913 @chapter Yanking Text Back
9915 @cindex Text retrieval
9916 @cindex Retrieving text
9917 @cindex Pasting text
9919 Whenever you cut text out of a buffer with a `kill' command in GNU Emacs,
9920 you can bring it back with a `yank' command. The text that is cut out of
9921 the buffer is put in the kill ring and the yank commands insert the
9922 appropriate contents of the kill ring back into a buffer (not necessarily
9923 the original buffer).
9925 A simple @kbd{C-y} (@code{yank}) command inserts the first item from
9926 the kill ring into the current buffer. If the @kbd{C-y} command is
9927 followed immediately by @kbd{M-y}, the first element is replaced by
9928 the second element. Successive @kbd{M-y} commands replace the second
9929 element with the third, fourth, or fifth element, and so on. When the
9930 last element in the kill ring is reached, it is replaced by the first
9931 element and the cycle is repeated. (Thus the kill ring is called a
9932 `ring' rather than just a `list'. However, the actual data structure
9933 that holds the text is a list.
9934 @xref{Kill Ring, , Handling the Kill Ring}, for the details of how the
9935 list is handled as a ring.)
9938 * Kill Ring Overview::
9939 * kill-ring-yank-pointer:: The kill ring is a list.
9940 * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable.
9943 @node Kill Ring Overview
9944 @section Kill Ring Overview
9945 @cindex Kill ring overview
9947 The kill ring is a list of textual strings. This is what it looks like:
9950 ("some text" "a different piece of text" "yet more text")
9953 If this were the contents of my kill ring and I pressed @kbd{C-y}, the
9954 string of characters saying @samp{some text} would be inserted in this
9955 buffer where my cursor is located.
9957 The @code{yank} command is also used for duplicating text by copying it.
9958 The copied text is not cut from the buffer, but a copy of it is put on the
9959 kill ring and is inserted by yanking it back.
9961 Three functions are used for bringing text back from the kill ring:
9962 @code{yank}, which is usually bound to @kbd{C-y}; @code{yank-pop},
9963 which is usually bound to @kbd{M-y}; and @code{rotate-yank-pointer},
9964 which is used by the two other functions.
9966 These functions refer to the kill ring through a variable called the
9967 @code{kill-ring-yank-pointer}. Indeed, the insertion code for both the
9968 @code{yank} and @code{yank-pop} functions is:
9971 (insert (car kill-ring-yank-pointer))
9975 (Well, no more. In GNU Emacs 22, the function has been replaced by
9976 @code{insert-for-yank} which calls @code{insert-for-yank-1}
9977 repetitively for each @code{yank-handler} segment. In turn,
9978 @code{insert-for-yank-1} strips text properties from the inserted text
9979 according to @code{yank-excluded-properties}. Otherwise, it is just
9980 like @code{insert}. We will stick with plain @code{insert} since it
9981 is easier to understand.)
9983 To begin to understand how @code{yank} and @code{yank-pop} work, it is
9984 first necessary to look at the @code{kill-ring-yank-pointer} variable.
9986 @node kill-ring-yank-pointer
9987 @section The @code{kill-ring-yank-pointer} Variable
9989 @code{kill-ring-yank-pointer} is a variable, just as @code{kill-ring} is
9990 a variable. It points to something by being bound to the value of what
9991 it points to, like any other Lisp variable.
9994 Thus, if the value of the kill ring is:
9997 ("some text" "a different piece of text" "yet more text")
10002 and the @code{kill-ring-yank-pointer} points to the second clause, the
10003 value of @code{kill-ring-yank-pointer} is:
10006 ("a different piece of text" "yet more text")
10009 As explained in the previous chapter (@pxref{List Implementation}), the
10010 computer does not keep two different copies of the text being pointed to
10011 by both the @code{kill-ring} and the @code{kill-ring-yank-pointer}. The
10012 words ``a different piece of text'' and ``yet more text'' are not
10013 duplicated. Instead, the two Lisp variables point to the same pieces of
10014 text. Here is a diagram:
10016 @c cons-cell-diagram #5
10020 kill-ring kill-ring-yank-pointer
10022 | ___ ___ | ___ ___ ___ ___
10023 ---> | | | --> | | | | | |
10024 |___|___|----> |___|___|--> |___|___|--> nil
10027 | | --> "yet more text"
10029 | --> "a different piece of text"
10036 @ifset print-postscript-figures
10039 @center @image{cons-5}
10043 @ifclear print-postscript-figures
10047 kill-ring kill-ring-yank-pointer
10049 | ___ ___ | ___ ___ ___ ___
10050 ---> | | | --> | | | | | |
10051 |___|___|----> |___|___|--> |___|___|--> nil
10054 | | --> "yet more text"
10056 | --> "a different piece of text
10065 Both the variable @code{kill-ring} and the variable
10066 @code{kill-ring-yank-pointer} are pointers. But the kill ring itself is
10067 usually described as if it were actually what it is composed of. The
10068 @code{kill-ring} is spoken of as if it were the list rather than that it
10069 points to the list. Conversely, the @code{kill-ring-yank-pointer} is
10070 spoken of as pointing to a list.
10072 These two ways of talking about the same thing sound confusing at first but
10073 make sense on reflection. The kill ring is generally thought of as the
10074 complete structure of data that holds the information of what has recently
10075 been cut out of the Emacs buffers. The @code{kill-ring-yank-pointer}
10076 on the other hand, serves to indicate---that is, to `point to'---that part
10077 of the kill ring of which the first element (the @sc{car}) will be
10081 In GNU Emacs 22, the @code{kill-new} function calls
10083 @code{(setq kill-ring-yank-pointer kill-ring)}
10085 (defun rotate-yank-pointer (arg)
10086 "Rotate the yanking point in the kill ring.
10087 With argument, rotate that many kills forward (or backward, if negative)."
10089 (current-kill arg))
10091 (defun current-kill (n &optional do-not-move)
10092 "Rotate the yanking point by N places, and then return that kill.
10093 If N is zero, `interprogram-paste-function' is set, and calling it
10094 returns a string, then that string is added to the front of the
10095 kill ring and returned as the latest kill.
10096 If optional arg DO-NOT-MOVE is non-nil, then don't actually move the
10097 yanking point; just return the Nth kill forward."
10098 (let ((interprogram-paste (and (= n 0)
10099 interprogram-paste-function
10100 (funcall interprogram-paste-function))))
10101 (if interprogram-paste
10103 ;; Disable the interprogram cut function when we add the new
10104 ;; text to the kill ring, so Emacs doesn't try to own the
10105 ;; selection, with identical text.
10106 (let ((interprogram-cut-function nil))
10107 (kill-new interprogram-paste))
10108 interprogram-paste)
10109 (or kill-ring (error "Kill ring is empty"))
10110 (let ((ARGth-kill-element
10111 (nthcdr (mod (- n (length kill-ring-yank-pointer))
10112 (length kill-ring))
10115 (setq kill-ring-yank-pointer ARGth-kill-element))
10116 (car ARGth-kill-element)))))
10121 @node yank nthcdr Exercises
10122 @section Exercises with @code{yank} and @code{nthcdr}
10126 Using @kbd{C-h v} (@code{describe-variable}), look at the value of
10127 your kill ring. Add several items to your kill ring; look at its
10128 value again. Using @kbd{M-y} (@code{yank-pop)}, move all the way
10129 around the kill ring. How many items were in your kill ring? Find
10130 the value of @code{kill-ring-max}. Was your kill ring full, or could
10131 you have kept more blocks of text within it?
10134 Using @code{nthcdr} and @code{car}, construct a series of expressions
10135 to return the first, second, third, and fourth elements of a list.
10138 @node Loops & Recursion
10139 @chapter Loops and Recursion
10140 @cindex Loops and recursion
10141 @cindex Recursion and loops
10142 @cindex Repetition (loops)
10144 Emacs Lisp has two primary ways to cause an expression, or a series of
10145 expressions, to be evaluated repeatedly: one uses a @code{while}
10146 loop, and the other uses @dfn{recursion}.
10148 Repetition can be very valuable. For example, to move forward four
10149 sentences, you need only write a program that will move forward one
10150 sentence and then repeat the process four times. Since a computer does
10151 not get bored or tired, such repetitive action does not have the
10152 deleterious effects that excessive or the wrong kinds of repetition can
10155 People mostly write Emacs Lisp functions using @code{while} loops and
10156 their kin; but you can use recursion, which provides a very powerful
10157 way to think about and then to solve problems@footnote{You can write
10158 recursive functions to be frugal or wasteful of mental or computer
10159 resources; as it happens, methods that people find easy---that are
10160 frugal of `mental resources'---sometimes use considerable computer
10161 resources. Emacs was designed to run on machines that we now consider
10162 limited and its default settings are conservative. You may want to
10163 increase the values of @code{max-specpdl-size} and
10164 @code{max-lisp-eval-depth}. In my @file{.emacs} file, I set them to
10165 15 and 30 times their default value.}.
10168 * while:: Causing a stretch of code to repeat.
10170 * Recursion:: Causing a function to call itself.
10171 * Looping exercise::
10175 @section @code{while}
10179 The @code{while} special form tests whether the value returned by
10180 evaluating its first argument is true or false. This is similar to what
10181 the Lisp interpreter does with an @code{if}; what the interpreter does
10182 next, however, is different.
10184 In a @code{while} expression, if the value returned by evaluating the
10185 first argument is false, the Lisp interpreter skips the rest of the
10186 expression (the @dfn{body} of the expression) and does not evaluate it.
10187 However, if the value is true, the Lisp interpreter evaluates the body
10188 of the expression and then again tests whether the first argument to
10189 @code{while} is true or false. If the value returned by evaluating the
10190 first argument is again true, the Lisp interpreter again evaluates the
10191 body of the expression.
10194 The template for a @code{while} expression looks like this:
10198 (while @var{true-or-false-test}
10204 * Looping with while:: Repeat so long as test returns true.
10205 * Loop Example:: A @code{while} loop that uses a list.
10206 * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}.
10207 * Incrementing Loop:: A loop with an incrementing counter.
10208 * Incrementing Loop Details::
10209 * Decrementing Loop:: A loop with a decrementing counter.
10213 @node Looping with while
10214 @unnumberedsubsec Looping with @code{while}
10217 So long as the true-or-false-test of the @code{while} expression
10218 returns a true value when it is evaluated, the body is repeatedly
10219 evaluated. This process is called a loop since the Lisp interpreter
10220 repeats the same thing again and again, like an airplane doing a loop.
10221 When the result of evaluating the true-or-false-test is false, the
10222 Lisp interpreter does not evaluate the rest of the @code{while}
10223 expression and `exits the loop'.
10225 Clearly, if the value returned by evaluating the first argument to
10226 @code{while} is always true, the body following will be evaluated
10227 again and again @dots{} and again @dots{} forever. Conversely, if the
10228 value returned is never true, the expressions in the body will never
10229 be evaluated. The craft of writing a @code{while} loop consists of
10230 choosing a mechanism such that the true-or-false-test returns true
10231 just the number of times that you want the subsequent expressions to
10232 be evaluated, and then have the test return false.
10234 The value returned by evaluating a @code{while} is the value of the
10235 true-or-false-test. An interesting consequence of this is that a
10236 @code{while} loop that evaluates without error will return @code{nil}
10237 or false regardless of whether it has looped 1 or 100 times or none at
10238 all. A @code{while} expression that evaluates successfully never
10239 returns a true value! What this means is that @code{while} is always
10240 evaluated for its side effects, which is to say, the consequences of
10241 evaluating the expressions within the body of the @code{while} loop.
10242 This makes sense. It is not the mere act of looping that is desired,
10243 but the consequences of what happens when the expressions in the loop
10244 are repeatedly evaluated.
10247 @subsection A @code{while} Loop and a List
10249 A common way to control a @code{while} loop is to test whether a list
10250 has any elements. If it does, the loop is repeated; but if it does not,
10251 the repetition is ended. Since this is an important technique, we will
10252 create a short example to illustrate it.
10254 A simple way to test whether a list has elements is to evaluate the
10255 list: if it has no elements, it is an empty list and will return the
10256 empty list, @code{()}, which is a synonym for @code{nil} or false. On
10257 the other hand, a list with elements will return those elements when it
10258 is evaluated. Since Emacs Lisp considers as true any value that is not
10259 @code{nil}, a list that returns elements will test true in a
10263 For example, you can set the variable @code{empty-list} to @code{nil} by
10264 evaluating the following @code{setq} expression:
10267 (setq empty-list ())
10271 After evaluating the @code{setq} expression, you can evaluate the
10272 variable @code{empty-list} in the usual way, by placing the cursor after
10273 the symbol and typing @kbd{C-x C-e}; @code{nil} will appear in your
10280 On the other hand, if you set a variable to be a list with elements, the
10281 list will appear when you evaluate the variable, as you can see by
10282 evaluating the following two expressions:
10286 (setq animals '(gazelle giraffe lion tiger))
10292 Thus, to create a @code{while} loop that tests whether there are any
10293 items in the list @code{animals}, the first part of the loop will be
10304 When the @code{while} tests its first argument, the variable
10305 @code{animals} is evaluated. It returns a list. So long as the list
10306 has elements, the @code{while} considers the results of the test to be
10307 true; but when the list is empty, it considers the results of the test
10310 To prevent the @code{while} loop from running forever, some mechanism
10311 needs to be provided to empty the list eventually. An oft-used
10312 technique is to have one of the subsequent forms in the @code{while}
10313 expression set the value of the list to be the @sc{cdr} of the list.
10314 Each time the @code{cdr} function is evaluated, the list will be made
10315 shorter, until eventually only the empty list will be left. At this
10316 point, the test of the @code{while} loop will return false, and the
10317 arguments to the @code{while} will no longer be evaluated.
10319 For example, the list of animals bound to the variable @code{animals}
10320 can be set to be the @sc{cdr} of the original list with the
10321 following expression:
10324 (setq animals (cdr animals))
10328 If you have evaluated the previous expressions and then evaluate this
10329 expression, you will see @code{(giraffe lion tiger)} appear in the echo
10330 area. If you evaluate the expression again, @code{(lion tiger)} will
10331 appear in the echo area. If you evaluate it again and yet again,
10332 @code{(tiger)} appears and then the empty list, shown by @code{nil}.
10334 A template for a @code{while} loop that uses the @code{cdr} function
10335 repeatedly to cause the true-or-false-test eventually to test false
10340 (while @var{test-whether-list-is-empty}
10342 @var{set-list-to-cdr-of-list})
10346 This test and use of @code{cdr} can be put together in a function that
10347 goes through a list and prints each element of the list on a line of its
10350 @node print-elements-of-list
10351 @subsection An Example: @code{print-elements-of-list}
10352 @findex print-elements-of-list
10354 The @code{print-elements-of-list} function illustrates a @code{while}
10357 @cindex @file{*scratch*} buffer
10358 The function requires several lines for its output. If you are
10359 reading this in a recent instance of GNU Emacs,
10360 @c GNU Emacs 21, GNU Emacs 22, or a later version,
10361 you can evaluate the following expression inside of Info, as usual.
10363 If you are using an earlier version of Emacs, you need to copy the
10364 necessary expressions to your @file{*scratch*} buffer and evaluate
10365 them there. This is because the echo area had only one line in the
10368 You can copy the expressions by marking the beginning of the region
10369 with @kbd{C-@key{SPC}} (@code{set-mark-command}), moving the cursor to
10370 the end of the region and then copying the region using @kbd{M-w}
10371 (@code{kill-ring-save}, which calls @code{copy-region-as-kill} and
10372 then provides visual feedback). In the @file{*scratch*}
10373 buffer, you can yank the expressions back by typing @kbd{C-y}
10376 After you have copied the expressions to the @file{*scratch*} buffer,
10377 evaluate each expression in turn. Be sure to evaluate the last
10378 expression, @code{(print-elements-of-list animals)}, by typing
10379 @kbd{C-u C-x C-e}, that is, by giving an argument to
10380 @code{eval-last-sexp}. This will cause the result of the evaluation
10381 to be printed in the @file{*scratch*} buffer instead of being printed
10382 in the echo area. (Otherwise you will see something like this in your
10383 echo area: @code{^Jgazelle^J^Jgiraffe^J^Jlion^J^Jtiger^Jnil}, in which
10384 each @samp{^J} stands for a `newline'.)
10387 In a recent instance of GNU Emacs, you can evaluate these expressions
10388 directly in the Info buffer, and the echo area will grow to show the
10393 (setq animals '(gazelle giraffe lion tiger))
10395 (defun print-elements-of-list (list)
10396 "Print each element of LIST on a line of its own."
10399 (setq list (cdr list))))
10401 (print-elements-of-list animals)
10407 When you evaluate the three expressions in sequence, you will see
10423 Each element of the list is printed on a line of its own (that is what
10424 the function @code{print} does) and then the value returned by the
10425 function is printed. Since the last expression in the function is the
10426 @code{while} loop, and since @code{while} loops always return
10427 @code{nil}, a @code{nil} is printed after the last element of the list.
10429 @node Incrementing Loop
10430 @subsection A Loop with an Incrementing Counter
10432 A loop is not useful unless it stops when it ought. Besides
10433 controlling a loop with a list, a common way of stopping a loop is to
10434 write the first argument as a test that returns false when the correct
10435 number of repetitions are complete. This means that the loop must
10436 have a counter---an expression that counts how many times the loop
10440 @node Incrementing Loop Details
10441 @unnumberedsubsec Details of an Incrementing Loop
10444 The test for a loop with an incrementing counter can be an expression
10445 such as @code{(< count desired-number)} which returns @code{t} for
10446 true if the value of @code{count} is less than the
10447 @code{desired-number} of repetitions and @code{nil} for false if the
10448 value of @code{count} is equal to or is greater than the
10449 @code{desired-number}. The expression that increments the count can
10450 be a simple @code{setq} such as @code{(setq count (1+ count))}, where
10451 @code{1+} is a built-in function in Emacs Lisp that adds 1 to its
10452 argument. (The expression @w{@code{(1+ count)}} has the same result
10453 as @w{@code{(+ count 1)}}, but is easier for a human to read.)
10456 The template for a @code{while} loop controlled by an incrementing
10457 counter looks like this:
10461 @var{set-count-to-initial-value}
10462 (while (< count desired-number) ; @r{true-or-false-test}
10464 (setq count (1+ count))) ; @r{incrementer}
10469 Note that you need to set the initial value of @code{count}; usually it
10473 * Incrementing Example:: Counting pebbles in a triangle.
10474 * Inc Example parts:: The parts of the function definition.
10475 * Inc Example altogether:: Putting the function definition together.
10478 @node Incrementing Example
10479 @unnumberedsubsubsec Example with incrementing counter
10481 Suppose you are playing on the beach and decide to make a triangle of
10482 pebbles, putting one pebble in the first row, two in the second row,
10483 three in the third row and so on, like this:
10501 @bullet{} @bullet{}
10502 @bullet{} @bullet{} @bullet{}
10503 @bullet{} @bullet{} @bullet{} @bullet{}
10510 (About 2500 years ago, Pythagoras and others developed the beginnings of
10511 number theory by considering questions such as this.)
10513 Suppose you want to know how many pebbles you will need to make a
10514 triangle with 7 rows?
10516 Clearly, what you need to do is add up the numbers from 1 to 7. There
10517 are two ways to do this; start with the smallest number, one, and add up
10518 the list in sequence, 1, 2, 3, 4 and so on; or start with the largest
10519 number and add the list going down: 7, 6, 5, 4 and so on. Because both
10520 mechanisms illustrate common ways of writing @code{while} loops, we will
10521 create two examples, one counting up and the other counting down. In
10522 this first example, we will start with 1 and add 2, 3, 4 and so on.
10524 If you are just adding up a short list of numbers, the easiest way to do
10525 it is to add up all the numbers at once. However, if you do not know
10526 ahead of time how many numbers your list will have, or if you want to be
10527 prepared for a very long list, then you need to design your addition so
10528 that what you do is repeat a simple process many times instead of doing
10529 a more complex process once.
10531 For example, instead of adding up all the pebbles all at once, what you
10532 can do is add the number of pebbles in the first row, 1, to the number
10533 in the second row, 2, and then add the total of those two rows to the
10534 third row, 3. Then you can add the number in the fourth row, 4, to the
10535 total of the first three rows; and so on.
10537 The critical characteristic of the process is that each repetitive
10538 action is simple. In this case, at each step we add only two numbers,
10539 the number of pebbles in the row and the total already found. This
10540 process of adding two numbers is repeated again and again until the last
10541 row has been added to the total of all the preceding rows. In a more
10542 complex loop the repetitive action might not be so simple, but it will
10543 be simpler than doing everything all at once.
10545 @node Inc Example parts
10546 @unnumberedsubsubsec The parts of the function definition
10548 The preceding analysis gives us the bones of our function definition:
10549 first, we will need a variable that we can call @code{total} that will
10550 be the total number of pebbles. This will be the value returned by
10553 Second, we know that the function will require an argument: this
10554 argument will be the total number of rows in the triangle. It can be
10555 called @code{number-of-rows}.
10557 Finally, we need a variable to use as a counter. We could call this
10558 variable @code{counter}, but a better name is @code{row-number}. That
10559 is because what the counter does in this function is count rows, and a
10560 program should be written to be as understandable as possible.
10562 When the Lisp interpreter first starts evaluating the expressions in the
10563 function, the value of @code{total} should be set to zero, since we have
10564 not added anything to it. Then the function should add the number of
10565 pebbles in the first row to the total, and then add the number of
10566 pebbles in the second to the total, and then add the number of
10567 pebbles in the third row to the total, and so on, until there are no
10568 more rows left to add.
10570 Both @code{total} and @code{row-number} are used only inside the
10571 function, so they can be declared as local variables with @code{let}
10572 and given initial values. Clearly, the initial value for @code{total}
10573 should be 0. The initial value of @code{row-number} should be 1,
10574 since we start with the first row. This means that the @code{let}
10575 statement will look like this:
10585 After the internal variables are declared and bound to their initial
10586 values, we can begin the @code{while} loop. The expression that serves
10587 as the test should return a value of @code{t} for true so long as the
10588 @code{row-number} is less than or equal to the @code{number-of-rows}.
10589 (If the expression tests true only so long as the row number is less
10590 than the number of rows in the triangle, the last row will never be
10591 added to the total; hence the row number has to be either less than or
10592 equal to the number of rows.)
10595 @findex <= @r{(less than or equal)}
10596 Lisp provides the @code{<=} function that returns true if the value of
10597 its first argument is less than or equal to the value of its second
10598 argument and false otherwise. So the expression that the @code{while}
10599 will evaluate as its test should look like this:
10602 (<= row-number number-of-rows)
10605 The total number of pebbles can be found by repeatedly adding the number
10606 of pebbles in a row to the total already found. Since the number of
10607 pebbles in the row is equal to the row number, the total can be found by
10608 adding the row number to the total. (Clearly, in a more complex
10609 situation, the number of pebbles in the row might be related to the row
10610 number in a more complicated way; if this were the case, the row number
10611 would be replaced by the appropriate expression.)
10614 (setq total (+ total row-number))
10618 What this does is set the new value of @code{total} to be equal to the
10619 sum of adding the number of pebbles in the row to the previous total.
10621 After setting the value of @code{total}, the conditions need to be
10622 established for the next repetition of the loop, if there is one. This
10623 is done by incrementing the value of the @code{row-number} variable,
10624 which serves as a counter. After the @code{row-number} variable has
10625 been incremented, the true-or-false-test at the beginning of the
10626 @code{while} loop tests whether its value is still less than or equal to
10627 the value of the @code{number-of-rows} and if it is, adds the new value
10628 of the @code{row-number} variable to the @code{total} of the previous
10629 repetition of the loop.
10632 The built-in Emacs Lisp function @code{1+} adds 1 to a number, so the
10633 @code{row-number} variable can be incremented with this expression:
10636 (setq row-number (1+ row-number))
10639 @node Inc Example altogether
10640 @unnumberedsubsubsec Putting the function definition together
10642 We have created the parts for the function definition; now we need to
10646 First, the contents of the @code{while} expression:
10650 (while (<= row-number number-of-rows) ; @r{true-or-false-test}
10651 (setq total (+ total row-number))
10652 (setq row-number (1+ row-number))) ; @r{incrementer}
10656 Along with the @code{let} expression varlist, this very nearly
10657 completes the body of the function definition. However, it requires
10658 one final element, the need for which is somewhat subtle.
10660 The final touch is to place the variable @code{total} on a line by
10661 itself after the @code{while} expression. Otherwise, the value returned
10662 by the whole function is the value of the last expression that is
10663 evaluated in the body of the @code{let}, and this is the value
10664 returned by the @code{while}, which is always @code{nil}.
10666 This may not be evident at first sight. It almost looks as if the
10667 incrementing expression is the last expression of the whole function.
10668 But that expression is part of the body of the @code{while}; it is the
10669 last element of the list that starts with the symbol @code{while}.
10670 Moreover, the whole of the @code{while} loop is a list within the body
10674 In outline, the function will look like this:
10678 (defun @var{name-of-function} (@var{argument-list})
10679 "@var{documentation}@dots{}"
10680 (let (@var{varlist})
10681 (while (@var{true-or-false-test})
10682 @var{body-of-while}@dots{} )
10683 @dots{} )) ; @r{Need final expression here.}
10687 The result of evaluating the @code{let} is what is going to be returned
10688 by the @code{defun} since the @code{let} is not embedded within any
10689 containing list, except for the @code{defun} as a whole. However, if
10690 the @code{while} is the last element of the @code{let} expression, the
10691 function will always return @code{nil}. This is not what we want!
10692 Instead, what we want is the value of the variable @code{total}. This
10693 is returned by simply placing the symbol as the last element of the list
10694 starting with @code{let}. It gets evaluated after the preceding
10695 elements of the list are evaluated, which means it gets evaluated after
10696 it has been assigned the correct value for the total.
10698 It may be easier to see this by printing the list starting with
10699 @code{let} all on one line. This format makes it evident that the
10700 @var{varlist} and @code{while} expressions are the second and third
10701 elements of the list starting with @code{let}, and the @code{total} is
10706 (let (@var{varlist}) (while (@var{true-or-false-test}) @var{body-of-while}@dots{} ) total)
10711 Putting everything together, the @code{triangle} function definition
10716 (defun triangle (number-of-rows) ; @r{Version with}
10717 ; @r{ incrementing counter.}
10718 "Add up the number of pebbles in a triangle.
10719 The first row has one pebble, the second row two pebbles,
10720 the third row three pebbles, and so on.
10721 The argument is NUMBER-OF-ROWS."
10726 (while (<= row-number number-of-rows)
10727 (setq total (+ total row-number))
10728 (setq row-number (1+ row-number)))
10734 After you have installed @code{triangle} by evaluating the function, you
10735 can try it out. Here are two examples:
10746 The sum of the first four numbers is 10 and the sum of the first seven
10749 @node Decrementing Loop
10750 @subsection Loop with a Decrementing Counter
10752 Another common way to write a @code{while} loop is to write the test
10753 so that it determines whether a counter is greater than zero. So long
10754 as the counter is greater than zero, the loop is repeated. But when
10755 the counter is equal to or less than zero, the loop is stopped. For
10756 this to work, the counter has to start out greater than zero and then
10757 be made smaller and smaller by a form that is evaluated
10760 The test will be an expression such as @code{(> counter 0)} which
10761 returns @code{t} for true if the value of @code{counter} is greater
10762 than zero, and @code{nil} for false if the value of @code{counter} is
10763 equal to or less than zero. The expression that makes the number
10764 smaller and smaller can be a simple @code{setq} such as @code{(setq
10765 counter (1- counter))}, where @code{1-} is a built-in function in
10766 Emacs Lisp that subtracts 1 from its argument.
10769 The template for a decrementing @code{while} loop looks like this:
10773 (while (> counter 0) ; @r{true-or-false-test}
10775 (setq counter (1- counter))) ; @r{decrementer}
10780 * Decrementing Example:: More pebbles on the beach.
10781 * Dec Example parts:: The parts of the function definition.
10782 * Dec Example altogether:: Putting the function definition together.
10785 @node Decrementing Example
10786 @unnumberedsubsubsec Example with decrementing counter
10788 To illustrate a loop with a decrementing counter, we will rewrite the
10789 @code{triangle} function so the counter decreases to zero.
10791 This is the reverse of the earlier version of the function. In this
10792 case, to find out how many pebbles are needed to make a triangle with
10793 3 rows, add the number of pebbles in the third row, 3, to the number
10794 in the preceding row, 2, and then add the total of those two rows to
10795 the row that precedes them, which is 1.
10797 Likewise, to find the number of pebbles in a triangle with 7 rows, add
10798 the number of pebbles in the seventh row, 7, to the number in the
10799 preceding row, which is 6, and then add the total of those two rows to
10800 the row that precedes them, which is 5, and so on. As in the previous
10801 example, each addition only involves adding two numbers, the total of
10802 the rows already added up and the number of pebbles in the row that is
10803 being added to the total. This process of adding two numbers is
10804 repeated again and again until there are no more pebbles to add.
10806 We know how many pebbles to start with: the number of pebbles in the
10807 last row is equal to the number of rows. If the triangle has seven
10808 rows, the number of pebbles in the last row is 7. Likewise, we know how
10809 many pebbles are in the preceding row: it is one less than the number in
10812 @node Dec Example parts
10813 @unnumberedsubsubsec The parts of the function definition
10815 We start with three variables: the total number of rows in the
10816 triangle; the number of pebbles in a row; and the total number of
10817 pebbles, which is what we want to calculate. These variables can be
10818 named @code{number-of-rows}, @code{number-of-pebbles-in-row}, and
10819 @code{total}, respectively.
10821 Both @code{total} and @code{number-of-pebbles-in-row} are used only
10822 inside the function and are declared with @code{let}. The initial
10823 value of @code{total} should, of course, be zero. However, the
10824 initial value of @code{number-of-pebbles-in-row} should be equal to
10825 the number of rows in the triangle, since the addition will start with
10829 This means that the beginning of the @code{let} expression will look
10835 (number-of-pebbles-in-row number-of-rows))
10840 The total number of pebbles can be found by repeatedly adding the number
10841 of pebbles in a row to the total already found, that is, by repeatedly
10842 evaluating the following expression:
10845 (setq total (+ total number-of-pebbles-in-row))
10849 After the @code{number-of-pebbles-in-row} is added to the @code{total},
10850 the @code{number-of-pebbles-in-row} should be decremented by one, since
10851 the next time the loop repeats, the preceding row will be
10852 added to the total.
10854 The number of pebbles in a preceding row is one less than the number of
10855 pebbles in a row, so the built-in Emacs Lisp function @code{1-} can be
10856 used to compute the number of pebbles in the preceding row. This can be
10857 done with the following expression:
10861 (setq number-of-pebbles-in-row
10862 (1- number-of-pebbles-in-row))
10866 Finally, we know that the @code{while} loop should stop making repeated
10867 additions when there are no pebbles in a row. So the test for
10868 the @code{while} loop is simply:
10871 (while (> number-of-pebbles-in-row 0)
10874 @node Dec Example altogether
10875 @unnumberedsubsubsec Putting the function definition together
10877 We can put these expressions together to create a function definition
10878 that works. However, on examination, we find that one of the local
10879 variables is unneeded!
10882 The function definition looks like this:
10886 ;;; @r{First subtractive version.}
10887 (defun triangle (number-of-rows)
10888 "Add up the number of pebbles in a triangle."
10890 (number-of-pebbles-in-row number-of-rows))
10891 (while (> number-of-pebbles-in-row 0)
10892 (setq total (+ total number-of-pebbles-in-row))
10893 (setq number-of-pebbles-in-row
10894 (1- number-of-pebbles-in-row)))
10899 As written, this function works.
10901 However, we do not need @code{number-of-pebbles-in-row}.
10903 @cindex Argument as local variable
10904 When the @code{triangle} function is evaluated, the symbol
10905 @code{number-of-rows} will be bound to a number, giving it an initial
10906 value. That number can be changed in the body of the function as if
10907 it were a local variable, without any fear that such a change will
10908 effect the value of the variable outside of the function. This is a
10909 very useful characteristic of Lisp; it means that the variable
10910 @code{number-of-rows} can be used anywhere in the function where
10911 @code{number-of-pebbles-in-row} is used.
10914 Here is a second version of the function written a bit more cleanly:
10918 (defun triangle (number) ; @r{Second version.}
10919 "Return sum of numbers 1 through NUMBER inclusive."
10921 (while (> number 0)
10922 (setq total (+ total number))
10923 (setq number (1- number)))
10928 In brief, a properly written @code{while} loop will consist of three parts:
10932 A test that will return false after the loop has repeated itself the
10933 correct number of times.
10936 An expression the evaluation of which will return the value desired
10937 after being repeatedly evaluated.
10940 An expression to change the value passed to the true-or-false-test so
10941 that the test returns false after the loop has repeated itself the right
10945 @node dolist dotimes
10946 @section Save your time: @code{dolist} and @code{dotimes}
10948 In addition to @code{while}, both @code{dolist} and @code{dotimes}
10949 provide for looping. Sometimes these are quicker to write than the
10950 equivalent @code{while} loop. Both are Lisp macros. (@xref{Macros, ,
10951 Macros, elisp, The GNU Emacs Lisp Reference Manual}. )
10953 @code{dolist} works like a @code{while} loop that `@sc{cdr}s down a
10954 list': @code{dolist} automatically shortens the list each time it
10955 loops---takes the @sc{cdr} of the list---and binds the @sc{car} of
10956 each shorter version of the list to the first of its arguments.
10958 @code{dotimes} loops a specific number of times: you specify the number.
10966 @unnumberedsubsec The @code{dolist} Macro
10969 Suppose, for example, you want to reverse a list, so that
10970 ``first'' ``second'' ``third'' becomes ``third'' ``second'' ``first''.
10973 In practice, you would use the @code{reverse} function, like this:
10977 (setq animals '(gazelle giraffe lion tiger))
10985 Here is how you could reverse the list using a @code{while} loop:
10989 (setq animals '(gazelle giraffe lion tiger))
10991 (defun reverse-list-with-while (list)
10992 "Using while, reverse the order of LIST."
10993 (let (value) ; make sure list starts empty
10995 (setq value (cons (car list) value))
10996 (setq list (cdr list)))
10999 (reverse-list-with-while animals)
11005 And here is how you could use the @code{dolist} macro:
11009 (setq animals '(gazelle giraffe lion tiger))
11011 (defun reverse-list-with-dolist (list)
11012 "Using dolist, reverse the order of LIST."
11013 (let (value) ; make sure list starts empty
11014 (dolist (element list value)
11015 (setq value (cons element value)))))
11017 (reverse-list-with-dolist animals)
11023 In Info, you can place your cursor after the closing parenthesis of
11024 each expression and type @kbd{C-x C-e}; in each case, you should see
11027 (tiger lion giraffe gazelle)
11033 For this example, the existing @code{reverse} function is obviously best.
11034 The @code{while} loop is just like our first example (@pxref{Loop
11035 Example, , A @code{while} Loop and a List}). The @code{while} first
11036 checks whether the list has elements; if so, it constructs a new list
11037 by adding the first element of the list to the existing list (which in
11038 the first iteration of the loop is @code{nil}). Since the second
11039 element is prepended in front of the first element, and the third
11040 element is prepended in front of the second element, the list is reversed.
11042 In the expression using a @code{while} loop,
11043 the @w{@code{(setq list (cdr list))}}
11044 expression shortens the list, so the @code{while} loop eventually
11045 stops. In addition, it provides the @code{cons} expression with a new
11046 first element by creating a new and shorter list at each repetition of
11049 The @code{dolist} expression does very much the same as the
11050 @code{while} expression, except that the @code{dolist} macro does some
11051 of the work you have to do when writing a @code{while} expression.
11053 Like a @code{while} loop, a @code{dolist} loops. What is different is
11054 that it automatically shortens the list each time it loops---it
11055 `@sc{cdr}s down the list' on its own---and it automatically binds
11056 the @sc{car} of each shorter version of the list to the first of its
11059 In the example, the @sc{car} of each shorter version of the list is
11060 referred to using the symbol @samp{element}, the list itself is called
11061 @samp{list}, and the value returned is called @samp{value}. The
11062 remainder of the @code{dolist} expression is the body.
11064 The @code{dolist} expression binds the @sc{car} of each shorter
11065 version of the list to @code{element} and then evaluates the body of
11066 the expression; and repeats the loop. The result is returned in
11070 @unnumberedsubsec The @code{dotimes} Macro
11073 The @code{dotimes} macro is similar to @code{dolist}, except that it
11074 loops a specific number of times.
11076 The first argument to @code{dotimes} is assigned the numbers 0, 1, 2
11077 and so forth each time around the loop, and the value of the third
11078 argument is returned. You need to provide the value of the second
11079 argument, which is how many times the macro loops.
11082 For example, the following binds the numbers from 0 up to, but not
11083 including, the number 3 to the first argument, @var{number}, and then
11084 constructs a list of the three numbers. (The first number is 0, the
11085 second number is 1, and the third number is 2; this makes a total of
11086 three numbers in all, starting with zero as the first number.)
11090 (let (value) ; otherwise a value is a void variable
11091 (dotimes (number 3 value)
11092 (setq value (cons number value))))
11099 @code{dotimes} returns @code{value}, so the way to use
11100 @code{dotimes} is to operate on some expression @var{number} number of
11101 times and then return the result, either as a list or an atom.
11104 Here is an example of a @code{defun} that uses @code{dotimes} to add
11105 up the number of pebbles in a triangle.
11109 (defun triangle-using-dotimes (number-of-rows)
11110 "Using dotimes, add up the number of pebbles in a triangle."
11111 (let ((total 0)) ; otherwise a total is a void variable
11112 (dotimes (number number-of-rows total)
11113 (setq total (+ total (1+ number))))))
11115 (triangle-using-dotimes 4)
11123 A recursive function contains code that tells the Lisp interpreter to
11124 call a program that runs exactly like itself, but with slightly
11125 different arguments. The code runs exactly the same because it has
11126 the same name. However, even though the program has the same name, it
11127 is not the same entity. It is different. In the jargon, it is a
11128 different `instance'.
11130 Eventually, if the program is written correctly, the `slightly
11131 different arguments' will become sufficiently different from the first
11132 arguments that the final instance will stop.
11135 * Building Robots:: Same model, different serial number ...
11136 * Recursive Definition Parts:: Walk until you stop ...
11137 * Recursion with list:: Using a list as the test whether to recurse.
11138 * Recursive triangle function::
11139 * Recursion with cond::
11140 * Recursive Patterns:: Often used templates.
11141 * No Deferment:: Don't store up work ...
11142 * No deferment solution::
11145 @node Building Robots
11146 @subsection Building Robots: Extending the Metaphor
11147 @cindex Building robots
11148 @cindex Robots, building
11150 It is sometimes helpful to think of a running program as a robot that
11151 does a job. In doing its job, a recursive function calls on a second
11152 robot to help it. The second robot is identical to the first in every
11153 way, except that the second robot helps the first and has been
11154 passed different arguments than the first.
11156 In a recursive function, the second robot may call a third; and the
11157 third may call a fourth, and so on. Each of these is a different
11158 entity; but all are clones.
11160 Since each robot has slightly different instructions---the arguments
11161 will differ from one robot to the next---the last robot should know
11164 Let's expand on the metaphor in which a computer program is a robot.
11166 A function definition provides the blueprints for a robot. When you
11167 install a function definition, that is, when you evaluate a
11168 @code{defun} macro, you install the necessary equipment to build
11169 robots. It is as if you were in a factory, setting up an assembly
11170 line. Robots with the same name are built according to the same
11171 blueprints. So they have, as it were, the same `model number', but a
11172 different `serial number'.
11174 We often say that a recursive function `calls itself'. What we mean
11175 is that the instructions in a recursive function cause the Lisp
11176 interpreter to run a different function that has the same name and
11177 does the same job as the first, but with different arguments.
11179 It is important that the arguments differ from one instance to the
11180 next; otherwise, the process will never stop.
11182 @node Recursive Definition Parts
11183 @subsection The Parts of a Recursive Definition
11184 @cindex Parts of a Recursive Definition
11185 @cindex Recursive Definition Parts
11187 A recursive function typically contains a conditional expression which
11192 A true-or-false-test that determines whether the function is called
11193 again, here called the @dfn{do-again-test}.
11196 The name of the function. When this name is called, a new instance of
11197 the function---a new robot, as it were---is created and told what to do.
11200 An expression that returns a different value each time the function is
11201 called, here called the @dfn{next-step-expression}. Consequently, the
11202 argument (or arguments) passed to the new instance of the function
11203 will be different from that passed to the previous instance. This
11204 causes the conditional expression, the @dfn{do-again-test}, to test
11205 false after the correct number of repetitions.
11208 Recursive functions can be much simpler than any other kind of
11209 function. Indeed, when people first start to use them, they often look
11210 so mysteriously simple as to be incomprehensible. Like riding a
11211 bicycle, reading a recursive function definition takes a certain knack
11212 which is hard at first but then seems simple.
11215 There are several different common recursive patterns. A very simple
11216 pattern looks like this:
11220 (defun @var{name-of-recursive-function} (@var{argument-list})
11221 "@var{documentation}@dots{}"
11222 (if @var{do-again-test}
11224 (@var{name-of-recursive-function}
11225 @var{next-step-expression})))
11229 Each time a recursive function is evaluated, a new instance of it is
11230 created and told what to do. The arguments tell the instance what to do.
11232 An argument is bound to the value of the next-step-expression. Each
11233 instance runs with a different value of the next-step-expression.
11235 The value in the next-step-expression is used in the do-again-test.
11237 The value returned by the next-step-expression is passed to the new
11238 instance of the function, which evaluates it (or some
11239 transmogrification of it) to determine whether to continue or stop.
11240 The next-step-expression is designed so that the do-again-test returns
11241 false when the function should no longer be repeated.
11243 The do-again-test is sometimes called the @dfn{stop condition},
11244 since it stops the repetitions when it tests false.
11246 @node Recursion with list
11247 @subsection Recursion with a List
11249 The example of a @code{while} loop that printed the elements of a list
11250 of numbers can be written recursively. Here is the code, including
11251 an expression to set the value of the variable @code{animals} to a list.
11253 If you are reading this in Info in Emacs, you can evaluate this
11254 expression directly in Info. Otherwise, you must copy the example
11255 to the @file{*scratch*} buffer and evaluate each expression there.
11256 Use @kbd{C-u C-x C-e} to evaluate the
11257 @code{(print-elements-recursively animals)} expression so that the
11258 results are printed in the buffer; otherwise the Lisp interpreter will
11259 try to squeeze the results into the one line of the echo area.
11261 Also, place your cursor immediately after the last closing parenthesis
11262 of the @code{print-elements-recursively} function, before the comment.
11263 Otherwise, the Lisp interpreter will try to evaluate the comment.
11265 @findex print-elements-recursively
11268 (setq animals '(gazelle giraffe lion tiger))
11270 (defun print-elements-recursively (list)
11271 "Print each element of LIST on a line of its own.
11273 (when list ; @r{do-again-test}
11274 (print (car list)) ; @r{body}
11275 (print-elements-recursively ; @r{recursive call}
11276 (cdr list)))) ; @r{next-step-expression}
11278 (print-elements-recursively animals)
11282 The @code{print-elements-recursively} function first tests whether
11283 there is any content in the list; if there is, the function prints the
11284 first element of the list, the @sc{car} of the list. Then the
11285 function `invokes itself', but gives itself as its argument, not the
11286 whole list, but the second and subsequent elements of the list, the
11287 @sc{cdr} of the list.
11289 Put another way, if the list is not empty, the function invokes
11290 another instance of code that is similar to the initial code, but is a
11291 different thread of execution, with different arguments than the first
11294 Put in yet another way, if the list is not empty, the first robot
11295 assembles a second robot and tells it what to do; the second robot is
11296 a different individual from the first, but is the same model.
11298 When the second evaluation occurs, the @code{when} expression is
11299 evaluated and if true, prints the first element of the list it
11300 receives as its argument (which is the second element of the original
11301 list). Then the function `calls itself' with the @sc{cdr} of the list
11302 it is invoked with, which (the second time around) is the @sc{cdr} of
11303 the @sc{cdr} of the original list.
11305 Note that although we say that the function `calls itself', what we
11306 mean is that the Lisp interpreter assembles and instructs a new
11307 instance of the program. The new instance is a clone of the first,
11308 but is a separate individual.
11310 Each time the function `invokes itself', it invokes itself on a
11311 shorter version of the original list. It creates a new instance that
11312 works on a shorter list.
11314 Eventually, the function invokes itself on an empty list. It creates
11315 a new instance whose argument is @code{nil}. The conditional expression
11316 tests the value of @code{list}. Since the value of @code{list} is
11317 @code{nil}, the @code{when} expression tests false so the then-part is
11318 not evaluated. The function as a whole then returns @code{nil}.
11321 When you evaluate the expression @code{(print-elements-recursively
11322 animals)} in the @file{*scratch*} buffer, you see this result:
11338 @node Recursive triangle function
11339 @subsection Recursion in Place of a Counter
11340 @findex triangle-recursively
11343 The @code{triangle} function described in a previous section can also
11344 be written recursively. It looks like this:
11348 (defun triangle-recursively (number)
11349 "Return the sum of the numbers 1 through NUMBER inclusive.
11351 (if (= number 1) ; @r{do-again-test}
11353 (+ number ; @r{else-part}
11354 (triangle-recursively ; @r{recursive call}
11355 (1- number))))) ; @r{next-step-expression}
11357 (triangle-recursively 7)
11362 You can install this function by evaluating it and then try it by
11363 evaluating @code{(triangle-recursively 7)}. (Remember to put your
11364 cursor immediately after the last parenthesis of the function
11365 definition, before the comment.) The function evaluates to 28.
11367 To understand how this function works, let's consider what happens in the
11368 various cases when the function is passed 1, 2, 3, or 4 as the value of
11372 * Recursive Example arg of 1 or 2::
11373 * Recursive Example arg of 3 or 4::
11377 @node Recursive Example arg of 1 or 2
11378 @unnumberedsubsubsec An argument of 1 or 2
11381 First, what happens if the value of the argument is 1?
11383 The function has an @code{if} expression after the documentation
11384 string. It tests whether the value of @code{number} is equal to 1; if
11385 so, Emacs evaluates the then-part of the @code{if} expression, which
11386 returns the number 1 as the value of the function. (A triangle with
11387 one row has one pebble in it.)
11389 Suppose, however, that the value of the argument is 2. In this case,
11390 Emacs evaluates the else-part of the @code{if} expression.
11393 The else-part consists of an addition, the recursive call to
11394 @code{triangle-recursively} and a decrementing action; and it looks like
11398 (+ number (triangle-recursively (1- number)))
11401 When Emacs evaluates this expression, the innermost expression is
11402 evaluated first; then the other parts in sequence. Here are the steps
11406 @item Step 1 @w{ } Evaluate the innermost expression.
11408 The innermost expression is @code{(1- number)} so Emacs decrements the
11409 value of @code{number} from 2 to 1.
11411 @item Step 2 @w{ } Evaluate the @code{triangle-recursively} function.
11413 The Lisp interpreter creates an individual instance of
11414 @code{triangle-recursively}. It does not matter that this function is
11415 contained within itself. Emacs passes the result Step 1 as the
11416 argument used by this instance of the @code{triangle-recursively}
11419 In this case, Emacs evaluates @code{triangle-recursively} with an
11420 argument of 1. This means that this evaluation of
11421 @code{triangle-recursively} returns 1.
11423 @item Step 3 @w{ } Evaluate the value of @code{number}.
11425 The variable @code{number} is the second element of the list that
11426 starts with @code{+}; its value is 2.
11428 @item Step 4 @w{ } Evaluate the @code{+} expression.
11430 The @code{+} expression receives two arguments, the first
11431 from the evaluation of @code{number} (Step 3) and the second from the
11432 evaluation of @code{triangle-recursively} (Step 2).
11434 The result of the addition is the sum of 2 plus 1, and the number 3 is
11435 returned, which is correct. A triangle with two rows has three
11439 @node Recursive Example arg of 3 or 4
11440 @unnumberedsubsubsec An argument of 3 or 4
11442 Suppose that @code{triangle-recursively} is called with an argument of
11446 @item Step 1 @w{ } Evaluate the do-again-test.
11448 The @code{if} expression is evaluated first. This is the do-again
11449 test and returns false, so the else-part of the @code{if} expression
11450 is evaluated. (Note that in this example, the do-again-test causes
11451 the function to call itself when it tests false, not when it tests
11454 @item Step 2 @w{ } Evaluate the innermost expression of the else-part.
11456 The innermost expression of the else-part is evaluated, which decrements
11457 3 to 2. This is the next-step-expression.
11459 @item Step 3 @w{ } Evaluate the @code{triangle-recursively} function.
11461 The number 2 is passed to the @code{triangle-recursively} function.
11463 We already know what happens when Emacs evaluates @code{triangle-recursively} with
11464 an argument of 2. After going through the sequence of actions described
11465 earlier, it returns a value of 3. So that is what will happen here.
11467 @item Step 4 @w{ } Evaluate the addition.
11469 3 will be passed as an argument to the addition and will be added to the
11470 number with which the function was called, which is 3.
11474 The value returned by the function as a whole will be 6.
11476 Now that we know what will happen when @code{triangle-recursively} is
11477 called with an argument of 3, it is evident what will happen if it is
11478 called with an argument of 4:
11482 In the recursive call, the evaluation of
11485 (triangle-recursively (1- 4))
11490 will return the value of evaluating
11493 (triangle-recursively 3)
11497 which is 6 and this value will be added to 4 by the addition in the
11502 The value returned by the function as a whole will be 10.
11504 Each time @code{triangle-recursively} is evaluated, it evaluates a
11505 version of itself---a different instance of itself---with a smaller
11506 argument, until the argument is small enough so that it does not
11509 Note that this particular design for a recursive function
11510 requires that operations be deferred.
11512 Before @code{(triangle-recursively 7)} can calculate its answer, it
11513 must call @code{(triangle-recursively 6)}; and before
11514 @code{(triangle-recursively 6)} can calculate its answer, it must call
11515 @code{(triangle-recursively 5)}; and so on. That is to say, the
11516 calculation that @code{(triangle-recursively 7)} makes must be
11517 deferred until @code{(triangle-recursively 6)} makes its calculation;
11518 and @code{(triangle-recursively 6)} must defer until
11519 @code{(triangle-recursively 5)} completes; and so on.
11521 If each of these instances of @code{triangle-recursively} are thought
11522 of as different robots, the first robot must wait for the second to
11523 complete its job, which must wait until the third completes, and so
11526 There is a way around this kind of waiting, which we will discuss in
11527 @ref{No Deferment, , Recursion without Deferments}.
11529 @node Recursion with cond
11530 @subsection Recursion Example Using @code{cond}
11533 The version of @code{triangle-recursively} described earlier is written
11534 with the @code{if} special form. It can also be written using another
11535 special form called @code{cond}. The name of the special form
11536 @code{cond} is an abbreviation of the word @samp{conditional}.
11538 Although the @code{cond} special form is not used as often in the
11539 Emacs Lisp sources as @code{if}, it is used often enough to justify
11543 The template for a @code{cond} expression looks like this:
11553 where the @var{body} is a series of lists.
11556 Written out more fully, the template looks like this:
11561 (@var{first-true-or-false-test} @var{first-consequent})
11562 (@var{second-true-or-false-test} @var{second-consequent})
11563 (@var{third-true-or-false-test} @var{third-consequent})
11568 When the Lisp interpreter evaluates the @code{cond} expression, it
11569 evaluates the first element (the @sc{car} or true-or-false-test) of
11570 the first expression in a series of expressions within the body of the
11573 If the true-or-false-test returns @code{nil} the rest of that
11574 expression, the consequent, is skipped and the true-or-false-test of the
11575 next expression is evaluated. When an expression is found whose
11576 true-or-false-test returns a value that is not @code{nil}, the
11577 consequent of that expression is evaluated. The consequent can be one
11578 or more expressions. If the consequent consists of more than one
11579 expression, the expressions are evaluated in sequence and the value of
11580 the last one is returned. If the expression does not have a consequent,
11581 the value of the true-or-false-test is returned.
11583 If none of the true-or-false-tests test true, the @code{cond} expression
11584 returns @code{nil}.
11587 Written using @code{cond}, the @code{triangle} function looks like this:
11591 (defun triangle-using-cond (number)
11592 (cond ((<= number 0) 0)
11595 (+ number (triangle-using-cond (1- number))))))
11600 In this example, the @code{cond} returns 0 if the number is less than or
11601 equal to 0, it returns 1 if the number is 1 and it evaluates @code{(+
11602 number (triangle-using-cond (1- number)))} if the number is greater than
11605 @node Recursive Patterns
11606 @subsection Recursive Patterns
11607 @cindex Recursive Patterns
11609 Here are three common recursive patterns. Each involves a list.
11610 Recursion does not need to involve lists, but Lisp is designed for lists
11611 and this provides a sense of its primal capabilities.
11620 @unnumberedsubsubsec Recursive Pattern: @emph{every}
11621 @cindex Every, type of recursive pattern
11622 @cindex Recursive pattern: every
11624 In the @code{every} recursive pattern, an action is performed on every
11628 The basic pattern is:
11632 If a list be empty, return @code{nil}.
11634 Else, act on the beginning of the list (the @sc{car} of the list)
11637 through a recursive call by the function on the rest (the
11638 @sc{cdr}) of the list,
11640 and, optionally, combine the acted-on element, using @code{cons},
11641 with the results of acting on the rest.
11650 (defun square-each (numbers-list)
11651 "Square each of a NUMBERS LIST, recursively."
11652 (if (not numbers-list) ; do-again-test
11655 (* (car numbers-list) (car numbers-list))
11656 (square-each (cdr numbers-list))))) ; next-step-expression
11660 (square-each '(1 2 3))
11667 If @code{numbers-list} is empty, do nothing. But if it has content,
11668 construct a list combining the square of the first number in the list
11669 with the result of the recursive call.
11671 (The example follows the pattern exactly: @code{nil} is returned if
11672 the numbers' list is empty. In practice, you would write the
11673 conditional so it carries out the action when the numbers' list is not
11676 The @code{print-elements-recursively} function (@pxref{Recursion with
11677 list, , Recursion with a List}) is another example of an @code{every}
11678 pattern, except in this case, rather than bring the results together
11679 using @code{cons}, we print each element of output.
11682 The @code{print-elements-recursively} function looks like this:
11686 (setq animals '(gazelle giraffe lion tiger))
11690 (defun print-elements-recursively (list)
11691 "Print each element of LIST on a line of its own.
11693 (when list ; @r{do-again-test}
11694 (print (car list)) ; @r{body}
11695 (print-elements-recursively ; @r{recursive call}
11696 (cdr list)))) ; @r{next-step-expression}
11698 (print-elements-recursively animals)
11703 The pattern for @code{print-elements-recursively} is:
11707 When the list is empty, do nothing.
11709 But when the list has at least one element,
11712 act on the beginning of the list (the @sc{car} of the list),
11714 and make a recursive call on the rest (the @sc{cdr}) of the list.
11719 @unnumberedsubsubsec Recursive Pattern: @emph{accumulate}
11720 @cindex Accumulate, type of recursive pattern
11721 @cindex Recursive pattern: accumulate
11723 Another recursive pattern is called the @code{accumulate} pattern. In
11724 the @code{accumulate} recursive pattern, an action is performed on
11725 every element of a list and the result of that action is accumulated
11726 with the results of performing the action on the other elements.
11728 This is very like the `every' pattern using @code{cons}, except that
11729 @code{cons} is not used, but some other combiner.
11736 If a list be empty, return zero or some other constant.
11738 Else, act on the beginning of the list (the @sc{car} of the list),
11741 and combine that acted-on element, using @code{+} or
11742 some other combining function, with
11744 a recursive call by the function on the rest (the @sc{cdr}) of the list.
11749 Here is an example:
11753 (defun add-elements (numbers-list)
11754 "Add the elements of NUMBERS-LIST together."
11755 (if (not numbers-list)
11757 (+ (car numbers-list) (add-elements (cdr numbers-list)))))
11761 (add-elements '(1 2 3 4))
11766 @xref{Files List, , Making a List of Files}, for an example of the
11767 accumulate pattern.
11770 @unnumberedsubsubsec Recursive Pattern: @emph{keep}
11771 @cindex Keep, type of recursive pattern
11772 @cindex Recursive pattern: keep
11774 A third recursive pattern is called the @code{keep} pattern.
11775 In the @code{keep} recursive pattern, each element of a list is tested;
11776 the element is acted on and the results are kept only if the element
11779 Again, this is very like the `every' pattern, except the element is
11780 skipped unless it meets a criterion.
11783 The pattern has three parts:
11787 If a list be empty, return @code{nil}.
11789 Else, if the beginning of the list (the @sc{car} of the list) passes
11793 act on that element and combine it, using @code{cons} with
11795 a recursive call by the function on the rest (the @sc{cdr}) of the list.
11798 Otherwise, if the beginning of the list (the @sc{car} of the list) fails
11802 skip on that element,
11804 and, recursively call the function on the rest (the @sc{cdr}) of the list.
11809 Here is an example that uses @code{cond}:
11813 (defun keep-three-letter-words (word-list)
11814 "Keep three letter words in WORD-LIST."
11816 ;; First do-again-test: stop-condition
11817 ((not word-list) nil)
11819 ;; Second do-again-test: when to act
11820 ((eq 3 (length (symbol-name (car word-list))))
11821 ;; combine acted-on element with recursive call on shorter list
11822 (cons (car word-list) (keep-three-letter-words (cdr word-list))))
11824 ;; Third do-again-test: when to skip element;
11825 ;; recursively call shorter list with next-step expression
11826 (t (keep-three-letter-words (cdr word-list)))))
11830 (keep-three-letter-words '(one two three four five six))
11831 @result{} (one two six)
11835 It goes without saying that you need not use @code{nil} as the test for
11836 when to stop; and you can, of course, combine these patterns.
11839 @subsection Recursion without Deferments
11840 @cindex Deferment in recursion
11841 @cindex Recursion without Deferments
11843 Let's consider again what happens with the @code{triangle-recursively}
11844 function. We will find that the intermediate calculations are
11845 deferred until all can be done.
11848 Here is the function definition:
11852 (defun triangle-recursively (number)
11853 "Return the sum of the numbers 1 through NUMBER inclusive.
11855 (if (= number 1) ; @r{do-again-test}
11857 (+ number ; @r{else-part}
11858 (triangle-recursively ; @r{recursive call}
11859 (1- number))))) ; @r{next-step-expression}
11863 What happens when we call this function with a argument of 7?
11865 The first instance of the @code{triangle-recursively} function adds
11866 the number 7 to the value returned by a second instance of
11867 @code{triangle-recursively}, an instance that has been passed an
11868 argument of 6. That is to say, the first calculation is:
11871 (+ 7 (triangle-recursively 6))
11875 The first instance of @code{triangle-recursively}---you may want to
11876 think of it as a little robot---cannot complete its job. It must hand
11877 off the calculation for @code{(triangle-recursively 6)} to a second
11878 instance of the program, to a second robot. This second individual is
11879 completely different from the first one; it is, in the jargon, a
11880 `different instantiation'. Or, put another way, it is a different
11881 robot. It is the same model as the first; it calculates triangle
11882 numbers recursively; but it has a different serial number.
11884 And what does @code{(triangle-recursively 6)} return? It returns the
11885 number 6 added to the value returned by evaluating
11886 @code{triangle-recursively} with an argument of 5. Using the robot
11887 metaphor, it asks yet another robot to help it.
11893 (+ 7 6 (triangle-recursively 5))
11897 And what happens next?
11900 (+ 7 6 5 (triangle-recursively 4))
11903 Each time @code{triangle-recursively} is called, except for the last
11904 time, it creates another instance of the program---another robot---and
11905 asks it to make a calculation.
11908 Eventually, the full addition is set up and performed:
11914 This design for the function defers the calculation of the first step
11915 until the second can be done, and defers that until the third can be
11916 done, and so on. Each deferment means the computer must remember what
11917 is being waited on. This is not a problem when there are only a few
11918 steps, as in this example. But it can be a problem when there are
11921 @node No deferment solution
11922 @subsection No Deferment Solution
11923 @cindex No deferment solution
11924 @cindex Defermentless solution
11925 @cindex Solution without deferment
11927 The solution to the problem of deferred operations is to write in a
11928 manner that does not defer operations@footnote{The phrase @dfn{tail
11929 recursive} is used to describe such a process, one that uses
11930 `constant space'.}. This requires
11931 writing to a different pattern, often one that involves writing two
11932 function definitions, an `initialization' function and a `helper'
11935 The `initialization' function sets up the job; the `helper' function
11939 Here are the two function definitions for adding up numbers. They are
11940 so simple, I find them hard to understand.
11944 (defun triangle-initialization (number)
11945 "Return the sum of the numbers 1 through NUMBER inclusive.
11946 This is the `initialization' component of a two function
11947 duo that uses recursion."
11948 (triangle-recursive-helper 0 0 number))
11954 (defun triangle-recursive-helper (sum counter number)
11955 "Return SUM, using COUNTER, through NUMBER inclusive.
11956 This is the `helper' component of a two function duo
11957 that uses recursion."
11958 (if (> counter number)
11960 (triangle-recursive-helper (+ sum counter) ; @r{sum}
11961 (1+ counter) ; @r{counter}
11962 number))) ; @r{number}
11967 Install both function definitions by evaluating them, then call
11968 @code{triangle-initialization} with 2 rows:
11972 (triangle-initialization 2)
11977 The `initialization' function calls the first instance of the `helper'
11978 function with three arguments: zero, zero, and a number which is the
11979 number of rows in the triangle.
11981 The first two arguments passed to the `helper' function are
11982 initialization values. These values are changed when
11983 @code{triangle-recursive-helper} invokes new instances.@footnote{The
11984 jargon is mildly confusing: @code{triangle-recursive-helper} uses a
11985 process that is iterative in a procedure that is recursive. The
11986 process is called iterative because the computer need only record the
11987 three values, @code{sum}, @code{counter}, and @code{number}; the
11988 procedure is recursive because the function `calls itself'. On the
11989 other hand, both the process and the procedure used by
11990 @code{triangle-recursively} are called recursive. The word
11991 `recursive' has different meanings in the two contexts.}
11993 Let's see what happens when we have a triangle that has one row. (This
11994 triangle will have one pebble in it!)
11997 @code{triangle-initialization} will call its helper with
11998 the arguments @w{@code{0 0 1}}. That function will run the conditional
11999 test whether @code{(> counter number)}:
12007 and find that the result is false, so it will invoke
12008 the else-part of the @code{if} clause:
12012 (triangle-recursive-helper
12013 (+ sum counter) ; @r{sum plus counter} @result{} @r{sum}
12014 (1+ counter) ; @r{increment counter} @result{} @r{counter}
12015 number) ; @r{number stays the same}
12021 which will first compute:
12025 (triangle-recursive-helper (+ 0 0) ; @r{sum}
12026 (1+ 0) ; @r{counter}
12030 (triangle-recursive-helper 0 1 1)
12034 Again, @code{(> counter number)} will be false, so again, the Lisp
12035 interpreter will evaluate @code{triangle-recursive-helper}, creating a
12036 new instance with new arguments.
12039 This new instance will be;
12043 (triangle-recursive-helper
12044 (+ sum counter) ; @r{sum plus counter} @result{} @r{sum}
12045 (1+ counter) ; @r{increment counter} @result{} @r{counter}
12046 number) ; @r{number stays the same}
12050 (triangle-recursive-helper 1 2 1)
12054 In this case, the @code{(> counter number)} test will be true! So the
12055 instance will return the value of the sum, which will be 1, as
12058 Now, let's pass @code{triangle-initialization} an argument
12059 of 2, to find out how many pebbles there are in a triangle with two rows.
12061 That function calls @code{(triangle-recursive-helper 0 0 2)}.
12064 In stages, the instances called will be:
12068 @r{sum counter number}
12069 (triangle-recursive-helper 0 1 2)
12071 (triangle-recursive-helper 1 2 2)
12073 (triangle-recursive-helper 3 3 2)
12077 When the last instance is called, the @code{(> counter number)} test
12078 will be true, so the instance will return the value of @code{sum},
12081 This kind of pattern helps when you are writing functions that can use
12082 many resources in a computer.
12085 @node Looping exercise
12086 @section Looping Exercise
12090 Write a function similar to @code{triangle} in which each row has a
12091 value which is the square of the row number. Use a @code{while} loop.
12094 Write a function similar to @code{triangle} that multiplies instead of
12098 Rewrite these two functions recursively. Rewrite these functions
12101 @c comma in printed title causes problem in Info cross reference
12103 Write a function for Texinfo mode that creates an index entry at the
12104 beginning of a paragraph for every @samp{@@dfn} within the paragraph.
12105 (In a Texinfo file, @samp{@@dfn} marks a definition. This book is
12106 written in Texinfo.)
12108 Many of the functions you will need are described in two of the
12109 previous chapters, @ref{Cutting & Storing Text, , Cutting and Storing
12110 Text}, and @ref{Yanking, , Yanking Text Back}. If you use
12111 @code{forward-paragraph} to put the index entry at the beginning of
12112 the paragraph, you will have to use @w{@kbd{C-h f}}
12113 (@code{describe-function}) to find out how to make the command go
12116 For more information, see
12118 @ref{Indicating, , Indicating Definitions, texinfo}.
12121 @ref{Indicating, , Indicating, texinfo, Texinfo Manual}, which goes to
12122 a Texinfo manual in the current directory. Or, if you are on the
12124 @uref{http://www.gnu.org/software/texinfo/manual/texinfo/}
12127 ``Indicating Definitions, Commands, etc.'' in @cite{Texinfo, The GNU
12128 Documentation Format}.
12132 @node Regexp Search
12133 @chapter Regular Expression Searches
12134 @cindex Searches, illustrating
12135 @cindex Regular expression searches
12136 @cindex Patterns, searching for
12137 @cindex Motion by sentence and paragraph
12138 @cindex Sentences, movement by
12139 @cindex Paragraphs, movement by
12141 Regular expression searches are used extensively in GNU Emacs. The
12142 two functions, @code{forward-sentence} and @code{forward-paragraph},
12143 illustrate these searches well. They use regular expressions to find
12144 where to move point. The phrase `regular expression' is often written
12147 Regular expression searches are described in @ref{Regexp Search, ,
12148 Regular Expression Search, emacs, The GNU Emacs Manual}, as well as in
12149 @ref{Regular Expressions, , , elisp, The GNU Emacs Lisp Reference
12150 Manual}. In writing this chapter, I am presuming that you have at
12151 least a mild acquaintance with them. The major point to remember is
12152 that regular expressions permit you to search for patterns as well as
12153 for literal strings of characters. For example, the code in
12154 @code{forward-sentence} searches for the pattern of possible
12155 characters that could mark the end of a sentence, and moves point to
12158 Before looking at the code for the @code{forward-sentence} function, it
12159 is worth considering what the pattern that marks the end of a sentence
12160 must be. The pattern is discussed in the next section; following that
12161 is a description of the regular expression search function,
12162 @code{re-search-forward}. The @code{forward-sentence} function
12163 is described in the section following. Finally, the
12164 @code{forward-paragraph} function is described in the last section of
12165 this chapter. @code{forward-paragraph} is a complex function that
12166 introduces several new features.
12169 * sentence-end:: The regular expression for @code{sentence-end}.
12170 * re-search-forward:: Very similar to @code{search-forward}.
12171 * forward-sentence:: A straightforward example of regexp search.
12172 * forward-paragraph:: A somewhat complex example.
12173 * etags:: How to create your own @file{TAGS} table.
12175 * re-search Exercises::
12179 @section The Regular Expression for @code{sentence-end}
12180 @findex sentence-end
12182 The symbol @code{sentence-end} is bound to the pattern that marks the
12183 end of a sentence. What should this regular expression be?
12185 Clearly, a sentence may be ended by a period, a question mark, or an
12186 exclamation mark. Indeed, in English, only clauses that end with one
12187 of those three characters should be considered the end of a sentence.
12188 This means that the pattern should include the character set:
12194 However, we do not want @code{forward-sentence} merely to jump to a
12195 period, a question mark, or an exclamation mark, because such a character
12196 might be used in the middle of a sentence. A period, for example, is
12197 used after abbreviations. So other information is needed.
12199 According to convention, you type two spaces after every sentence, but
12200 only one space after a period, a question mark, or an exclamation mark in
12201 the body of a sentence. So a period, a question mark, or an exclamation
12202 mark followed by two spaces is a good indicator of an end of sentence.
12203 However, in a file, the two spaces may instead be a tab or the end of a
12204 line. This means that the regular expression should include these three
12205 items as alternatives.
12208 This group of alternatives will look like this:
12219 Here, @samp{$} indicates the end of the line, and I have pointed out
12220 where the tab and two spaces are inserted in the expression. Both are
12221 inserted by putting the actual characters into the expression.
12223 Two backslashes, @samp{\\}, are required before the parentheses and
12224 vertical bars: the first backslash quotes the following backslash in
12225 Emacs; and the second indicates that the following character, the
12226 parenthesis or the vertical bar, is special.
12229 Also, a sentence may be followed by one or more carriage returns, like
12240 Like tabs and spaces, a carriage return is inserted into a regular
12241 expression by inserting it literally. The asterisk indicates that the
12242 @key{RET} is repeated zero or more times.
12244 But a sentence end does not consist only of a period, a question mark or
12245 an exclamation mark followed by appropriate space: a closing quotation
12246 mark or a closing brace of some kind may precede the space. Indeed more
12247 than one such mark or brace may precede the space. These require a
12248 expression that looks like this:
12254 In this expression, the first @samp{]} is the first character in the
12255 expression; the second character is @samp{"}, which is preceded by a
12256 @samp{\} to tell Emacs the @samp{"} is @emph{not} special. The last
12257 three characters are @samp{'}, @samp{)}, and @samp{@}}.
12259 All this suggests what the regular expression pattern for matching the
12260 end of a sentence should be; and, indeed, if we evaluate
12261 @code{sentence-end} we find that it returns the following value:
12266 @result{} "[.?!][]\"')@}]*\\($\\| \\| \\)[
12272 (Well, not in GNU Emacs 22; that is because of an effort to make the
12273 process simpler and to handle more glyphs and languages. When the
12274 value of @code{sentence-end} is @code{nil}, then use the value defined
12275 by the function @code{sentence-end}. (Here is a use of the difference
12276 between a value and a function in Emacs Lisp.) The function returns a
12277 value constructed from the variables @code{sentence-end-base},
12278 @code{sentence-end-double-space}, @code{sentence-end-without-period},
12279 and @code{sentence-end-without-space}. The critical variable is
12280 @code{sentence-end-base}; its global value is similar to the one
12281 described above but it also contains two additional quotation marks.
12282 These have differing degrees of curliness. The
12283 @code{sentence-end-without-period} variable, when true, tells Emacs
12284 that a sentence may end without a period, such as text in Thai.)
12288 (Note that here the @key{TAB}, two spaces, and @key{RET} are shown
12289 literally in the pattern.)
12291 This regular expression can be deciphered as follows:
12295 The first part of the pattern is the three characters, a period, a question
12296 mark and an exclamation mark, within square brackets. The pattern must
12297 begin with one or other of these characters.
12300 The second part of the pattern is the group of closing braces and
12301 quotation marks, which can appear zero or more times. These may follow
12302 the period, question mark or exclamation mark. In a regular expression,
12303 the backslash, @samp{\}, followed by the double quotation mark,
12304 @samp{"}, indicates the class of string-quote characters. Usually, the
12305 double quotation mark is the only character in this class. The
12306 asterisk, @samp{*}, indicates that the items in the previous group (the
12307 group surrounded by square brackets, @samp{[]}) may be repeated zero or
12310 @item \\($\\| \\| \\)
12311 The third part of the pattern is one or other of: either the end of a
12312 line, or two blank spaces, or a tab. The double back-slashes are used
12313 to prevent Emacs from reading the parentheses and vertical bars as part
12314 of the search pattern; the parentheses are used to mark the group and
12315 the vertical bars are used to indicated that the patterns to either side
12316 of them are alternatives. The dollar sign is used to indicate the end
12317 of a line and both the two spaces and the tab are each inserted as is to
12318 indicate what they are.
12321 Finally, the last part of the pattern indicates that the end of the line
12322 or the whitespace following the period, question mark or exclamation
12323 mark may, but need not, be followed by one or more carriage returns. In
12324 the pattern, the carriage return is inserted as an actual carriage
12325 return between square brackets but here it is shown as @key{RET}.
12329 @node re-search-forward
12330 @section The @code{re-search-forward} Function
12331 @findex re-search-forward
12333 The @code{re-search-forward} function is very like the
12334 @code{search-forward} function. (@xref{search-forward, , The
12335 @code{search-forward} Function}.)
12337 @code{re-search-forward} searches for a regular expression. If the
12338 search is successful, it leaves point immediately after the last
12339 character in the target. If the search is backwards, it leaves point
12340 just before the first character in the target. You may tell
12341 @code{re-search-forward} to return @code{t} for true. (Moving point
12342 is therefore a `side effect'.)
12344 Like @code{search-forward}, the @code{re-search-forward} function takes
12349 The first argument is the regular expression that the function searches
12350 for. The regular expression will be a string between quotation marks.
12353 The optional second argument limits how far the function will search; it is a
12354 bound, which is specified as a position in the buffer.
12357 The optional third argument specifies how the function responds to
12358 failure: @code{nil} as the third argument causes the function to
12359 signal an error (and print a message) when the search fails; any other
12360 value causes it to return @code{nil} if the search fails and @code{t}
12361 if the search succeeds.
12364 The optional fourth argument is the repeat count. A negative repeat
12365 count causes @code{re-search-forward} to search backwards.
12369 The template for @code{re-search-forward} looks like this:
12373 (re-search-forward "@var{regular-expression}"
12374 @var{limit-of-search}
12375 @var{what-to-do-if-search-fails}
12376 @var{repeat-count})
12380 The second, third, and fourth arguments are optional. However, if you
12381 want to pass a value to either or both of the last two arguments, you
12382 must also pass a value to all the preceding arguments. Otherwise, the
12383 Lisp interpreter will mistake which argument you are passing the value
12387 In the @code{forward-sentence} function, the regular expression will be
12388 the value of the variable @code{sentence-end}. In simple form, that is:
12392 "[.?!][]\"')@}]*\\($\\| \\| \\)[
12398 The limit of the search will be the end of the paragraph (since a
12399 sentence cannot go beyond a paragraph). If the search fails, the
12400 function will return @code{nil}; and the repeat count will be provided
12401 by the argument to the @code{forward-sentence} function.
12403 @node forward-sentence
12404 @section @code{forward-sentence}
12405 @findex forward-sentence
12407 The command to move the cursor forward a sentence is a straightforward
12408 illustration of how to use regular expression searches in Emacs Lisp.
12409 Indeed, the function looks longer and more complicated than it is; this
12410 is because the function is designed to go backwards as well as forwards;
12411 and, optionally, over more than one sentence. The function is usually
12412 bound to the key command @kbd{M-e}.
12415 * Complete forward-sentence::
12416 * fwd-sentence while loops:: Two @code{while} loops.
12417 * fwd-sentence re-search:: A regular expression search.
12421 @node Complete forward-sentence
12422 @unnumberedsubsec Complete @code{forward-sentence} function definition
12426 Here is the code for @code{forward-sentence}:
12431 (defun forward-sentence (&optional arg)
12432 "Move forward to next `sentence-end'. With argument, repeat.
12433 With negative argument, move backward repeatedly to `sentence-beginning'.
12435 The variable `sentence-end' is a regular expression that matches ends of
12436 sentences. Also, every paragraph boundary terminates sentences as well."
12440 (or arg (setq arg 1))
12441 (let ((opoint (point))
12442 (sentence-end (sentence-end)))
12444 (let ((pos (point))
12445 (par-beg (save-excursion (start-of-paragraph-text) (point))))
12446 (if (and (re-search-backward sentence-end par-beg t)
12447 (or (< (match-end 0) pos)
12448 (re-search-backward sentence-end par-beg t)))
12449 (goto-char (match-end 0))
12450 (goto-char par-beg)))
12451 (setq arg (1+ arg)))
12455 (let ((par-end (save-excursion (end-of-paragraph-text) (point))))
12456 (if (re-search-forward sentence-end par-end t)
12457 (skip-chars-backward " \t\n")
12458 (goto-char par-end)))
12459 (setq arg (1- arg)))
12460 (constrain-to-field nil opoint t)))
12468 (defun forward-sentence (&optional arg)
12469 "Move forward to next sentence-end. With argument, repeat.
12470 With negative argument, move backward repeatedly to sentence-beginning.
12471 Sentence ends are identified by the value of sentence-end
12472 treated as a regular expression. Also, every paragraph boundary
12473 terminates sentences as well."
12477 (or arg (setq arg 1))
12480 (save-excursion (start-of-paragraph-text) (point))))
12481 (if (re-search-backward
12482 (concat sentence-end "[^ \t\n]") par-beg t)
12483 (goto-char (1- (match-end 0)))
12484 (goto-char par-beg)))
12485 (setq arg (1+ arg)))
12488 (save-excursion (end-of-paragraph-text) (point))))
12489 (if (re-search-forward sentence-end par-end t)
12490 (skip-chars-backward " \t\n")
12491 (goto-char par-end)))
12492 (setq arg (1- arg))))
12497 The function looks long at first sight and it is best to look at its
12498 skeleton first, and then its muscle. The way to see the skeleton is to
12499 look at the expressions that start in the left-most columns:
12503 (defun forward-sentence (&optional arg)
12504 "@var{documentation}@dots{}"
12506 (or arg (setq arg 1))
12507 (let ((opoint (point)) (sentence-end (sentence-end)))
12509 (let ((pos (point))
12510 (par-beg (save-excursion (start-of-paragraph-text) (point))))
12511 @var{rest-of-body-of-while-loop-when-going-backwards}
12513 (let ((par-end (save-excursion (end-of-paragraph-text) (point))))
12514 @var{rest-of-body-of-while-loop-when-going-forwards}
12515 @var{handle-forms-and-equivalent}
12519 This looks much simpler! The function definition consists of
12520 documentation, an @code{interactive} expression, an @code{or}
12521 expression, a @code{let} expression, and @code{while} loops.
12523 Let's look at each of these parts in turn.
12525 We note that the documentation is thorough and understandable.
12527 The function has an @code{interactive "p"} declaration. This means
12528 that the processed prefix argument, if any, is passed to the
12529 function as its argument. (This will be a number.) If the function
12530 is not passed an argument (it is optional) then the argument
12531 @code{arg} will be bound to 1.
12533 When @code{forward-sentence} is called non-interactively without an
12534 argument, @code{arg} is bound to @code{nil}. The @code{or} expression
12535 handles this. What it does is either leave the value of @code{arg} as
12536 it is, but only if @code{arg} is bound to a value; or it sets the
12537 value of @code{arg} to 1, in the case when @code{arg} is bound to
12540 Next is a @code{let}. That specifies the values of two local
12541 variables, @code{point} and @code{sentence-end}. The local value of
12542 point, from before the search, is used in the
12543 @code{constrain-to-field} function which handles forms and
12544 equivalents. The @code{sentence-end} variable is set by the
12545 @code{sentence-end} function.
12547 @node fwd-sentence while loops
12548 @unnumberedsubsec The @code{while} loops
12550 Two @code{while} loops follow. The first @code{while} has a
12551 true-or-false-test that tests true if the prefix argument for
12552 @code{forward-sentence} is a negative number. This is for going
12553 backwards. The body of this loop is similar to the body of the second
12554 @code{while} clause, but it is not exactly the same. We will skip
12555 this @code{while} loop and concentrate on the second @code{while}
12559 The second @code{while} loop is for moving point forward. Its skeleton
12564 (while (> arg 0) ; @r{true-or-false-test}
12566 (if (@var{true-or-false-test})
12569 (setq arg (1- arg)))) ; @code{while} @r{loop decrementer}
12573 The @code{while} loop is of the decrementing kind.
12574 (@xref{Decrementing Loop, , A Loop with a Decrementing Counter}.) It
12575 has a true-or-false-test that tests true so long as the counter (in
12576 this case, the variable @code{arg}) is greater than zero; and it has a
12577 decrementer that subtracts 1 from the value of the counter every time
12580 If no prefix argument is given to @code{forward-sentence}, which is
12581 the most common way the command is used, this @code{while} loop will
12582 run once, since the value of @code{arg} will be 1.
12584 The body of the @code{while} loop consists of a @code{let} expression,
12585 which creates and binds a local variable, and has, as its body, an
12586 @code{if} expression.
12589 The body of the @code{while} loop looks like this:
12594 (save-excursion (end-of-paragraph-text) (point))))
12595 (if (re-search-forward sentence-end par-end t)
12596 (skip-chars-backward " \t\n")
12597 (goto-char par-end)))
12601 The @code{let} expression creates and binds the local variable
12602 @code{par-end}. As we shall see, this local variable is designed to
12603 provide a bound or limit to the regular expression search. If the
12604 search fails to find a proper sentence ending in the paragraph, it will
12605 stop on reaching the end of the paragraph.
12607 But first, let us examine how @code{par-end} is bound to the value of
12608 the end of the paragraph. What happens is that the @code{let} sets the
12609 value of @code{par-end} to the value returned when the Lisp interpreter
12610 evaluates the expression
12614 (save-excursion (end-of-paragraph-text) (point))
12619 In this expression, @code{(end-of-paragraph-text)} moves point to the
12620 end of the paragraph, @code{(point)} returns the value of point, and then
12621 @code{save-excursion} restores point to its original position. Thus,
12622 the @code{let} binds @code{par-end} to the value returned by the
12623 @code{save-excursion} expression, which is the position of the end of
12624 the paragraph. (The @code{end-of-paragraph-text} function uses
12625 @code{forward-paragraph}, which we will discuss shortly.)
12628 Emacs next evaluates the body of the @code{let}, which is an @code{if}
12629 expression that looks like this:
12633 (if (re-search-forward sentence-end par-end t) ; @r{if-part}
12634 (skip-chars-backward " \t\n") ; @r{then-part}
12635 (goto-char par-end))) ; @r{else-part}
12639 The @code{if} tests whether its first argument is true and if so,
12640 evaluates its then-part; otherwise, the Emacs Lisp interpreter
12641 evaluates the else-part. The true-or-false-test of the @code{if}
12642 expression is the regular expression search.
12644 It may seem odd to have what looks like the `real work' of
12645 the @code{forward-sentence} function buried here, but this is a common
12646 way this kind of operation is carried out in Lisp.
12648 @node fwd-sentence re-search
12649 @unnumberedsubsec The regular expression search
12651 The @code{re-search-forward} function searches for the end of the
12652 sentence, that is, for the pattern defined by the @code{sentence-end}
12653 regular expression. If the pattern is found---if the end of the sentence is
12654 found---then the @code{re-search-forward} function does two things:
12658 The @code{re-search-forward} function carries out a side effect, which
12659 is to move point to the end of the occurrence found.
12662 The @code{re-search-forward} function returns a value of true. This is
12663 the value received by the @code{if}, and means that the search was
12668 The side effect, the movement of point, is completed before the
12669 @code{if} function is handed the value returned by the successful
12670 conclusion of the search.
12672 When the @code{if} function receives the value of true from a successful
12673 call to @code{re-search-forward}, the @code{if} evaluates the then-part,
12674 which is the expression @code{(skip-chars-backward " \t\n")}. This
12675 expression moves backwards over any blank spaces, tabs or carriage
12676 returns until a printed character is found and then leaves point after
12677 the character. Since point has already been moved to the end of the
12678 pattern that marks the end of the sentence, this action leaves point
12679 right after the closing printed character of the sentence, which is
12682 On the other hand, if the @code{re-search-forward} function fails to
12683 find a pattern marking the end of the sentence, the function returns
12684 false. The false then causes the @code{if} to evaluate its third
12685 argument, which is @code{(goto-char par-end)}: it moves point to the
12686 end of the paragraph.
12688 (And if the text is in a form or equivalent, and point may not move
12689 fully, then the @code{constrain-to-field} function comes into play.)
12691 Regular expression searches are exceptionally useful and the pattern
12692 illustrated by @code{re-search-forward}, in which the search is the
12693 test of an @code{if} expression, is handy. You will see or write code
12694 incorporating this pattern often.
12696 @node forward-paragraph
12697 @section @code{forward-paragraph}: a Goldmine of Functions
12698 @findex forward-paragraph
12702 (defun forward-paragraph (&optional arg)
12703 "Move forward to end of paragraph.
12704 With argument ARG, do it ARG times;
12705 a negative argument ARG = -N means move backward N paragraphs.
12707 A line which `paragraph-start' matches either separates paragraphs
12708 \(if `paragraph-separate' matches it also) or is the first line of a paragraph.
12709 A paragraph end is the beginning of a line which is not part of the paragraph
12710 to which the end of the previous line belongs, or the end of the buffer.
12711 Returns the count of paragraphs left to move."
12713 (or arg (setq arg 1))
12714 (let* ((opoint (point))
12715 (fill-prefix-regexp
12716 (and fill-prefix (not (equal fill-prefix ""))
12717 (not paragraph-ignore-fill-prefix)
12718 (regexp-quote fill-prefix)))
12719 ;; Remove ^ from paragraph-start and paragraph-sep if they are there.
12720 ;; These regexps shouldn't be anchored, because we look for them
12721 ;; starting at the left-margin. This allows paragraph commands to
12722 ;; work normally with indented text.
12723 ;; This hack will not find problem cases like "whatever\\|^something".
12724 (parstart (if (and (not (equal "" paragraph-start))
12725 (equal ?^ (aref paragraph-start 0)))
12726 (substring paragraph-start 1)
12728 (parsep (if (and (not (equal "" paragraph-separate))
12729 (equal ?^ (aref paragraph-separate 0)))
12730 (substring paragraph-separate 1)
12731 paragraph-separate))
12733 (if fill-prefix-regexp
12734 (concat parsep "\\|"
12735 fill-prefix-regexp "[ \t]*$")
12737 ;; This is used for searching.
12738 (sp-parstart (concat "^[ \t]*\\(?:" parstart "\\|" parsep "\\)"))
12740 (while (and (< arg 0) (not (bobp)))
12741 (if (and (not (looking-at parsep))
12742 (re-search-backward "^\n" (max (1- (point)) (point-min)) t)
12743 (looking-at parsep))
12744 (setq arg (1+ arg))
12745 (setq start (point))
12746 ;; Move back over paragraph-separating lines.
12747 (forward-char -1) (beginning-of-line)
12748 (while (and (not (bobp))
12749 (progn (move-to-left-margin)
12750 (looking-at parsep)))
12754 (setq arg (1+ arg))
12755 ;; Go to end of the previous (non-separating) line.
12757 ;; Search back for line that starts or separates paragraphs.
12758 (if (if fill-prefix-regexp
12759 ;; There is a fill prefix; it overrides parstart.
12760 (let (multiple-lines)
12761 (while (and (progn (beginning-of-line) (not (bobp)))
12762 (progn (move-to-left-margin)
12763 (not (looking-at parsep)))
12764 (looking-at fill-prefix-regexp))
12765 (unless (= (point) start)
12766 (setq multiple-lines t))
12768 (move-to-left-margin)
12769 ;; This deleted code caused a long hanging-indent line
12770 ;; not to be filled together with the following lines.
12771 ;; ;; Don't move back over a line before the paragraph
12772 ;; ;; which doesn't start with fill-prefix
12773 ;; ;; unless that is the only line we've moved over.
12774 ;; (and (not (looking-at fill-prefix-regexp))
12776 ;; (forward-line 1))
12778 (while (and (re-search-backward sp-parstart nil 1)
12779 (setq found-start t)
12780 ;; Found a candidate, but need to check if it is a
12782 (progn (setq start (point))
12783 (move-to-left-margin)
12784 (not (looking-at parsep)))
12785 (not (and (looking-at parstart)
12786 (or (not use-hard-newlines)
12789 (1- start) 'hard)))))
12790 (setq found-start nil)
12795 ;; Move forward over paragraph separators.
12796 ;; We know this cannot reach the place we started
12797 ;; because we know we moved back over a non-separator.
12798 (while (and (not (eobp))
12799 (progn (move-to-left-margin)
12800 (looking-at parsep)))
12802 ;; If line before paragraph is just margin, back up to there.
12804 (if (> (current-column) (current-left-margin))
12806 (skip-chars-backward " \t")
12808 (forward-line 1))))
12809 ;; No starter or separator line => use buffer beg.
12810 (goto-char (point-min))))))
12812 (while (and (> arg 0) (not (eobp)))
12813 ;; Move forward over separator lines...
12814 (while (and (not (eobp))
12815 (progn (move-to-left-margin) (not (eobp)))
12816 (looking-at parsep))
12818 (unless (eobp) (setq arg (1- arg)))
12819 ;; ... and one more line.
12821 (if fill-prefix-regexp
12822 ;; There is a fill prefix; it overrides parstart.
12823 (while (and (not (eobp))
12824 (progn (move-to-left-margin) (not (eobp)))
12825 (not (looking-at parsep))
12826 (looking-at fill-prefix-regexp))
12828 (while (and (re-search-forward sp-parstart nil 1)
12829 (progn (setq start (match-beginning 0))
12832 (progn (move-to-left-margin)
12833 (not (looking-at parsep)))
12834 (or (not (looking-at parstart))
12835 (and use-hard-newlines
12836 (not (get-text-property (1- start) 'hard)))))
12838 (if (< (point) (point-max))
12839 (goto-char start))))
12840 (constrain-to-field nil opoint t)
12841 ;; Return the number of steps that could not be done.
12845 The @code{forward-paragraph} function moves point forward to the end
12846 of the paragraph. It is usually bound to @kbd{M-@}} and makes use of a
12847 number of functions that are important in themselves, including
12848 @code{let*}, @code{match-beginning}, and @code{looking-at}.
12850 The function definition for @code{forward-paragraph} is considerably
12851 longer than the function definition for @code{forward-sentence}
12852 because it works with a paragraph, each line of which may begin with a
12855 A fill prefix consists of a string of characters that are repeated at
12856 the beginning of each line. For example, in Lisp code, it is a
12857 convention to start each line of a paragraph-long comment with
12858 @samp{;;; }. In Text mode, four blank spaces make up another common
12859 fill prefix, creating an indented paragraph. (@xref{Fill Prefix, , ,
12860 emacs, The GNU Emacs Manual}, for more information about fill
12863 The existence of a fill prefix means that in addition to being able to
12864 find the end of a paragraph whose lines begin on the left-most
12865 column, the @code{forward-paragraph} function must be able to find the
12866 end of a paragraph when all or many of the lines in the buffer begin
12867 with the fill prefix.
12869 Moreover, it is sometimes practical to ignore a fill prefix that
12870 exists, especially when blank lines separate paragraphs.
12871 This is an added complication.
12874 * forward-paragraph in brief:: Key parts of the function definition.
12875 * fwd-para let:: The @code{let*} expression.
12876 * fwd-para while:: The forward motion @code{while} loop.
12880 @node forward-paragraph in brief
12881 @unnumberedsubsec Shortened @code{forward-paragraph} function definition
12884 Rather than print all of the @code{forward-paragraph} function, we
12885 will only print parts of it. Read without preparation, the function
12889 In outline, the function looks like this:
12893 (defun forward-paragraph (&optional arg)
12894 "@var{documentation}@dots{}"
12896 (or arg (setq arg 1))
12899 (while (and (< arg 0) (not (bobp))) ; @r{backward-moving-code}
12901 (while (and (> arg 0) (not (eobp))) ; @r{forward-moving-code}
12906 The first parts of the function are routine: the function's argument
12907 list consists of one optional argument. Documentation follows.
12909 The lower case @samp{p} in the @code{interactive} declaration means
12910 that the processed prefix argument, if any, is passed to the function.
12911 This will be a number, and is the repeat count of how many paragraphs
12912 point will move. The @code{or} expression in the next line handles
12913 the common case when no argument is passed to the function, which occurs
12914 if the function is called from other code rather than interactively.
12915 This case was described earlier. (@xref{forward-sentence, The
12916 @code{forward-sentence} function}.) Now we reach the end of the
12917 familiar part of this function.
12920 @unnumberedsubsec The @code{let*} expression
12922 The next line of the @code{forward-paragraph} function begins a
12923 @code{let*} expression. This is a different than @code{let}. The
12924 symbol is @code{let*} not @code{let}.
12926 The @code{let*} special form is like @code{let} except that Emacs sets
12927 each variable in sequence, one after another, and variables in the
12928 latter part of the varlist can make use of the values to which Emacs
12929 set variables in the earlier part of the varlist.
12932 ( refappend save-excursion, , code save-excursion in code append-to-buffer .)
12935 (@ref{append save-excursion, , @code{save-excursion} in @code{append-to-buffer}}.)
12937 In the @code{let*} expression in this function, Emacs binds a total of
12938 seven variables: @code{opoint}, @code{fill-prefix-regexp},
12939 @code{parstart}, @code{parsep}, @code{sp-parstart}, @code{start}, and
12940 @code{found-start}.
12942 The variable @code{parsep} appears twice, first, to remove instances
12943 of @samp{^}, and second, to handle fill prefixes.
12945 The variable @code{opoint} is just the value of @code{point}. As you
12946 can guess, it is used in a @code{constrain-to-field} expression, just
12947 as in @code{forward-sentence}.
12949 The variable @code{fill-prefix-regexp} is set to the value returned by
12950 evaluating the following list:
12955 (not (equal fill-prefix ""))
12956 (not paragraph-ignore-fill-prefix)
12957 (regexp-quote fill-prefix))
12962 This is an expression whose first element is the @code{and} special form.
12964 As we learned earlier (@pxref{kill-new function, , The @code{kill-new}
12965 function}), the @code{and} special form evaluates each of its
12966 arguments until one of the arguments returns a value of @code{nil}, in
12967 which case the @code{and} expression returns @code{nil}; however, if
12968 none of the arguments returns a value of @code{nil}, the value
12969 resulting from evaluating the last argument is returned. (Since such
12970 a value is not @code{nil}, it is considered true in Lisp.) In other
12971 words, an @code{and} expression returns a true value only if all its
12972 arguments are true.
12975 In this case, the variable @code{fill-prefix-regexp} is bound to a
12976 non-@code{nil} value only if the following four expressions produce a
12977 true (i.e., a non-@code{nil}) value when they are evaluated; otherwise,
12978 @code{fill-prefix-regexp} is bound to @code{nil}.
12982 When this variable is evaluated, the value of the fill prefix, if any,
12983 is returned. If there is no fill prefix, this variable returns
12986 @item (not (equal fill-prefix "")
12987 This expression checks whether an existing fill prefix is an empty
12988 string, that is, a string with no characters in it. An empty string is
12989 not a useful fill prefix.
12991 @item (not paragraph-ignore-fill-prefix)
12992 This expression returns @code{nil} if the variable
12993 @code{paragraph-ignore-fill-prefix} has been turned on by being set to a
12994 true value such as @code{t}.
12996 @item (regexp-quote fill-prefix)
12997 This is the last argument to the @code{and} special form. If all the
12998 arguments to the @code{and} are true, the value resulting from
12999 evaluating this expression will be returned by the @code{and} expression
13000 and bound to the variable @code{fill-prefix-regexp},
13003 @findex regexp-quote
13005 The result of evaluating this @code{and} expression successfully is that
13006 @code{fill-prefix-regexp} will be bound to the value of
13007 @code{fill-prefix} as modified by the @code{regexp-quote} function.
13008 What @code{regexp-quote} does is read a string and return a regular
13009 expression that will exactly match the string and match nothing else.
13010 This means that @code{fill-prefix-regexp} will be set to a value that
13011 will exactly match the fill prefix if the fill prefix exists.
13012 Otherwise, the variable will be set to @code{nil}.
13014 The next two local variables in the @code{let*} expression are
13015 designed to remove instances of @samp{^} from @code{parstart} and
13016 @code{parsep}, the local variables which indicate the paragraph start
13017 and the paragraph separator. The next expression sets @code{parsep}
13018 again. That is to handle fill prefixes.
13020 This is the setting that requires the definition call @code{let*}
13021 rather than @code{let}. The true-or-false-test for the @code{if}
13022 depends on whether the variable @code{fill-prefix-regexp} evaluates to
13023 @code{nil} or some other value.
13025 If @code{fill-prefix-regexp} does not have a value, Emacs evaluates
13026 the else-part of the @code{if} expression and binds @code{parsep} to
13027 its local value. (@code{parsep} is a regular expression that matches
13028 what separates paragraphs.)
13030 But if @code{fill-prefix-regexp} does have a value, Emacs evaluates
13031 the then-part of the @code{if} expression and binds @code{parsep} to a
13032 regular expression that includes the @code{fill-prefix-regexp} as part
13035 Specifically, @code{parsep} is set to the original value of the
13036 paragraph separate regular expression concatenated with an alternative
13037 expression that consists of the @code{fill-prefix-regexp} followed by
13038 optional whitespace to the end of the line. The whitespace is defined
13039 by @w{@code{"[ \t]*$"}}.) The @samp{\\|} defines this portion of the
13040 regexp as an alternative to @code{parsep}.
13042 According to a comment in the code, the next local variable,
13043 @code{sp-parstart}, is used for searching, and then the final two,
13044 @code{start} and @code{found-start}, are set to @code{nil}.
13046 Now we get into the body of the @code{let*}. The first part of the body
13047 of the @code{let*} deals with the case when the function is given a
13048 negative argument and is therefore moving backwards. We will skip this
13051 @node fwd-para while
13052 @unnumberedsubsec The forward motion @code{while} loop
13054 The second part of the body of the @code{let*} deals with forward
13055 motion. It is a @code{while} loop that repeats itself so long as the
13056 value of @code{arg} is greater than zero. In the most common use of
13057 the function, the value of the argument is 1, so the body of the
13058 @code{while} loop is evaluated exactly once, and the cursor moves
13059 forward one paragraph.
13062 (while (and (> arg 0) (not (eobp)))
13064 ;; Move forward over separator lines...
13065 (while (and (not (eobp))
13066 (progn (move-to-left-margin) (not (eobp)))
13067 (looking-at parsep))
13069 (unless (eobp) (setq arg (1- arg)))
13070 ;; ... and one more line.
13073 (if fill-prefix-regexp
13074 ;; There is a fill prefix; it overrides parstart.
13075 (while (and (not (eobp))
13076 (progn (move-to-left-margin) (not (eobp)))
13077 (not (looking-at parsep))
13078 (looking-at fill-prefix-regexp))
13081 (while (and (re-search-forward sp-parstart nil 1)
13082 (progn (setq start (match-beginning 0))
13085 (progn (move-to-left-margin)
13086 (not (looking-at parsep)))
13087 (or (not (looking-at parstart))
13088 (and use-hard-newlines
13089 (not (get-text-property (1- start) 'hard)))))
13092 (if (< (point) (point-max))
13093 (goto-char start))))
13096 This part handles three situations: when point is between paragraphs,
13097 when there is a fill prefix and when there is no fill prefix.
13100 The @code{while} loop looks like this:
13104 ;; @r{going forwards and not at the end of the buffer}
13105 (while (and (> arg 0) (not (eobp)))
13107 ;; @r{between paragraphs}
13108 ;; Move forward over separator lines...
13109 (while (and (not (eobp))
13110 (progn (move-to-left-margin) (not (eobp)))
13111 (looking-at parsep))
13113 ;; @r{This decrements the loop}
13114 (unless (eobp) (setq arg (1- arg)))
13115 ;; ... and one more line.
13120 (if fill-prefix-regexp
13121 ;; There is a fill prefix; it overrides parstart;
13122 ;; we go forward line by line
13123 (while (and (not (eobp))
13124 (progn (move-to-left-margin) (not (eobp)))
13125 (not (looking-at parsep))
13126 (looking-at fill-prefix-regexp))
13131 ;; There is no fill prefix;
13132 ;; we go forward character by character
13133 (while (and (re-search-forward sp-parstart nil 1)
13134 (progn (setq start (match-beginning 0))
13137 (progn (move-to-left-margin)
13138 (not (looking-at parsep)))
13139 (or (not (looking-at parstart))
13140 (and use-hard-newlines
13141 (not (get-text-property (1- start) 'hard)))))
13146 ;; and if there is no fill prefix and if we are not at the end,
13147 ;; go to whatever was found in the regular expression search
13149 (if (< (point) (point-max))
13150 (goto-char start))))
13155 We can see that this is a decrementing counter @code{while} loop,
13156 using the expression @code{(setq arg (1- arg))} as the decrementer.
13157 That expression is not far from the @code{while}, but is hidden in
13158 another Lisp macro, an @code{unless} macro. Unless we are at the end
13159 of the buffer---that is what the @code{eobp} function determines; it
13160 is an abbreviation of @samp{End Of Buffer P}---we decrease the value
13161 of @code{arg} by one.
13163 (If we are at the end of the buffer, we cannot go forward any more and
13164 the next loop of the @code{while} expression will test false since the
13165 test is an @code{and} with @code{(not (eobp))}. The @code{not}
13166 function means exactly as you expect; it is another name for
13167 @code{null}, a function that returns true when its argument is false.)
13169 Interestingly, the loop count is not decremented until we leave the
13170 space between paragraphs, unless we come to the end of buffer or stop
13171 seeing the local value of the paragraph separator.
13173 That second @code{while} also has a @code{(move-to-left-margin)}
13174 expression. The function is self-explanatory. It is inside a
13175 @code{progn} expression and not the last element of its body, so it is
13176 only invoked for its side effect, which is to move point to the left
13177 margin of the current line.
13180 The @code{looking-at} function is also self-explanatory; it returns
13181 true if the text after point matches the regular expression given as
13184 The rest of the body of the loop looks difficult at first, but makes
13185 sense as you come to understand it.
13188 First consider what happens if there is a fill prefix:
13192 (if fill-prefix-regexp
13193 ;; There is a fill prefix; it overrides parstart;
13194 ;; we go forward line by line
13195 (while (and (not (eobp))
13196 (progn (move-to-left-margin) (not (eobp)))
13197 (not (looking-at parsep))
13198 (looking-at fill-prefix-regexp))
13204 This expression moves point forward line by line so long
13205 as four conditions are true:
13209 Point is not at the end of the buffer.
13212 We can move to the left margin of the text and are
13213 not at the end of the buffer.
13216 The text following point does not separate paragraphs.
13219 The pattern following point is the fill prefix regular expression.
13222 The last condition may be puzzling, until you remember that point was
13223 moved to the beginning of the line early in the @code{forward-paragraph}
13224 function. This means that if the text has a fill prefix, the
13225 @code{looking-at} function will see it.
13228 Consider what happens when there is no fill prefix.
13232 (while (and (re-search-forward sp-parstart nil 1)
13233 (progn (setq start (match-beginning 0))
13236 (progn (move-to-left-margin)
13237 (not (looking-at parsep)))
13238 (or (not (looking-at parstart))
13239 (and use-hard-newlines
13240 (not (get-text-property (1- start) 'hard)))))
13246 This @code{while} loop has us searching forward for
13247 @code{sp-parstart}, which is the combination of possible whitespace
13248 with a the local value of the start of a paragraph or of a paragraph
13249 separator. (The latter two are within an expression starting
13250 @code{\(?:} so that they are not referenced by the
13251 @code{match-beginning} function.)
13254 The two expressions,
13258 (setq start (match-beginning 0))
13264 mean go to the start of the text matched by the regular expression
13267 The @code{(match-beginning 0)} expression is new. It returns a number
13268 specifying the location of the start of the text that was matched by
13271 The @code{match-beginning} function is used here because of a
13272 characteristic of a forward search: a successful forward search,
13273 regardless of whether it is a plain search or a regular expression
13274 search, moves point to the end of the text that is found. In this
13275 case, a successful search moves point to the end of the pattern for
13276 @code{sp-parstart}.
13278 However, we want to put point at the end of the current paragraph, not
13279 somewhere else. Indeed, since the search possibly includes the
13280 paragraph separator, point may end up at the beginning of the next one
13281 unless we use an expression that includes @code{match-beginning}.
13283 @findex match-beginning
13284 When given an argument of 0, @code{match-beginning} returns the
13285 position that is the start of the text matched by the most recent
13286 search. In this case, the most recent search looks for
13287 @code{sp-parstart}. The @code{(match-beginning 0)} expression returns
13288 the beginning position of that pattern, rather than the end position
13291 (Incidentally, when passed a positive number as an argument, the
13292 @code{match-beginning} function returns the location of point at that
13293 parenthesized expression in the last search unless that parenthesized
13294 expression begins with @code{\(?:}. I don't know why @code{\(?:}
13295 appears here since the argument is 0.)
13298 The last expression when there is no fill prefix is
13302 (if (< (point) (point-max))
13303 (goto-char start))))
13308 This says that if there is no fill prefix and if we are not at the
13309 end, point should move to the beginning of whatever was found by the
13310 regular expression search for @code{sp-parstart}.
13312 The full definition for the @code{forward-paragraph} function not only
13313 includes code for going forwards, but also code for going backwards.
13315 If you are reading this inside of GNU Emacs and you want to see the
13316 whole function, you can type @kbd{C-h f} (@code{describe-function})
13317 and the name of the function. This gives you the function
13318 documentation and the name of the library containing the function's
13319 source. Place point over the name of the library and press the RET
13320 key; you will be taken directly to the source. (Be sure to install
13321 your sources! Without them, you are like a person who tries to drive
13322 a car with his eyes shut!)
13325 @section Create Your Own @file{TAGS} File
13327 @cindex @file{TAGS} file, create own
13329 Besides @kbd{C-h f} (@code{describe-function}), another way to see the
13330 source of a function is to type @kbd{M-.} (@code{find-tag}) and the
13331 name of the function when prompted for it. This is a good habit to
13332 get into. The @kbd{M-.} (@code{find-tag}) command takes you directly
13333 to the source for a function, variable, or node. The function depends
13334 on tags tables to tell it where to go.
13336 If the @code{find-tag} function first asks you for the name of a
13337 @file{TAGS} table, give it the name of a @file{TAGS} file such as
13338 @file{/usr/local/src/emacs/src/TAGS}. (The exact path to your
13339 @file{TAGS} file depends on how your copy of Emacs was installed. I
13340 just told you the location that provides both my C and my Emacs Lisp
13343 You can also create your own @file{TAGS} file for directories that
13346 You often need to build and install tags tables yourself. They are
13347 not built automatically. A tags table is called a @file{TAGS} file;
13348 the name is in upper case letters.
13350 You can create a @file{TAGS} file by calling the @code{etags} program
13351 that comes as a part of the Emacs distribution. Usually, @code{etags}
13352 is compiled and installed when Emacs is built. (@code{etags} is not
13353 an Emacs Lisp function or a part of Emacs; it is a C program.)
13356 To create a @file{TAGS} file, first switch to the directory in which
13357 you want to create the file. In Emacs you can do this with the
13358 @kbd{M-x cd} command, or by visiting a file in the directory, or by
13359 listing the directory with @kbd{C-x d} (@code{dired}). Then run the
13360 compile command, with @w{@code{etags *.el}} as the command to execute
13363 M-x compile RET etags *.el RET
13367 to create a @file{TAGS} file for Emacs Lisp.
13369 For example, if you have a large number of files in your
13370 @file{~/emacs} directory, as I do---I have 137 @file{.el} files in it,
13371 of which I load 12---you can create a @file{TAGS} file for the Emacs
13372 Lisp files in that directory.
13375 The @code{etags} program takes all the usual shell `wildcards'. For
13376 example, if you have two directories for which you want a single
13377 @file{TAGS} file, type @w{@code{etags *.el ../elisp/*.el}}, where
13378 @file{../elisp/} is the second directory:
13381 M-x compile RET etags *.el ../elisp/*.el RET
13388 M-x compile RET etags --help RET
13392 to see a list of the options accepted by @code{etags} as well as a
13393 list of supported languages.
13395 The @code{etags} program handles more than 20 languages, including
13396 Emacs Lisp, Common Lisp, Scheme, C, C++, Ada, Fortran, HTML, Java,
13397 LaTeX, Pascal, Perl, PostScript, Python, TeX, Texinfo, makefiles, and
13398 most assemblers. The program has no switches for specifying the
13399 language; it recognizes the language in an input file according to its
13400 file name and contents.
13402 @file{etags} is very helpful when you are writing code yourself and
13403 want to refer back to functions you have already written. Just run
13404 @code{etags} again at intervals as you write new functions, so they
13405 become part of the @file{TAGS} file.
13407 If you think an appropriate @file{TAGS} file already exists for what
13408 you want, but do not know where it is, you can use the @code{locate}
13409 program to attempt to find it.
13411 Type @w{@kbd{M-x locate @key{RET} TAGS @key{RET}}} and Emacs will list
13412 for you the full path names of all your @file{TAGS} files. On my
13413 system, this command lists 34 @file{TAGS} files. On the other hand, a
13414 `plain vanilla' system I recently installed did not contain any
13417 If the tags table you want has been created, you can use the @code{M-x
13418 visit-tags-table} command to specify it. Otherwise, you will need to
13419 create the tag table yourself and then use @code{M-x
13422 @subsubheading Building Tags in the Emacs sources
13423 @cindex Building Tags in the Emacs sources
13424 @cindex Tags in the Emacs sources
13427 The GNU Emacs sources come with a @file{Makefile} that contains a
13428 sophisticated @code{etags} command that creates, collects, and merges
13429 tags tables from all over the Emacs sources and puts the information
13430 into one @file{TAGS} file in the @file{src/} directory. (The
13431 @file{src/} directory is below the top level of your Emacs directory.)
13434 To build this @file{TAGS} file, go to the top level of your Emacs
13435 source directory and run the compile command @code{make tags}:
13438 M-x compile RET make tags RET
13442 (The @code{make tags} command works well with the GNU Emacs sources,
13443 as well as with some other source packages.)
13445 For more information, see @ref{Tags, , Tag Tables, emacs, The GNU Emacs
13448 @node Regexp Review
13451 Here is a brief summary of some recently introduced functions.
13455 Repeatedly evaluate the body of the expression so long as the first
13456 element of the body tests true. Then return @code{nil}. (The
13457 expression is evaluated only for its side effects.)
13466 (insert (format "foo is %d.\n" foo))
13467 (setq foo (1- foo))))
13469 @result{} foo is 2.
13476 (The @code{insert} function inserts its arguments at point; the
13477 @code{format} function returns a string formatted from its arguments
13478 the way @code{message} formats its arguments; @code{\n} produces a new
13481 @item re-search-forward
13482 Search for a pattern, and if the pattern is found, move point to rest
13486 Takes four arguments, like @code{search-forward}:
13490 A regular expression that specifies the pattern to search for.
13491 (Remember to put quotation marks around this argument!)
13494 Optionally, the limit of the search.
13497 Optionally, what to do if the search fails, return @code{nil} or an
13501 Optionally, how many times to repeat the search; if negative, the
13502 search goes backwards.
13506 Bind some variables locally to particular values,
13507 and then evaluate the remaining arguments, returning the value of the
13508 last one. While binding the local variables, use the local values of
13509 variables bound earlier, if any.
13518 (message "`bar' is %d." bar))
13519 @result{} `bar' is 21.
13523 @item match-beginning
13524 Return the position of the start of the text found by the last regular
13528 Return @code{t} for true if the text after point matches the argument,
13529 which should be a regular expression.
13532 Return @code{t} for true if point is at the end of the accessible part
13533 of a buffer. The end of the accessible part is the end of the buffer
13534 if the buffer is not narrowed; it is the end of the narrowed part if
13535 the buffer is narrowed.
13539 @node re-search Exercises
13540 @section Exercises with @code{re-search-forward}
13544 Write a function to search for a regular expression that matches two
13545 or more blank lines in sequence.
13548 Write a function to search for duplicated words, such as `the the'.
13549 @xref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs
13550 Manual}, for information on how to write a regexp (a regular
13551 expression) to match a string that is composed of two identical
13552 halves. You can devise several regexps; some are better than others.
13553 The function I use is described in an appendix, along with several
13554 regexps. @xref{the-the, , @code{the-the} Duplicated Words Function}.
13557 @node Counting Words
13558 @chapter Counting: Repetition and Regexps
13559 @cindex Repetition for word counting
13560 @cindex Regular expressions for word counting
13562 Repetition and regular expression searches are powerful tools that you
13563 often use when you write code in Emacs Lisp. This chapter illustrates
13564 the use of regular expression searches through the construction of
13565 word count commands using @code{while} loops and recursion.
13568 * Why Count Words::
13569 * @value{COUNT-WORDS}:: Use a regexp, but find a problem.
13570 * recursive-count-words:: Start with case of no words in region.
13571 * Counting Exercise::
13575 @node Why Count Words
13576 @unnumberedsec Counting words
13579 The standard Emacs distribution contains functions for counting the
13580 number of lines and words within a region.
13582 Certain types of writing ask you to count words. Thus, if you write
13583 an essay, you may be limited to 800 words; if you write a novel, you
13584 may discipline yourself to write 1000 words a day. It seems odd, but
13585 for a long time, Emacs lacked a word count command. Perhaps people used
13586 Emacs mostly for code or types of documentation that did not require
13587 word counts; or perhaps they restricted themselves to the operating
13588 system word count command, @code{wc}. Alternatively, people may have
13589 followed the publishers' convention and computed a word count by
13590 dividing the number of characters in a document by five.
13592 There are many ways to implement a command to count words. Here are
13593 some examples, which you may wish to compare with the standard Emacs
13594 command, @code{count-words-region}.
13596 @node @value{COUNT-WORDS}
13597 @section The @code{@value{COUNT-WORDS}} Function
13598 @findex @value{COUNT-WORDS}
13600 A word count command could count words in a line, paragraph, region,
13601 or buffer. What should the command cover? You could design the
13602 command to count the number of words in a complete buffer. However,
13603 the Emacs tradition encourages flexibility---you may want to count
13604 words in just a section, rather than all of a buffer. So it makes
13605 more sense to design the command to count the number of words in a
13606 region. Once you have a command to count words in a region, you can,
13607 if you wish, count words in a whole buffer by marking it with
13608 @w{@kbd{C-x h}} (@code{mark-whole-buffer}).
13610 Clearly, counting words is a repetitive act: starting from the
13611 beginning of the region, you count the first word, then the second
13612 word, then the third word, and so on, until you reach the end of the
13613 region. This means that word counting is ideally suited to recursion
13614 or to a @code{while} loop.
13617 * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop.
13618 * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}.
13622 @node Design @value{COUNT-WORDS}
13623 @unnumberedsubsec Designing @code{@value{COUNT-WORDS}}
13626 First, we will implement the word count command with a @code{while}
13627 loop, then with recursion. The command will, of course, be
13631 The template for an interactive function definition is, as always:
13635 (defun @var{name-of-function} (@var{argument-list})
13636 "@var{documentation}@dots{}"
13637 (@var{interactive-expression}@dots{})
13642 What we need to do is fill in the slots.
13644 The name of the function should be self-explanatory and similar to the
13645 existing @code{count-lines-region} name. This makes the name easier
13646 to remember. @code{count-words-region} is the obvious choice. Since
13647 that name is now used for the standard Emacs command to count words, we
13648 will name our implementation @code{@value{COUNT-WORDS}}.
13650 The function counts words within a region. This means that the
13651 argument list must contain symbols that are bound to the two
13652 positions, the beginning and end of the region. These two positions
13653 can be called @samp{beginning} and @samp{end} respectively. The first
13654 line of the documentation should be a single sentence, since that is
13655 all that is printed as documentation by a command such as
13656 @code{apropos}. The interactive expression will be of the form
13657 @samp{(interactive "r")}, since that will cause Emacs to pass the
13658 beginning and end of the region to the function's argument list. All
13661 The body of the function needs to be written to do three tasks:
13662 first, to set up conditions under which the @code{while} loop can
13663 count words, second, to run the @code{while} loop, and third, to send
13664 a message to the user.
13666 When a user calls @code{@value{COUNT-WORDS}}, point may be at the
13667 beginning or the end of the region. However, the counting process
13668 must start at the beginning of the region. This means we will want
13669 to put point there if it is not already there. Executing
13670 @code{(goto-char beginning)} ensures this. Of course, we will want to
13671 return point to its expected position when the function finishes its
13672 work. For this reason, the body must be enclosed in a
13673 @code{save-excursion} expression.
13675 The central part of the body of the function consists of a
13676 @code{while} loop in which one expression jumps point forward word by
13677 word, and another expression counts those jumps. The true-or-false-test
13678 of the @code{while} loop should test true so long as point should jump
13679 forward, and false when point is at the end of the region.
13681 We could use @code{(forward-word 1)} as the expression for moving point
13682 forward word by word, but it is easier to see what Emacs identifies as a
13683 `word' if we use a regular expression search.
13685 A regular expression search that finds the pattern for which it is
13686 searching leaves point after the last character matched. This means
13687 that a succession of successful word searches will move point forward
13690 As a practical matter, we want the regular expression search to jump
13691 over whitespace and punctuation between words as well as over the
13692 words themselves. A regexp that refuses to jump over interword
13693 whitespace would never jump more than one word! This means that
13694 the regexp should include the whitespace and punctuation that follows
13695 a word, if any, as well as the word itself. (A word may end a buffer
13696 and not have any following whitespace or punctuation, so that part of
13697 the regexp must be optional.)
13699 Thus, what we want for the regexp is a pattern defining one or more
13700 word constituent characters followed, optionally, by one or more
13701 characters that are not word constituents. The regular expression for
13709 The buffer's syntax table determines which characters are and are not
13710 word constituents. For more information about syntax,
13711 @pxref{Syntax Tables, , Syntax Tables, elisp, The GNU Emacs Lisp
13715 The search expression looks like this:
13718 (re-search-forward "\\w+\\W*")
13722 (Note that paired backslashes precede the @samp{w} and @samp{W}. A
13723 single backslash has special meaning to the Emacs Lisp interpreter.
13724 It indicates that the following character is interpreted differently
13725 than usual. For example, the two characters, @samp{\n}, stand for
13726 @samp{newline}, rather than for a backslash followed by @samp{n}. Two
13727 backslashes in a row stand for an ordinary, `unspecial' backslash, so
13728 Emacs Lisp interpreter ends of seeing a single backslash followed by a
13729 letter. So it discovers the letter is special.)
13731 We need a counter to count how many words there are; this variable
13732 must first be set to 0 and then incremented each time Emacs goes
13733 around the @code{while} loop. The incrementing expression is simply:
13736 (setq count (1+ count))
13739 Finally, we want to tell the user how many words there are in the
13740 region. The @code{message} function is intended for presenting this
13741 kind of information to the user. The message has to be phrased so
13742 that it reads properly regardless of how many words there are in the
13743 region: we don't want to say that ``there are 1 words in the region''.
13744 The conflict between singular and plural is ungrammatical. We can
13745 solve this problem by using a conditional expression that evaluates
13746 different messages depending on the number of words in the region.
13747 There are three possibilities: no words in the region, one word in the
13748 region, and more than one word. This means that the @code{cond}
13749 special form is appropriate.
13752 All this leads to the following function definition:
13756 ;;; @r{First version; has bugs!}
13757 (defun @value{COUNT-WORDS} (beginning end)
13758 "Print number of words in the region.
13759 Words are defined as at least one word-constituent
13760 character followed by at least one character that
13761 is not a word-constituent. The buffer's syntax
13762 table determines which characters these are."
13764 (message "Counting words in region ... ")
13768 ;;; @r{1. Set up appropriate conditions.}
13770 (goto-char beginning)
13775 ;;; @r{2. Run the} while @r{loop.}
13776 (while (< (point) end)
13777 (re-search-forward "\\w+\\W*")
13778 (setq count (1+ count)))
13782 ;;; @r{3. Send a message to the user.}
13783 (cond ((zerop count)
13785 "The region does NOT have any words."))
13788 "The region has 1 word."))
13791 "The region has %d words." count))))))
13796 As written, the function works, but not in all circumstances.
13798 @node Whitespace Bug
13799 @subsection The Whitespace Bug in @code{@value{COUNT-WORDS}}
13801 The @code{@value{COUNT-WORDS}} command described in the preceding
13802 section has two bugs, or rather, one bug with two manifestations.
13803 First, if you mark a region containing only whitespace in the middle
13804 of some text, the @code{@value{COUNT-WORDS}} command tells you that the
13805 region contains one word! Second, if you mark a region containing
13806 only whitespace at the end of the buffer or the accessible portion of
13807 a narrowed buffer, the command displays an error message that looks
13811 Search failed: "\\w+\\W*"
13814 If you are reading this in Info in GNU Emacs, you can test for these
13817 First, evaluate the function in the usual manner to install it.
13819 Here is a copy of the definition. Place your cursor after the closing
13820 parenthesis and type @kbd{C-x C-e} to install it.
13824 ;; @r{First version; has bugs!}
13825 (defun @value{COUNT-WORDS} (beginning end)
13826 "Print number of words in the region.
13827 Words are defined as at least one word-constituent character followed
13828 by at least one character that is not a word-constituent. The buffer's
13829 syntax table determines which characters these are."
13833 (message "Counting words in region ... ")
13837 ;;; @r{1. Set up appropriate conditions.}
13839 (goto-char beginning)
13844 ;;; @r{2. Run the} while @r{loop.}
13845 (while (< (point) end)
13846 (re-search-forward "\\w+\\W*")
13847 (setq count (1+ count)))
13851 ;;; @r{3. Send a message to the user.}
13852 (cond ((zerop count)
13853 (message "The region does NOT have any words."))
13854 ((= 1 count) (message "The region has 1 word."))
13855 (t (message "The region has %d words." count))))))
13861 If you wish, you can also install this keybinding by evaluating it:
13864 (global-set-key "\C-c=" '@value{COUNT-WORDS})
13867 To conduct the first test, set mark and point to the beginning and end
13868 of the following line and then type @kbd{C-c =} (or @kbd{M-x
13869 @value{COUNT-WORDS}} if you have not bound @kbd{C-c =}):
13876 Emacs will tell you, correctly, that the region has three words.
13878 Repeat the test, but place mark at the beginning of the line and place
13879 point just @emph{before} the word @samp{one}. Again type the command
13880 @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}). Emacs should tell you
13881 that the region has no words, since it is composed only of the
13882 whitespace at the beginning of the line. But instead Emacs tells you
13883 that the region has one word!
13885 For the third test, copy the sample line to the end of the
13886 @file{*scratch*} buffer and then type several spaces at the end of the
13887 line. Place mark right after the word @samp{three} and point at the
13888 end of line. (The end of the line will be the end of the buffer.)
13889 Type @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}) as you did before.
13890 Again, Emacs should tell you that the region has no words, since it is
13891 composed only of the whitespace at the end of the line. Instead,
13892 Emacs displays an error message saying @samp{Search failed}.
13894 The two bugs stem from the same problem.
13896 Consider the first manifestation of the bug, in which the command
13897 tells you that the whitespace at the beginning of the line contains
13898 one word. What happens is this: The @code{M-x @value{COUNT-WORDS}}
13899 command moves point to the beginning of the region. The @code{while}
13900 tests whether the value of point is smaller than the value of
13901 @code{end}, which it is. Consequently, the regular expression search
13902 looks for and finds the first word. It leaves point after the word.
13903 @code{count} is set to one. The @code{while} loop repeats; but this
13904 time the value of point is larger than the value of @code{end}, the
13905 loop is exited; and the function displays a message saying the number
13906 of words in the region is one. In brief, the regular expression
13907 search looks for and finds the word even though it is outside
13910 In the second manifestation of the bug, the region is whitespace at
13911 the end of the buffer. Emacs says @samp{Search failed}. What happens
13912 is that the true-or-false-test in the @code{while} loop tests true, so
13913 the search expression is executed. But since there are no more words
13914 in the buffer, the search fails.
13916 In both manifestations of the bug, the search extends or attempts to
13917 extend outside of the region.
13919 The solution is to limit the search to the region---this is a fairly
13920 simple action, but as you may have come to expect, it is not quite as
13921 simple as you might think.
13923 As we have seen, the @code{re-search-forward} function takes a search
13924 pattern as its first argument. But in addition to this first,
13925 mandatory argument, it accepts three optional arguments. The optional
13926 second argument bounds the search. The optional third argument, if
13927 @code{t}, causes the function to return @code{nil} rather than signal
13928 an error if the search fails. The optional fourth argument is a
13929 repeat count. (In Emacs, you can see a function's documentation by
13930 typing @kbd{C-h f}, the name of the function, and then @key{RET}.)
13932 In the @code{@value{COUNT-WORDS}} definition, the value of the end of
13933 the region is held by the variable @code{end} which is passed as an
13934 argument to the function. Thus, we can add @code{end} as an argument
13935 to the regular expression search expression:
13938 (re-search-forward "\\w+\\W*" end)
13941 However, if you make only this change to the @code{@value{COUNT-WORDS}}
13942 definition and then test the new version of the definition on a
13943 stretch of whitespace, you will receive an error message saying
13944 @samp{Search failed}.
13946 What happens is this: the search is limited to the region, and fails
13947 as you expect because there are no word-constituent characters in the
13948 region. Since it fails, we receive an error message. But we do not
13949 want to receive an error message in this case; we want to receive the
13950 message that "The region does NOT have any words."
13952 The solution to this problem is to provide @code{re-search-forward}
13953 with a third argument of @code{t}, which causes the function to return
13954 @code{nil} rather than signal an error if the search fails.
13956 However, if you make this change and try it, you will see the message
13957 ``Counting words in region ... '' and @dots{} you will keep on seeing
13958 that message @dots{}, until you type @kbd{C-g} (@code{keyboard-quit}).
13960 Here is what happens: the search is limited to the region, as before,
13961 and it fails because there are no word-constituent characters in the
13962 region, as expected. Consequently, the @code{re-search-forward}
13963 expression returns @code{nil}. It does nothing else. In particular,
13964 it does not move point, which it does as a side effect if it finds the
13965 search target. After the @code{re-search-forward} expression returns
13966 @code{nil}, the next expression in the @code{while} loop is evaluated.
13967 This expression increments the count. Then the loop repeats. The
13968 true-or-false-test tests true because the value of point is still less
13969 than the value of end, since the @code{re-search-forward} expression
13970 did not move point. @dots{} and the cycle repeats @dots{}
13972 The @code{@value{COUNT-WORDS}} definition requires yet another
13973 modification, to cause the true-or-false-test of the @code{while} loop
13974 to test false if the search fails. Put another way, there are two
13975 conditions that must be satisfied in the true-or-false-test before the
13976 word count variable is incremented: point must still be within the
13977 region and the search expression must have found a word to count.
13979 Since both the first condition and the second condition must be true
13980 together, the two expressions, the region test and the search
13981 expression, can be joined with an @code{and} special form and embedded in
13982 the @code{while} loop as the true-or-false-test, like this:
13985 (and (< (point) end) (re-search-forward "\\w+\\W*" end t))
13988 @c colon in printed section title causes problem in Info cross reference
13989 @c also trouble with an overfull hbox
13992 (For information about @code{and}, see
13993 @ref{kill-new function, , The @code{kill-new} function}.)
13997 (@xref{kill-new function, , The @code{kill-new} function}, for
13998 information about @code{and}.)
14001 The @code{re-search-forward} expression returns @code{t} if the search
14002 succeeds and as a side effect moves point. Consequently, as words are
14003 found, point is moved through the region. When the search expression
14004 fails to find another word, or when point reaches the end of the
14005 region, the true-or-false-test tests false, the @code{while} loop
14006 exits, and the @code{@value{COUNT-WORDS}} function displays one or
14007 other of its messages.
14009 After incorporating these final changes, the @code{@value{COUNT-WORDS}}
14010 works without bugs (or at least, without bugs that I have found!).
14011 Here is what it looks like:
14015 ;;; @r{Final version:} @code{while}
14016 (defun @value{COUNT-WORDS} (beginning end)
14017 "Print number of words in the region."
14019 (message "Counting words in region ... ")
14023 ;;; @r{1. Set up appropriate conditions.}
14026 (goto-char beginning)
14030 ;;; @r{2. Run the} while @r{loop.}
14031 (while (and (< (point) end)
14032 (re-search-forward "\\w+\\W*" end t))
14033 (setq count (1+ count)))
14037 ;;; @r{3. Send a message to the user.}
14038 (cond ((zerop count)
14040 "The region does NOT have any words."))
14043 "The region has 1 word."))
14046 "The region has %d words." count))))))
14050 @node recursive-count-words
14051 @section Count Words Recursively
14052 @cindex Count words recursively
14053 @cindex Recursively counting words
14054 @cindex Words, counted recursively
14056 You can write the function for counting words recursively as well as
14057 with a @code{while} loop. Let's see how this is done.
14059 First, we need to recognize that the @code{@value{COUNT-WORDS}}
14060 function has three jobs: it sets up the appropriate conditions for
14061 counting to occur; it counts the words in the region; and it sends a
14062 message to the user telling how many words there are.
14064 If we write a single recursive function to do everything, we will
14065 receive a message for every recursive call. If the region contains 13
14066 words, we will receive thirteen messages, one right after the other.
14067 We don't want this! Instead, we must write two functions to do the
14068 job, one of which (the recursive function) will be used inside of the
14069 other. One function will set up the conditions and display the
14070 message; the other will return the word count.
14072 Let us start with the function that causes the message to be displayed.
14073 We can continue to call this @code{@value{COUNT-WORDS}}.
14075 This is the function that the user will call. It will be interactive.
14076 Indeed, it will be similar to our previous versions of this
14077 function, except that it will call @code{recursive-count-words} to
14078 determine how many words are in the region.
14081 We can readily construct a template for this function, based on our
14086 ;; @r{Recursive version; uses regular expression search}
14087 (defun @value{COUNT-WORDS} (beginning end)
14088 "@var{documentation}@dots{}"
14089 (@var{interactive-expression}@dots{})
14093 ;;; @r{1. Set up appropriate conditions.}
14094 (@var{explanatory message})
14095 (@var{set-up functions}@dots{}
14099 ;;; @r{2. Count the words.}
14100 @var{recursive call}
14104 ;;; @r{3. Send a message to the user.}
14105 @var{message providing word count}))
14109 The definition looks straightforward, except that somehow the count
14110 returned by the recursive call must be passed to the message
14111 displaying the word count. A little thought suggests that this can be
14112 done by making use of a @code{let} expression: we can bind a variable
14113 in the varlist of a @code{let} expression to the number of words in
14114 the region, as returned by the recursive call; and then the
14115 @code{cond} expression, using binding, can display the value to the
14118 Often, one thinks of the binding within a @code{let} expression as
14119 somehow secondary to the `primary' work of a function. But in this
14120 case, what you might consider the `primary' job of the function,
14121 counting words, is done within the @code{let} expression.
14124 Using @code{let}, the function definition looks like this:
14128 (defun @value{COUNT-WORDS} (beginning end)
14129 "Print number of words in the region."
14134 ;;; @r{1. Set up appropriate conditions.}
14135 (message "Counting words in region ... ")
14137 (goto-char beginning)
14141 ;;; @r{2. Count the words.}
14142 (let ((count (recursive-count-words end)))
14146 ;;; @r{3. Send a message to the user.}
14147 (cond ((zerop count)
14149 "The region does NOT have any words."))
14152 "The region has 1 word."))
14155 "The region has %d words." count))))))
14159 Next, we need to write the recursive counting function.
14161 A recursive function has at least three parts: the `do-again-test', the
14162 `next-step-expression', and the recursive call.
14164 The do-again-test determines whether the function will or will not be
14165 called again. Since we are counting words in a region and can use a
14166 function that moves point forward for every word, the do-again-test
14167 can check whether point is still within the region. The do-again-test
14168 should find the value of point and determine whether point is before,
14169 at, or after the value of the end of the region. We can use the
14170 @code{point} function to locate point. Clearly, we must pass the
14171 value of the end of the region to the recursive counting function as an
14174 In addition, the do-again-test should also test whether the search finds a
14175 word. If it does not, the function should not call itself again.
14177 The next-step-expression changes a value so that when the recursive
14178 function is supposed to stop calling itself, it stops. More
14179 precisely, the next-step-expression changes a value so that at the
14180 right time, the do-again-test stops the recursive function from
14181 calling itself again. In this case, the next-step-expression can be
14182 the expression that moves point forward, word by word.
14184 The third part of a recursive function is the recursive call.
14186 Somewhere, also, we also need a part that does the `work' of the
14187 function, a part that does the counting. A vital part!
14190 But already, we have an outline of the recursive counting function:
14194 (defun recursive-count-words (region-end)
14195 "@var{documentation}@dots{}"
14196 @var{do-again-test}
14197 @var{next-step-expression}
14198 @var{recursive call})
14202 Now we need to fill in the slots. Let's start with the simplest cases
14203 first: if point is at or beyond the end of the region, there cannot
14204 be any words in the region, so the function should return zero.
14205 Likewise, if the search fails, there are no words to count, so the
14206 function should return zero.
14208 On the other hand, if point is within the region and the search
14209 succeeds, the function should call itself again.
14212 Thus, the do-again-test should look like this:
14216 (and (< (point) region-end)
14217 (re-search-forward "\\w+\\W*" region-end t))
14221 Note that the search expression is part of the do-again-test---the
14222 function returns @code{t} if its search succeeds and @code{nil} if it
14223 fails. (@xref{Whitespace Bug, , The Whitespace Bug in
14224 @code{@value{COUNT-WORDS}}}, for an explanation of how
14225 @code{re-search-forward} works.)
14227 The do-again-test is the true-or-false test of an @code{if} clause.
14228 Clearly, if the do-again-test succeeds, the then-part of the @code{if}
14229 clause should call the function again; but if it fails, the else-part
14230 should return zero since either point is outside the region or the
14231 search failed because there were no words to find.
14233 But before considering the recursive call, we need to consider the
14234 next-step-expression. What is it? Interestingly, it is the search
14235 part of the do-again-test.
14237 In addition to returning @code{t} or @code{nil} for the
14238 do-again-test, @code{re-search-forward} moves point forward as a side
14239 effect of a successful search. This is the action that changes the
14240 value of point so that the recursive function stops calling itself
14241 when point completes its movement through the region. Consequently,
14242 the @code{re-search-forward} expression is the next-step-expression.
14245 In outline, then, the body of the @code{recursive-count-words}
14246 function looks like this:
14250 (if @var{do-again-test-and-next-step-combined}
14252 @var{recursive-call-returning-count}
14258 How to incorporate the mechanism that counts?
14260 If you are not used to writing recursive functions, a question like
14261 this can be troublesome. But it can and should be approached
14264 We know that the counting mechanism should be associated in some way
14265 with the recursive call. Indeed, since the next-step-expression moves
14266 point forward by one word, and since a recursive call is made for
14267 each word, the counting mechanism must be an expression that adds one
14268 to the value returned by a call to @code{recursive-count-words}.
14271 Consider several cases:
14275 If there are two words in the region, the function should return
14276 a value resulting from adding one to the value returned when it counts
14277 the first word, plus the number returned when it counts the remaining
14278 words in the region, which in this case is one.
14281 If there is one word in the region, the function should return
14282 a value resulting from adding one to the value returned when it counts
14283 that word, plus the number returned when it counts the remaining
14284 words in the region, which in this case is zero.
14287 If there are no words in the region, the function should return zero.
14290 From the sketch we can see that the else-part of the @code{if} returns
14291 zero for the case of no words. This means that the then-part of the
14292 @code{if} must return a value resulting from adding one to the value
14293 returned from a count of the remaining words.
14296 The expression will look like this, where @code{1+} is a function that
14297 adds one to its argument.
14300 (1+ (recursive-count-words region-end))
14304 The whole @code{recursive-count-words} function will then look like
14309 (defun recursive-count-words (region-end)
14310 "@var{documentation}@dots{}"
14312 ;;; @r{1. do-again-test}
14313 (if (and (< (point) region-end)
14314 (re-search-forward "\\w+\\W*" region-end t))
14318 ;;; @r{2. then-part: the recursive call}
14319 (1+ (recursive-count-words region-end))
14321 ;;; @r{3. else-part}
14327 Let's examine how this works:
14329 If there are no words in the region, the else part of the @code{if}
14330 expression is evaluated and consequently the function returns zero.
14332 If there is one word in the region, the value of point is less than
14333 the value of @code{region-end} and the search succeeds. In this case,
14334 the true-or-false-test of the @code{if} expression tests true, and the
14335 then-part of the @code{if} expression is evaluated. The counting
14336 expression is evaluated. This expression returns a value (which will
14337 be the value returned by the whole function) that is the sum of one
14338 added to the value returned by a recursive call.
14340 Meanwhile, the next-step-expression has caused point to jump over the
14341 first (and in this case only) word in the region. This means that
14342 when @code{(recursive-count-words region-end)} is evaluated a second
14343 time, as a result of the recursive call, the value of point will be
14344 equal to or greater than the value of region end. So this time,
14345 @code{recursive-count-words} will return zero. The zero will be added
14346 to one, and the original evaluation of @code{recursive-count-words}
14347 will return one plus zero, which is one, which is the correct amount.
14349 Clearly, if there are two words in the region, the first call to
14350 @code{recursive-count-words} returns one added to the value returned
14351 by calling @code{recursive-count-words} on a region containing the
14352 remaining word---that is, it adds one to one, producing two, which is
14353 the correct amount.
14355 Similarly, if there are three words in the region, the first call to
14356 @code{recursive-count-words} returns one added to the value returned
14357 by calling @code{recursive-count-words} on a region containing the
14358 remaining two words---and so on and so on.
14362 With full documentation the two functions look like this:
14366 The recursive function:
14368 @findex recursive-count-words
14371 (defun recursive-count-words (region-end)
14372 "Number of words between point and REGION-END."
14376 ;;; @r{1. do-again-test}
14377 (if (and (< (point) region-end)
14378 (re-search-forward "\\w+\\W*" region-end t))
14382 ;;; @r{2. then-part: the recursive call}
14383 (1+ (recursive-count-words region-end))
14385 ;;; @r{3. else-part}
14396 ;;; @r{Recursive version}
14397 (defun @value{COUNT-WORDS} (beginning end)
14398 "Print number of words in the region.
14402 Words are defined as at least one word-constituent
14403 character followed by at least one character that is
14404 not a word-constituent. The buffer's syntax table
14405 determines which characters these are."
14409 (message "Counting words in region ... ")
14411 (goto-char beginning)
14412 (let ((count (recursive-count-words end)))
14415 (cond ((zerop count)
14417 "The region does NOT have any words."))
14421 (message "The region has 1 word."))
14424 "The region has %d words." count))))))
14428 @node Counting Exercise
14429 @section Exercise: Counting Punctuation
14431 Using a @code{while} loop, write a function to count the number of
14432 punctuation marks in a region---period, comma, semicolon, colon,
14433 exclamation mark, and question mark. Do the same using recursion.
14435 @node Words in a defun
14436 @chapter Counting Words in a @code{defun}
14437 @cindex Counting words in a @code{defun}
14438 @cindex Word counting in a @code{defun}
14440 Our next project is to count the number of words in a function
14441 definition. Clearly, this can be done using some variant of
14442 @code{@value{COUNT-WORDS}}. @xref{Counting Words, , Counting Words:
14443 Repetition and Regexps}. If we are just going to count the words in
14444 one definition, it is easy enough to mark the definition with the
14445 @kbd{C-M-h} (@code{mark-defun}) command, and then call
14446 @code{@value{COUNT-WORDS}}.
14448 However, I am more ambitious: I want to count the words and symbols in
14449 every definition in the Emacs sources and then print a graph that
14450 shows how many functions there are of each length: how many contain 40
14451 to 49 words or symbols, how many contain 50 to 59 words or symbols,
14452 and so on. I have often been curious how long a typical function is,
14453 and this will tell.
14456 * Divide and Conquer::
14457 * Words and Symbols:: What to count?
14458 * Syntax:: What constitutes a word or symbol?
14459 * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}.
14460 * Several defuns:: Counting several defuns in a file.
14461 * Find a File:: Do you want to look at a file?
14462 * lengths-list-file:: A list of the lengths of many definitions.
14463 * Several files:: Counting in definitions in different files.
14464 * Several files recursively:: Recursively counting in different files.
14465 * Prepare the data:: Prepare the data for display in a graph.
14469 @node Divide and Conquer
14470 @unnumberedsec Divide and Conquer
14473 Described in one phrase, the histogram project is daunting; but
14474 divided into numerous small steps, each of which we can take one at a
14475 time, the project becomes less fearsome. Let us consider what the
14480 First, write a function to count the words in one definition. This
14481 includes the problem of handling symbols as well as words.
14484 Second, write a function to list the numbers of words in each function
14485 in a file. This function can use the @code{count-words-in-defun}
14489 Third, write a function to list the numbers of words in each function
14490 in each of several files. This entails automatically finding the
14491 various files, switching to them, and counting the words in the
14492 definitions within them.
14495 Fourth, write a function to convert the list of numbers that we
14496 created in step three to a form that will be suitable for printing as
14500 Fifth, write a function to print the results as a graph.
14503 This is quite a project! But if we take each step slowly, it will not
14506 @node Words and Symbols
14507 @section What to Count?
14508 @cindex Words and symbols in defun
14510 When we first start thinking about how to count the words in a
14511 function definition, the first question is (or ought to be) what are
14512 we going to count? When we speak of `words' with respect to a Lisp
14513 function definition, we are actually speaking, in large part, of
14514 `symbols'. For example, the following @code{multiply-by-seven}
14515 function contains the five symbols @code{defun},
14516 @code{multiply-by-seven}, @code{number}, @code{*}, and @code{7}. In
14517 addition, in the documentation string, it contains the four words
14518 @samp{Multiply}, @samp{NUMBER}, @samp{by}, and @samp{seven}. The
14519 symbol @samp{number} is repeated, so the definition contains a total
14520 of ten words and symbols.
14524 (defun multiply-by-seven (number)
14525 "Multiply NUMBER by seven."
14531 However, if we mark the @code{multiply-by-seven} definition with
14532 @kbd{C-M-h} (@code{mark-defun}), and then call
14533 @code{@value{COUNT-WORDS}} on it, we will find that
14534 @code{@value{COUNT-WORDS}} claims the definition has eleven words, not
14535 ten! Something is wrong!
14537 The problem is twofold: @code{@value{COUNT-WORDS}} does not count the
14538 @samp{*} as a word, and it counts the single symbol,
14539 @code{multiply-by-seven}, as containing three words. The hyphens are
14540 treated as if they were interword spaces rather than intraword
14541 connectors: @samp{multiply-by-seven} is counted as if it were written
14542 @samp{multiply by seven}.
14544 The cause of this confusion is the regular expression search within
14545 the @code{@value{COUNT-WORDS}} definition that moves point forward word
14546 by word. In the canonical version of @code{@value{COUNT-WORDS}}, the
14554 This regular expression is a pattern defining one or more word
14555 constituent characters possibly followed by one or more characters
14556 that are not word constituents. What is meant by `word constituent
14557 characters' brings us to the issue of syntax, which is worth a section
14561 @section What Constitutes a Word or Symbol?
14562 @cindex Syntax categories and tables
14564 Emacs treats different characters as belonging to different
14565 @dfn{syntax categories}. For example, the regular expression,
14566 @samp{\\w+}, is a pattern specifying one or more @emph{word
14567 constituent} characters. Word constituent characters are members of
14568 one syntax category. Other syntax categories include the class of
14569 punctuation characters, such as the period and the comma, and the
14570 class of whitespace characters, such as the blank space and the tab
14571 character. (For more information, @pxref{Syntax Tables, , Syntax
14572 Tables, elisp, The GNU Emacs Lisp Reference Manual}.)
14574 Syntax tables specify which characters belong to which categories.
14575 Usually, a hyphen is not specified as a `word constituent character'.
14576 Instead, it is specified as being in the `class of characters that are
14577 part of symbol names but not words.' This means that the
14578 @code{@value{COUNT-WORDS}} function treats it in the same way it treats
14579 an interword white space, which is why @code{@value{COUNT-WORDS}}
14580 counts @samp{multiply-by-seven} as three words.
14582 There are two ways to cause Emacs to count @samp{multiply-by-seven} as
14583 one symbol: modify the syntax table or modify the regular expression.
14585 We could redefine a hyphen as a word constituent character by
14586 modifying the syntax table that Emacs keeps for each mode. This
14587 action would serve our purpose, except that a hyphen is merely the
14588 most common character within symbols that is not typically a word
14589 constituent character; there are others, too.
14591 Alternatively, we can redefine the regexp used in the
14592 @code{@value{COUNT-WORDS}} definition so as to include symbols. This
14593 procedure has the merit of clarity, but the task is a little tricky.
14596 The first part is simple enough: the pattern must match ``at least one
14597 character that is a word or symbol constituent''. Thus:
14600 "\\(\\w\\|\\s_\\)+"
14604 The @samp{\\(} is the first part of the grouping construct that
14605 includes the @samp{\\w} and the @samp{\\s_} as alternatives, separated
14606 by the @samp{\\|}. The @samp{\\w} matches any word-constituent
14607 character and the @samp{\\s_} matches any character that is part of a
14608 symbol name but not a word-constituent character. The @samp{+}
14609 following the group indicates that the word or symbol constituent
14610 characters must be matched at least once.
14612 However, the second part of the regexp is more difficult to design.
14613 What we want is to follow the first part with ``optionally one or more
14614 characters that are not constituents of a word or symbol''. At first,
14615 I thought I could define this with the following:
14618 "\\(\\W\\|\\S_\\)*"
14622 The upper case @samp{W} and @samp{S} match characters that are
14623 @emph{not} word or symbol constituents. Unfortunately, this
14624 expression matches any character that is either not a word constituent
14625 or not a symbol constituent. This matches any character!
14627 I then noticed that every word or symbol in my test region was
14628 followed by white space (blank space, tab, or newline). So I tried
14629 placing a pattern to match one or more blank spaces after the pattern
14630 for one or more word or symbol constituents. This failed, too. Words
14631 and symbols are often separated by whitespace, but in actual code
14632 parentheses may follow symbols and punctuation may follow words. So
14633 finally, I designed a pattern in which the word or symbol constituents
14634 are followed optionally by characters that are not white space and
14635 then followed optionally by white space.
14638 Here is the full regular expression:
14641 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
14644 @node count-words-in-defun
14645 @section The @code{count-words-in-defun} Function
14646 @cindex Counting words in a @code{defun}
14648 We have seen that there are several ways to write a
14649 @code{count-words-region} function. To write a
14650 @code{count-words-in-defun}, we need merely adapt one of these
14653 The version that uses a @code{while} loop is easy to understand, so I
14654 am going to adapt that. Because @code{count-words-in-defun} will be
14655 part of a more complex program, it need not be interactive and it need
14656 not display a message but just return the count. These considerations
14657 simplify the definition a little.
14659 On the other hand, @code{count-words-in-defun} will be used within a
14660 buffer that contains function definitions. Consequently, it is
14661 reasonable to ask that the function determine whether it is called
14662 when point is within a function definition, and if it is, to return
14663 the count for that definition. This adds complexity to the
14664 definition, but saves us from needing to pass arguments to the
14668 These considerations lead us to prepare the following template:
14672 (defun count-words-in-defun ()
14673 "@var{documentation}@dots{}"
14674 (@var{set up}@dots{}
14675 (@var{while loop}@dots{})
14676 @var{return count})
14681 As usual, our job is to fill in the slots.
14685 We are presuming that this function will be called within a buffer
14686 containing function definitions. Point will either be within a
14687 function definition or not. For @code{count-words-in-defun} to work,
14688 point must move to the beginning of the definition, a counter must
14689 start at zero, and the counting loop must stop when point reaches the
14690 end of the definition.
14692 The @code{beginning-of-defun} function searches backwards for an
14693 opening delimiter such as a @samp{(} at the beginning of a line, and
14694 moves point to that position, or else to the limit of the search. In
14695 practice, this means that @code{beginning-of-defun} moves point to the
14696 beginning of an enclosing or preceding function definition, or else to
14697 the beginning of the buffer. We can use @code{beginning-of-defun} to
14698 place point where we wish to start.
14700 The @code{while} loop requires a counter to keep track of the words or
14701 symbols being counted. A @code{let} expression can be used to create
14702 a local variable for this purpose, and bind it to an initial value of zero.
14704 The @code{end-of-defun} function works like @code{beginning-of-defun}
14705 except that it moves point to the end of the definition.
14706 @code{end-of-defun} can be used as part of an expression that
14707 determines the position of the end of the definition.
14709 The set up for @code{count-words-in-defun} takes shape rapidly: first
14710 we move point to the beginning of the definition, then we create a
14711 local variable to hold the count, and finally, we record the position
14712 of the end of the definition so the @code{while} loop will know when to stop
14716 The code looks like this:
14720 (beginning-of-defun)
14722 (end (save-excursion (end-of-defun) (point))))
14727 The code is simple. The only slight complication is likely to concern
14728 @code{end}: it is bound to the position of the end of the definition
14729 by a @code{save-excursion} expression that returns the value of point
14730 after @code{end-of-defun} temporarily moves it to the end of the
14733 The second part of the @code{count-words-in-defun}, after the set up,
14734 is the @code{while} loop.
14736 The loop must contain an expression that jumps point forward word by
14737 word and symbol by symbol, and another expression that counts the
14738 jumps. The true-or-false-test for the @code{while} loop should test
14739 true so long as point should jump forward, and false when point is at
14740 the end of the definition. We have already redefined the regular
14741 expression for this, so the loop is straightforward:
14745 (while (and (< (point) end)
14747 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*" end t))
14748 (setq count (1+ count)))
14752 The third part of the function definition returns the count of words
14753 and symbols. This part is the last expression within the body of the
14754 @code{let} expression, and can be, very simply, the local variable
14755 @code{count}, which when evaluated returns the count.
14758 Put together, the @code{count-words-in-defun} definition looks like this:
14760 @findex count-words-in-defun
14763 (defun count-words-in-defun ()
14764 "Return the number of words and symbols in a defun."
14765 (beginning-of-defun)
14767 (end (save-excursion (end-of-defun) (point))))
14771 (and (< (point) end)
14773 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
14775 (setq count (1+ count)))
14780 How to test this? The function is not interactive, but it is easy to
14781 put a wrapper around the function to make it interactive; we can use
14782 almost the same code as for the recursive version of
14783 @code{@value{COUNT-WORDS}}:
14787 ;;; @r{Interactive version.}
14788 (defun count-words-defun ()
14789 "Number of words and symbols in a function definition."
14792 "Counting words and symbols in function definition ... ")
14795 (let ((count (count-words-in-defun)))
14799 "The definition does NOT have any words or symbols."))
14804 "The definition has 1 word or symbol."))
14807 "The definition has %d words or symbols." count)))))
14813 Let's re-use @kbd{C-c =} as a convenient keybinding:
14816 (global-set-key "\C-c=" 'count-words-defun)
14819 Now we can try out @code{count-words-defun}: install both
14820 @code{count-words-in-defun} and @code{count-words-defun}, and set the
14821 keybinding, and then place the cursor within the following definition:
14825 (defun multiply-by-seven (number)
14826 "Multiply NUMBER by seven."
14833 Success! The definition has 10 words and symbols.
14835 The next problem is to count the numbers of words and symbols in
14836 several definitions within a single file.
14838 @node Several defuns
14839 @section Count Several @code{defuns} Within a File
14841 A file such as @file{simple.el} may have a hundred or more function
14842 definitions within it. Our long term goal is to collect statistics on
14843 many files, but as a first step, our immediate goal is to collect
14844 statistics on one file.
14846 The information will be a series of numbers, each number being the
14847 length of a function definition. We can store the numbers in a list.
14849 We know that we will want to incorporate the information regarding one
14850 file with information about many other files; this means that the
14851 function for counting definition lengths within one file need only
14852 return the list of lengths. It need not and should not display any
14855 The word count commands contain one expression to jump point forward
14856 word by word and another expression to count the jumps. The function
14857 to return the lengths of definitions can be designed to work the same
14858 way, with one expression to jump point forward definition by
14859 definition and another expression to construct the lengths' list.
14861 This statement of the problem makes it elementary to write the
14862 function definition. Clearly, we will start the count at the
14863 beginning of the file, so the first command will be @code{(goto-char
14864 (point-min))}. Next, we start the @code{while} loop; and the
14865 true-or-false test of the loop can be a regular expression search for
14866 the next function definition---so long as the search succeeds, point
14867 is moved forward and then the body of the loop is evaluated. The body
14868 needs an expression that constructs the lengths' list. @code{cons},
14869 the list construction command, can be used to create the list. That
14870 is almost all there is to it.
14873 Here is what this fragment of code looks like:
14877 (goto-char (point-min))
14878 (while (re-search-forward "^(defun" nil t)
14880 (cons (count-words-in-defun) lengths-list)))
14884 What we have left out is the mechanism for finding the file that
14885 contains the function definitions.
14887 In previous examples, we either used this, the Info file, or we
14888 switched back and forth to some other buffer, such as the
14889 @file{*scratch*} buffer.
14891 Finding a file is a new process that we have not yet discussed.
14894 @section Find a File
14895 @cindex Find a File
14897 To find a file in Emacs, you use the @kbd{C-x C-f} (@code{find-file})
14898 command. This command is almost, but not quite right for the lengths
14902 Let's look at the source for @code{find-file}:
14906 (defun find-file (filename)
14907 "Edit file FILENAME.
14908 Switch to a buffer visiting file FILENAME,
14909 creating one if none already exists."
14910 (interactive "FFind file: ")
14911 (switch-to-buffer (find-file-noselect filename)))
14916 (The most recent version of the @code{find-file} function definition
14917 permits you to specify optional wildcards to visit multiple files; that
14918 makes the definition more complex and we will not discuss it here,
14919 since it is not relevant. You can see its source using either
14920 @kbd{M-.} (@code{find-tag}) or @kbd{C-h f} (@code{describe-function}).)
14924 (defun find-file (filename &optional wildcards)
14925 "Edit file FILENAME.
14926 Switch to a buffer visiting file FILENAME,
14927 creating one if none already exists.
14928 Interactively, the default if you just type RET is the current directory,
14929 but the visited file name is available through the minibuffer history:
14930 type M-n to pull it into the minibuffer.
14932 Interactively, or if WILDCARDS is non-nil in a call from Lisp,
14933 expand wildcards (if any) and visit multiple files. You can
14934 suppress wildcard expansion by setting `find-file-wildcards' to nil.
14936 To visit a file without any kind of conversion and without
14937 automatically choosing a major mode, use \\[find-file-literally]."
14938 (interactive (find-file-read-args "Find file: " nil))
14939 (let ((value (find-file-noselect filename nil nil wildcards)))
14941 (mapcar 'switch-to-buffer (nreverse value))
14942 (switch-to-buffer value))))
14945 The definition I am showing possesses short but complete documentation
14946 and an interactive specification that prompts you for a file name when
14947 you use the command interactively. The body of the definition
14948 contains two functions, @code{find-file-noselect} and
14949 @code{switch-to-buffer}.
14951 According to its documentation as shown by @kbd{C-h f} (the
14952 @code{describe-function} command), the @code{find-file-noselect}
14953 function reads the named file into a buffer and returns the buffer.
14954 (Its most recent version includes an optional wildcards argument,
14955 too, as well as another to read a file literally and an other you
14956 suppress warning messages. These optional arguments are irrelevant.)
14958 However, the @code{find-file-noselect} function does not select the
14959 buffer in which it puts the file. Emacs does not switch its attention
14960 (or yours if you are using @code{find-file-noselect}) to the selected
14961 buffer. That is what @code{switch-to-buffer} does: it switches the
14962 buffer to which Emacs attention is directed; and it switches the
14963 buffer displayed in the window to the new buffer. We have discussed
14964 buffer switching elsewhere. (@xref{Switching Buffers}.)
14966 In this histogram project, we do not need to display each file on the
14967 screen as the program determines the length of each definition within
14968 it. Instead of employing @code{switch-to-buffer}, we can work with
14969 @code{set-buffer}, which redirects the attention of the computer
14970 program to a different buffer but does not redisplay it on the screen.
14971 So instead of calling on @code{find-file} to do the job, we must write
14972 our own expression.
14974 The task is easy: use @code{find-file-noselect} and @code{set-buffer}.
14976 @node lengths-list-file
14977 @section @code{lengths-list-file} in Detail
14979 The core of the @code{lengths-list-file} function is a @code{while}
14980 loop containing a function to move point forward `defun by defun' and
14981 a function to count the number of words and symbols in each defun.
14982 This core must be surrounded by functions that do various other tasks,
14983 including finding the file, and ensuring that point starts out at the
14984 beginning of the file. The function definition looks like this:
14985 @findex lengths-list-file
14989 (defun lengths-list-file (filename)
14990 "Return list of definitions' lengths within FILE.
14991 The returned list is a list of numbers.
14992 Each number is the number of words or
14993 symbols in one function definition."
14996 (message "Working on `%s' ... " filename)
14998 (let ((buffer (find-file-noselect filename))
15000 (set-buffer buffer)
15001 (setq buffer-read-only t)
15003 (goto-char (point-min))
15004 (while (re-search-forward "^(defun" nil t)
15006 (cons (count-words-in-defun) lengths-list)))
15007 (kill-buffer buffer)
15013 The function is passed one argument, the name of the file on which it
15014 will work. It has four lines of documentation, but no interactive
15015 specification. Since people worry that a computer is broken if they
15016 don't see anything going on, the first line of the body is a
15019 The next line contains a @code{save-excursion} that returns Emacs's
15020 attention to the current buffer when the function completes. This is
15021 useful in case you embed this function in another function that
15022 presumes point is restored to the original buffer.
15024 In the varlist of the @code{let} expression, Emacs finds the file and
15025 binds the local variable @code{buffer} to the buffer containing the
15026 file. At the same time, Emacs creates @code{lengths-list} as a local
15029 Next, Emacs switches its attention to the buffer.
15031 In the following line, Emacs makes the buffer read-only. Ideally,
15032 this line is not necessary. None of the functions for counting words
15033 and symbols in a function definition should change the buffer.
15034 Besides, the buffer is not going to be saved, even if it were changed.
15035 This line is entirely the consequence of great, perhaps excessive,
15036 caution. The reason for the caution is that this function and those
15037 it calls work on the sources for Emacs and it is inconvenient if they
15038 are inadvertently modified. It goes without saying that I did not
15039 realize a need for this line until an experiment went awry and started
15040 to modify my Emacs source files @dots{}
15042 Next comes a call to widen the buffer if it is narrowed. This
15043 function is usually not needed---Emacs creates a fresh buffer if none
15044 already exists; but if a buffer visiting the file already exists Emacs
15045 returns that one. In this case, the buffer may be narrowed and must
15046 be widened. If we wanted to be fully `user-friendly', we would
15047 arrange to save the restriction and the location of point, but we
15050 The @code{(goto-char (point-min))} expression moves point to the
15051 beginning of the buffer.
15053 Then comes a @code{while} loop in which the `work' of the function is
15054 carried out. In the loop, Emacs determines the length of each
15055 definition and constructs a lengths' list containing the information.
15057 Emacs kills the buffer after working through it. This is to save
15058 space inside of Emacs. My version of GNU Emacs 19 contained over 300
15059 source files of interest; GNU Emacs 22 contains over a thousand source
15060 files. Another function will apply @code{lengths-list-file} to each
15063 Finally, the last expression within the @code{let} expression is the
15064 @code{lengths-list} variable; its value is returned as the value of
15065 the whole function.
15067 You can try this function by installing it in the usual fashion. Then
15068 place your cursor after the following expression and type @kbd{C-x
15069 C-e} (@code{eval-last-sexp}).
15071 @c !!! 22.1.1 lisp sources location here
15074 "/usr/local/share/emacs/22.1.1/lisp/emacs-lisp/debug.el")
15078 (You may need to change the pathname of the file; the one here is for
15079 GNU Emacs version 22.1.1. To change the expression, copy it to
15080 the @file{*scratch*} buffer and edit it.
15084 (Also, to see the full length of the list, rather than a truncated
15085 version, you may have to evaluate the following:
15088 (custom-set-variables '(eval-expression-print-length nil))
15092 (@xref{defcustom, , Specifying Variables using @code{defcustom}}.
15093 Then evaluate the @code{lengths-list-file} expression.)
15096 The lengths' list for @file{debug.el} takes less than a second to
15097 produce and looks like this in GNU Emacs 22:
15100 (83 113 105 144 289 22 30 97 48 89 25 52 52 88 28 29 77 49 43 290 232 587)
15104 (Using my old machine, the version 19 lengths' list for @file{debug.el}
15105 took seven seconds to produce and looked like this:
15108 (75 41 80 62 20 45 44 68 45 12 34 235)
15111 (The newer version of @file{debug.el} contains more defuns than the
15112 earlier one; and my new machine is much faster than the old one.)
15114 Note that the length of the last definition in the file is first in
15117 @node Several files
15118 @section Count Words in @code{defuns} in Different Files
15120 In the previous section, we created a function that returns a list of
15121 the lengths of each definition in a file. Now, we want to define a
15122 function to return a master list of the lengths of the definitions in
15125 Working on each of a list of files is a repetitious act, so we can use
15126 either a @code{while} loop or recursion.
15129 * lengths-list-many-files:: Return a list of the lengths of defuns.
15130 * append:: Attach one list to another.
15134 @node lengths-list-many-files
15135 @unnumberedsubsec Determine the lengths of @code{defuns}
15138 The design using a @code{while} loop is routine. The argument passed
15139 the function is a list of files. As we saw earlier (@pxref{Loop
15140 Example}), you can write a @code{while} loop so that the body of the
15141 loop is evaluated if such a list contains elements, but to exit the
15142 loop if the list is empty. For this design to work, the body of the
15143 loop must contain an expression that shortens the list each time the
15144 body is evaluated, so that eventually the list is empty. The usual
15145 technique is to set the value of the list to the value of the @sc{cdr}
15146 of the list each time the body is evaluated.
15149 The template looks like this:
15153 (while @var{test-whether-list-is-empty}
15155 @var{set-list-to-cdr-of-list})
15159 Also, we remember that a @code{while} loop returns @code{nil} (the
15160 result of evaluating the true-or-false-test), not the result of any
15161 evaluation within its body. (The evaluations within the body of the
15162 loop are done for their side effects.) However, the expression that
15163 sets the lengths' list is part of the body---and that is the value
15164 that we want returned by the function as a whole. To do this, we
15165 enclose the @code{while} loop within a @code{let} expression, and
15166 arrange that the last element of the @code{let} expression contains
15167 the value of the lengths' list. (@xref{Incrementing Example, , Loop
15168 Example with an Incrementing Counter}.)
15170 @findex lengths-list-many-files
15172 These considerations lead us directly to the function itself:
15176 ;;; @r{Use @code{while} loop.}
15177 (defun lengths-list-many-files (list-of-files)
15178 "Return list of lengths of defuns in LIST-OF-FILES."
15181 (let (lengths-list)
15183 ;;; @r{true-or-false-test}
15184 (while list-of-files
15189 ;;; @r{Generate a lengths' list.}
15191 (expand-file-name (car list-of-files)))))
15195 ;;; @r{Make files' list shorter.}
15196 (setq list-of-files (cdr list-of-files)))
15198 ;;; @r{Return final value of lengths' list.}
15203 @code{expand-file-name} is a built-in function that converts a file
15204 name to the absolute, long, path name form. The function employs the
15205 name of the directory in which the function is called.
15207 @c !!! 22.1.1 lisp sources location here
15209 Thus, if @code{expand-file-name} is called on @code{debug.el} when
15210 Emacs is visiting the
15211 @file{/usr/local/share/emacs/22.1.1/lisp/emacs-lisp/} directory,
15221 @c !!! 22.1.1 lisp sources location here
15223 /usr/local/share/emacs/22.1.1/lisp/emacs-lisp/debug.el
15226 The only other new element of this function definition is the as yet
15227 unstudied function @code{append}, which merits a short section for
15231 @subsection The @code{append} Function
15234 The @code{append} function attaches one list to another. Thus,
15237 (append '(1 2 3 4) '(5 6 7 8))
15248 This is exactly how we want to attach two lengths' lists produced by
15249 @code{lengths-list-file} to each other. The results contrast with
15253 (cons '(1 2 3 4) '(5 6 7 8))
15258 which constructs a new list in which the first argument to @code{cons}
15259 becomes the first element of the new list:
15262 ((1 2 3 4) 5 6 7 8)
15265 @node Several files recursively
15266 @section Recursively Count Words in Different Files
15268 Besides a @code{while} loop, you can work on each of a list of files
15269 with recursion. A recursive version of @code{lengths-list-many-files}
15270 is short and simple.
15272 The recursive function has the usual parts: the `do-again-test', the
15273 `next-step-expression', and the recursive call. The `do-again-test'
15274 determines whether the function should call itself again, which it
15275 will do if the @code{list-of-files} contains any remaining elements;
15276 the `next-step-expression' resets the @code{list-of-files} to the
15277 @sc{cdr} of itself, so eventually the list will be empty; and the
15278 recursive call calls itself on the shorter list. The complete
15279 function is shorter than this description!
15280 @findex recursive-lengths-list-many-files
15284 (defun recursive-lengths-list-many-files (list-of-files)
15285 "Return list of lengths of each defun in LIST-OF-FILES."
15286 (if list-of-files ; @r{do-again-test}
15289 (expand-file-name (car list-of-files)))
15290 (recursive-lengths-list-many-files
15291 (cdr list-of-files)))))
15296 In a sentence, the function returns the lengths' list for the first of
15297 the @code{list-of-files} appended to the result of calling itself on
15298 the rest of the @code{list-of-files}.
15300 Here is a test of @code{recursive-lengths-list-many-files}, along with
15301 the results of running @code{lengths-list-file} on each of the files
15304 Install @code{recursive-lengths-list-many-files} and
15305 @code{lengths-list-file}, if necessary, and then evaluate the
15306 following expressions. You may need to change the files' pathnames;
15307 those here work when this Info file and the Emacs sources are located
15308 in their customary places. To change the expressions, copy them to
15309 the @file{*scratch*} buffer, edit them, and then evaluate them.
15311 The results are shown after the @samp{@result{}}. (These results are
15312 for files from Emacs version 22.1.1; files from other versions of
15313 Emacs may produce different results.)
15315 @c !!! 22.1.1 lisp sources location here
15318 (cd "/usr/local/share/emacs/22.1.1/")
15320 (lengths-list-file "./lisp/macros.el")
15321 @result{} (283 263 480 90)
15325 (lengths-list-file "./lisp/mail/mailalias.el")
15326 @result{} (38 32 29 95 178 180 321 218 324)
15330 (lengths-list-file "./lisp/makesum.el")
15335 (recursive-lengths-list-many-files
15336 '("./lisp/macros.el"
15337 "./lisp/mail/mailalias.el"
15338 "./lisp/makesum.el"))
15339 @result{} (283 263 480 90 38 32 29 95 178 180 321 218 324 85 181)
15343 The @code{recursive-lengths-list-many-files} function produces the
15346 The next step is to prepare the data in the list for display in a graph.
15348 @node Prepare the data
15349 @section Prepare the Data for Display in a Graph
15351 The @code{recursive-lengths-list-many-files} function returns a list
15352 of numbers. Each number records the length of a function definition.
15353 What we need to do now is transform this data into a list of numbers
15354 suitable for generating a graph. The new list will tell how many
15355 functions definitions contain less than 10 words and
15356 symbols, how many contain between 10 and 19 words and symbols, how
15357 many contain between 20 and 29 words and symbols, and so on.
15359 In brief, we need to go through the lengths' list produced by the
15360 @code{recursive-lengths-list-many-files} function and count the number
15361 of defuns within each range of lengths, and produce a list of those
15365 * Data for Display in Detail::
15366 * Sorting:: Sorting lists.
15367 * Files List:: Making a list of files.
15368 * Counting function definitions::
15372 @node Data for Display in Detail
15373 @unnumberedsubsec The Data for Display in Detail
15376 Based on what we have done before, we can readily foresee that it
15377 should not be too hard to write a function that `@sc{cdr}s' down the
15378 lengths' list, looks at each element, determines which length range it
15379 is in, and increments a counter for that range.
15381 However, before beginning to write such a function, we should consider
15382 the advantages of sorting the lengths' list first, so the numbers are
15383 ordered from smallest to largest. First, sorting will make it easier
15384 to count the numbers in each range, since two adjacent numbers will
15385 either be in the same length range or in adjacent ranges. Second, by
15386 inspecting a sorted list, we can discover the highest and lowest
15387 number, and thereby determine the largest and smallest length range
15391 @subsection Sorting Lists
15394 Emacs contains a function to sort lists, called (as you might guess)
15395 @code{sort}. The @code{sort} function takes two arguments, the list
15396 to be sorted, and a predicate that determines whether the first of
15397 two list elements is ``less'' than the second.
15399 As we saw earlier (@pxref{Wrong Type of Argument, , Using the Wrong
15400 Type Object as an Argument}), a predicate is a function that
15401 determines whether some property is true or false. The @code{sort}
15402 function will reorder a list according to whatever property the
15403 predicate uses; this means that @code{sort} can be used to sort
15404 non-numeric lists by non-numeric criteria---it can, for example,
15405 alphabetize a list.
15408 The @code{<} function is used when sorting a numeric list. For example,
15411 (sort '(4 8 21 17 33 7 21 7) '<)
15419 (4 7 7 8 17 21 21 33)
15423 (Note that in this example, both the arguments are quoted so that the
15424 symbols are not evaluated before being passed to @code{sort} as
15427 Sorting the list returned by the
15428 @code{recursive-lengths-list-many-files} function is straightforward;
15429 it uses the @code{<} function:
15433 In GNU Emacs 22, eval
15435 (cd "/usr/local/share/emacs/22.0.50/")
15437 (recursive-lengths-list-many-files
15438 '("./lisp/macros.el"
15439 "./lisp/mail/mailalias.el"
15440 "./lisp/makesum.el"))
15448 (recursive-lengths-list-many-files
15449 '("./lisp/macros.el"
15450 "./lisp/mailalias.el"
15451 "./lisp/makesum.el"))
15461 (29 32 38 85 90 95 178 180 181 218 263 283 321 324 480)
15465 (Note that in this example, the first argument to @code{sort} is not
15466 quoted, since the expression must be evaluated so as to produce the
15467 list that is passed to @code{sort}.)
15470 @subsection Making a List of Files
15472 The @code{recursive-lengths-list-many-files} function requires a list
15473 of files as its argument. For our test examples, we constructed such
15474 a list by hand; but the Emacs Lisp source directory is too large for
15475 us to do for that. Instead, we will write a function to do the job
15476 for us. In this function, we will use both a @code{while} loop and a
15479 @findex directory-files
15480 We did not have to write a function like this for older versions of
15481 GNU Emacs, since they placed all the @samp{.el} files in one
15482 directory. Instead, we were able to use the @code{directory-files}
15483 function, which lists the names of files that match a specified
15484 pattern within a single directory.
15486 However, recent versions of Emacs place Emacs Lisp files in
15487 sub-directories of the top level @file{lisp} directory. This
15488 re-arrangement eases navigation. For example, all the mail related
15489 files are in a @file{lisp} sub-directory called @file{mail}. But at
15490 the same time, this arrangement forces us to create a file listing
15491 function that descends into the sub-directories.
15493 @findex files-in-below-directory
15494 We can create this function, called @code{files-in-below-directory},
15495 using familiar functions such as @code{car}, @code{nthcdr}, and
15496 @code{substring} in conjunction with an existing function called
15497 @code{directory-files-and-attributes}. This latter function not only
15498 lists all the filenames in a directory, including the names
15499 of sub-directories, but also their attributes.
15501 To restate our goal: to create a function that will enable us
15502 to feed filenames to @code{recursive-lengths-list-many-files}
15503 as a list that looks like this (but with more elements):
15507 ("./lisp/macros.el"
15508 "./lisp/mail/rmail.el"
15509 "./lisp/makesum.el")
15513 The @code{directory-files-and-attributes} function returns a list of
15514 lists. Each of the lists within the main list consists of 13
15515 elements. The first element is a string that contains the name of the
15516 file---which, in GNU/Linux, may be a `directory file', that is to
15517 say, a file with the special attributes of a directory. The second
15518 element of the list is @code{t} for a directory, a string
15519 for symbolic link (the string is the name linked to), or @code{nil}.
15521 For example, the first @samp{.el} file in the @file{lisp/} directory
15522 is @file{abbrev.el}. Its name is
15523 @file{/usr/local/share/emacs/22.1.1/lisp/abbrev.el} and it is not a
15524 directory or a symbolic link.
15527 This is how @code{directory-files-and-attributes} lists that file and
15539 (20615 27034 579989 697000)
15541 (20615 26327 734791 805000)
15553 On the other hand, @file{mail/} is a directory within the @file{lisp/}
15554 directory. The beginning of its listing looks like this:
15565 (To learn about the different attributes, look at the documentation of
15566 @code{file-attributes}. Bear in mind that the @code{file-attributes}
15567 function does not list the filename, so its first element is
15568 @code{directory-files-and-attributes}'s second element.)
15570 We will want our new function, @code{files-in-below-directory}, to
15571 list the @samp{.el} files in the directory it is told to check, and in
15572 any directories below that directory.
15574 This gives us a hint on how to construct
15575 @code{files-in-below-directory}: within a directory, the function
15576 should add @samp{.el} filenames to a list; and if, within a directory,
15577 the function comes upon a sub-directory, it should go into that
15578 sub-directory and repeat its actions.
15580 However, we should note that every directory contains a name that
15581 refers to itself, called @file{.}, (``dot'') and a name that refers to
15582 its parent directory, called @file{..} (``double dot''). (In
15583 @file{/}, the root directory, @file{..} refers to itself, since
15584 @file{/} has no parent.) Clearly, we do not want our
15585 @code{files-in-below-directory} function to enter those directories,
15586 since they always lead us, directly or indirectly, to the current
15589 Consequently, our @code{files-in-below-directory} function must do
15594 Check to see whether it is looking at a filename that ends in
15595 @samp{.el}; and if so, add its name to a list.
15598 Check to see whether it is looking at a filename that is the name of a
15599 directory; and if so,
15603 Check to see whether it is looking at @file{.} or @file{..}; and if
15607 Or else, go into that directory and repeat the process.
15611 Let's write a function definition to do these tasks. We will use a
15612 @code{while} loop to move from one filename to another within a
15613 directory, checking what needs to be done; and we will use a recursive
15614 call to repeat the actions on each sub-directory. The recursive
15615 pattern is `accumulate'
15616 (@pxref{Accumulate, , Recursive Pattern: @emph{accumulate}}),
15617 using @code{append} as the combiner.
15620 (directory-files "/usr/local/src/emacs/lisp/" t "\\.el$")
15621 (shell-command "find /usr/local/src/emacs/lisp/ -name '*.el'")
15623 (directory-files "/usr/local/share/emacs/22.1.1/lisp/" t "\\.el$")
15624 (shell-command "find /usr/local/share/emacs/22.1.1/lisp/ -name '*.el'")
15627 @c /usr/local/share/emacs/22.1.1/lisp/
15630 Here is the function:
15634 (defun files-in-below-directory (directory)
15635 "List the .el files in DIRECTORY and in its sub-directories."
15636 ;; Although the function will be used non-interactively,
15637 ;; it will be easier to test if we make it interactive.
15638 ;; The directory will have a name such as
15639 ;; "/usr/local/share/emacs/22.1.1/lisp/"
15640 (interactive "DDirectory name: ")
15643 (let (el-files-list
15644 (current-directory-list
15645 (directory-files-and-attributes directory t)))
15646 ;; while we are in the current directory
15647 (while current-directory-list
15651 ;; check to see whether filename ends in `.el'
15652 ;; and if so, append its name to a list.
15653 ((equal ".el" (substring (car (car current-directory-list)) -3))
15654 (setq el-files-list
15655 (cons (car (car current-directory-list)) el-files-list)))
15658 ;; check whether filename is that of a directory
15659 ((eq t (car (cdr (car current-directory-list))))
15660 ;; decide whether to skip or recurse
15663 (substring (car (car current-directory-list)) -1))
15664 ;; then do nothing since filename is that of
15665 ;; current directory or parent, "." or ".."
15669 ;; else descend into the directory and repeat the process
15670 (setq el-files-list
15672 (files-in-below-directory
15673 (car (car current-directory-list)))
15675 ;; move to the next filename in the list; this also
15676 ;; shortens the list so the while loop eventually comes to an end
15677 (setq current-directory-list (cdr current-directory-list)))
15678 ;; return the filenames
15683 @c (files-in-below-directory "/usr/local/src/emacs/lisp/")
15684 @c (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/")
15686 The @code{files-in-below-directory} @code{directory-files} function
15687 takes one argument, the name of a directory.
15690 Thus, on my system,
15692 @c (length (files-in-below-directory "/usr/local/src/emacs/lisp/"))
15694 @c !!! 22.1.1 lisp sources location here
15698 (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/"))
15703 tells me that in and below my Lisp sources directory are 1031
15706 @code{files-in-below-directory} returns a list in reverse alphabetical
15707 order. An expression to sort the list in alphabetical order looks
15713 (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/")
15720 "Test how long it takes to find lengths of all sorted elisp defuns."
15721 (insert "\n" (current-time-string) "\n")
15724 (recursive-lengths-list-many-files
15725 (files-in-below-directory "/usr/local/src/emacs/lisp/"))
15727 (insert (format "%s" (current-time-string))))
15730 @node Counting function definitions
15731 @subsection Counting function definitions
15733 Our immediate goal is to generate a list that tells us how many
15734 function definitions contain fewer than 10 words and symbols, how many
15735 contain between 10 and 19 words and symbols, how many contain between
15736 20 and 29 words and symbols, and so on.
15738 With a sorted list of numbers, this is easy: count how many elements
15739 of the list are smaller than 10, then, after moving past the numbers
15740 just counted, count how many are smaller than 20, then, after moving
15741 past the numbers just counted, count how many are smaller than 30, and
15742 so on. Each of the numbers, 10, 20, 30, 40, and the like, is one
15743 larger than the top of that range. We can call the list of such
15744 numbers the @code{top-of-ranges} list.
15747 If we wished, we could generate this list automatically, but it is
15748 simpler to write a list manually. Here it is:
15749 @vindex top-of-ranges
15753 (defvar top-of-ranges
15756 110 120 130 140 150
15757 160 170 180 190 200
15758 210 220 230 240 250
15759 260 270 280 290 300)
15760 "List specifying ranges for `defuns-per-range'.")
15764 To change the ranges, we edit this list.
15766 Next, we need to write the function that creates the list of the
15767 number of definitions within each range. Clearly, this function must
15768 take the @code{sorted-lengths} and the @code{top-of-ranges} lists
15771 The @code{defuns-per-range} function must do two things again and
15772 again: it must count the number of definitions within a range
15773 specified by the current top-of-range value; and it must shift to the
15774 next higher value in the @code{top-of-ranges} list after counting the
15775 number of definitions in the current range. Since each of these
15776 actions is repetitive, we can use @code{while} loops for the job.
15777 One loop counts the number of definitions in the range defined by the
15778 current top-of-range value, and the other loop selects each of the
15779 top-of-range values in turn.
15781 Several entries of the @code{sorted-lengths} list are counted for each
15782 range; this means that the loop for the @code{sorted-lengths} list
15783 will be inside the loop for the @code{top-of-ranges} list, like a
15784 small gear inside a big gear.
15786 The inner loop counts the number of definitions within the range. It
15787 is a simple counting loop of the type we have seen before.
15788 (@xref{Incrementing Loop, , A loop with an incrementing counter}.)
15789 The true-or-false test of the loop tests whether the value from the
15790 @code{sorted-lengths} list is smaller than the current value of the
15791 top of the range. If it is, the function increments the counter and
15792 tests the next value from the @code{sorted-lengths} list.
15795 The inner loop looks like this:
15799 (while @var{length-element-smaller-than-top-of-range}
15800 (setq number-within-range (1+ number-within-range))
15801 (setq sorted-lengths (cdr sorted-lengths)))
15805 The outer loop must start with the lowest value of the
15806 @code{top-of-ranges} list, and then be set to each of the succeeding
15807 higher values in turn. This can be done with a loop like this:
15811 (while top-of-ranges
15812 @var{body-of-loop}@dots{}
15813 (setq top-of-ranges (cdr top-of-ranges)))
15818 Put together, the two loops look like this:
15822 (while top-of-ranges
15824 ;; @r{Count the number of elements within the current range.}
15825 (while @var{length-element-smaller-than-top-of-range}
15826 (setq number-within-range (1+ number-within-range))
15827 (setq sorted-lengths (cdr sorted-lengths)))
15829 ;; @r{Move to next range.}
15830 (setq top-of-ranges (cdr top-of-ranges)))
15834 In addition, in each circuit of the outer loop, Emacs should record
15835 the number of definitions within that range (the value of
15836 @code{number-within-range}) in a list. We can use @code{cons} for
15837 this purpose. (@xref{cons, , @code{cons}}.)
15839 The @code{cons} function works fine, except that the list it
15840 constructs will contain the number of definitions for the highest
15841 range at its beginning and the number of definitions for the lowest
15842 range at its end. This is because @code{cons} attaches new elements
15843 of the list to the beginning of the list, and since the two loops are
15844 working their way through the lengths' list from the lower end first,
15845 the @code{defuns-per-range-list} will end up largest number first.
15846 But we will want to print our graph with smallest values first and the
15847 larger later. The solution is to reverse the order of the
15848 @code{defuns-per-range-list}. We can do this using the
15849 @code{nreverse} function, which reverses the order of a list.
15856 (nreverse '(1 2 3 4))
15867 Note that the @code{nreverse} function is ``destructive''---that is,
15868 it changes the list to which it is applied; this contrasts with the
15869 @code{car} and @code{cdr} functions, which are non-destructive. In
15870 this case, we do not want the original @code{defuns-per-range-list},
15871 so it does not matter that it is destroyed. (The @code{reverse}
15872 function provides a reversed copy of a list, leaving the original list
15877 Put all together, the @code{defuns-per-range} looks like this:
15881 (defun defuns-per-range (sorted-lengths top-of-ranges)
15882 "SORTED-LENGTHS defuns in each TOP-OF-RANGES range."
15883 (let ((top-of-range (car top-of-ranges))
15884 (number-within-range 0)
15885 defuns-per-range-list)
15890 (while top-of-ranges
15896 ;; @r{Need number for numeric test.}
15897 (car sorted-lengths)
15898 (< (car sorted-lengths) top-of-range))
15902 ;; @r{Count number of definitions within current range.}
15903 (setq number-within-range (1+ number-within-range))
15904 (setq sorted-lengths (cdr sorted-lengths)))
15906 ;; @r{Exit inner loop but remain within outer loop.}
15910 (setq defuns-per-range-list
15911 (cons number-within-range defuns-per-range-list))
15912 (setq number-within-range 0) ; @r{Reset count to zero.}
15916 ;; @r{Move to next range.}
15917 (setq top-of-ranges (cdr top-of-ranges))
15918 ;; @r{Specify next top of range value.}
15919 (setq top-of-range (car top-of-ranges)))
15923 ;; @r{Exit outer loop and count the number of defuns larger than}
15924 ;; @r{ the largest top-of-range value.}
15925 (setq defuns-per-range-list
15927 (length sorted-lengths)
15928 defuns-per-range-list))
15932 ;; @r{Return a list of the number of definitions within each range,}
15933 ;; @r{ smallest to largest.}
15934 (nreverse defuns-per-range-list)))
15940 The function is straightforward except for one subtle feature. The
15941 true-or-false test of the inner loop looks like this:
15945 (and (car sorted-lengths)
15946 (< (car sorted-lengths) top-of-range))
15952 instead of like this:
15955 (< (car sorted-lengths) top-of-range)
15958 The purpose of the test is to determine whether the first item in the
15959 @code{sorted-lengths} list is less than the value of the top of the
15962 The simple version of the test works fine unless the
15963 @code{sorted-lengths} list has a @code{nil} value. In that case, the
15964 @code{(car sorted-lengths)} expression function returns
15965 @code{nil}. The @code{<} function cannot compare a number to
15966 @code{nil}, which is an empty list, so Emacs signals an error and
15967 stops the function from attempting to continue to execute.
15969 The @code{sorted-lengths} list always becomes @code{nil} when the
15970 counter reaches the end of the list. This means that any attempt to
15971 use the @code{defuns-per-range} function with the simple version of
15972 the test will fail.
15974 We solve the problem by using the @code{(car sorted-lengths)}
15975 expression in conjunction with the @code{and} expression. The
15976 @code{(car sorted-lengths)} expression returns a non-@code{nil}
15977 value so long as the list has at least one number within it, but
15978 returns @code{nil} if the list is empty. The @code{and} expression
15979 first evaluates the @code{(car sorted-lengths)} expression, and
15980 if it is @code{nil}, returns false @emph{without} evaluating the
15981 @code{<} expression. But if the @code{(car sorted-lengths)}
15982 expression returns a non-@code{nil} value, the @code{and} expression
15983 evaluates the @code{<} expression, and returns that value as the value
15984 of the @code{and} expression.
15986 @c colon in printed section title causes problem in Info cross reference
15987 This way, we avoid an error.
15990 (For information about @code{and}, see
15991 @ref{kill-new function, , The @code{kill-new} function}.)
15995 (@xref{kill-new function, , The @code{kill-new} function}, for
15996 information about @code{and}.)
15999 Here is a short test of the @code{defuns-per-range} function. First,
16000 evaluate the expression that binds (a shortened)
16001 @code{top-of-ranges} list to the list of values, then evaluate the
16002 expression for binding the @code{sorted-lengths} list, and then
16003 evaluate the @code{defuns-per-range} function.
16007 ;; @r{(Shorter list than we will use later.)}
16008 (setq top-of-ranges
16009 '(110 120 130 140 150
16010 160 170 180 190 200))
16012 (setq sorted-lengths
16013 '(85 86 110 116 122 129 154 176 179 200 265 300 300))
16015 (defuns-per-range sorted-lengths top-of-ranges)
16021 The list returned looks like this:
16024 (2 2 2 0 0 1 0 2 0 0 4)
16028 Indeed, there are two elements of the @code{sorted-lengths} list
16029 smaller than 110, two elements between 110 and 119, two elements
16030 between 120 and 129, and so on. There are four elements with a value
16033 @c The next step is to turn this numbers' list into a graph.
16034 @node Readying a Graph
16035 @chapter Readying a Graph
16036 @cindex Readying a graph
16037 @cindex Graph prototype
16038 @cindex Prototype graph
16039 @cindex Body of graph
16041 Our goal is to construct a graph showing the numbers of function
16042 definitions of various lengths in the Emacs lisp sources.
16044 As a practical matter, if you were creating a graph, you would
16045 probably use a program such as @code{gnuplot} to do the job.
16046 (@code{gnuplot} is nicely integrated into GNU Emacs.) In this case,
16047 however, we create one from scratch, and in the process we will
16048 re-acquaint ourselves with some of what we learned before and learn
16051 In this chapter, we will first write a simple graph printing function.
16052 This first definition will be a @dfn{prototype}, a rapidly written
16053 function that enables us to reconnoiter this unknown graph-making
16054 territory. We will discover dragons, or find that they are myth.
16055 After scouting the terrain, we will feel more confident and enhance
16056 the function to label the axes automatically.
16059 * Columns of a graph::
16060 * graph-body-print:: How to print the body of a graph.
16061 * recursive-graph-body-print::
16063 * Line Graph Exercise::
16067 @node Columns of a graph
16068 @unnumberedsec Printing the Columns of a Graph
16071 Since Emacs is designed to be flexible and work with all kinds of
16072 terminals, including character-only terminals, the graph will need to
16073 be made from one of the `typewriter' symbols. An asterisk will do; as
16074 we enhance the graph-printing function, we can make the choice of
16075 symbol a user option.
16077 We can call this function @code{graph-body-print}; it will take a
16078 @code{numbers-list} as its only argument. At this stage, we will not
16079 label the graph, but only print its body.
16081 The @code{graph-body-print} function inserts a vertical column of
16082 asterisks for each element in the @code{numbers-list}. The height of
16083 each line is determined by the value of that element of the
16084 @code{numbers-list}.
16086 Inserting columns is a repetitive act; that means that this function can
16087 be written either with a @code{while} loop or recursively.
16089 Our first challenge is to discover how to print a column of asterisks.
16090 Usually, in Emacs, we print characters onto a screen horizontally,
16091 line by line, by typing. We have two routes we can follow: write our
16092 own column-insertion function or discover whether one exists in Emacs.
16094 To see whether there is one in Emacs, we can use the @kbd{M-x apropos}
16095 command. This command is like the @kbd{C-h a} (@code{command-apropos})
16096 command, except that the latter finds only those functions that are
16097 commands. The @kbd{M-x apropos} command lists all symbols that match
16098 a regular expression, including functions that are not interactive.
16101 What we want to look for is some command that prints or inserts
16102 columns. Very likely, the name of the function will contain either
16103 the word `print' or the word `insert' or the word `column'.
16104 Therefore, we can simply type @kbd{M-x apropos RET
16105 print\|insert\|column RET} and look at the result. On my system, this
16106 command once too takes quite some time, and then produced a list of 79
16107 functions and variables. Now it does not take much time at all and
16108 produces a list of 211 functions and variables. Scanning down the
16109 list, the only function that looks as if it might do the job is
16110 @code{insert-rectangle}.
16113 Indeed, this is the function we want; its documentation says:
16118 Insert text of RECTANGLE with upper left corner at point.
16119 RECTANGLE's first line is inserted at point,
16120 its second line is inserted at a point vertically under point, etc.
16121 RECTANGLE should be a list of strings.
16122 After this command, the mark is at the upper left corner
16123 and point is at the lower right corner.
16127 We can run a quick test, to make sure it does what we expect of it.
16129 Here is the result of placing the cursor after the
16130 @code{insert-rectangle} expression and typing @kbd{C-u C-x C-e}
16131 (@code{eval-last-sexp}). The function inserts the strings
16132 @samp{"first"}, @samp{"second"}, and @samp{"third"} at and below
16133 point. Also the function returns @code{nil}.
16137 (insert-rectangle '("first" "second" "third"))first
16144 Of course, we won't be inserting the text of the
16145 @code{insert-rectangle} expression itself into the buffer in which we
16146 are making the graph, but will call the function from our program. We
16147 shall, however, have to make sure that point is in the buffer at the
16148 place where the @code{insert-rectangle} function will insert its
16151 If you are reading this in Info, you can see how this works by
16152 switching to another buffer, such as the @file{*scratch*} buffer,
16153 placing point somewhere in the buffer, typing @kbd{M-:}, typing the
16154 @code{insert-rectangle} expression into the minibuffer at the prompt,
16155 and then typing @key{RET}. This causes Emacs to evaluate the
16156 expression in the minibuffer, but to use as the value of point the
16157 position of point in the @file{*scratch*} buffer. (@kbd{M-:} is the
16158 keybinding for @code{eval-expression}. Also, @code{nil} does not
16159 appear in the @file{*scratch*} buffer since the expression is
16160 evaluated in the minibuffer.)
16162 We find when we do this that point ends up at the end of the last
16163 inserted line---that is to say, this function moves point as a
16164 side-effect. If we were to repeat the command, with point at this
16165 position, the next insertion would be below and to the right of the
16166 previous insertion. We don't want this! If we are going to make a
16167 bar graph, the columns need to be beside each other.
16169 So we discover that each cycle of the column-inserting @code{while}
16170 loop must reposition point to the place we want it, and that place
16171 will be at the top, not the bottom, of the column. Moreover, we
16172 remember that when we print a graph, we do not expect all the columns
16173 to be the same height. This means that the top of each column may be
16174 at a different height from the previous one. We cannot simply
16175 reposition point to the same line each time, but moved over to the
16176 right---or perhaps we can@dots{}
16178 We are planning to make the columns of the bar graph out of asterisks.
16179 The number of asterisks in the column is the number specified by the
16180 current element of the @code{numbers-list}. We need to construct a
16181 list of asterisks of the right length for each call to
16182 @code{insert-rectangle}. If this list consists solely of the requisite
16183 number of asterisks, then we will have position point the right number
16184 of lines above the base for the graph to print correctly. This could
16187 Alternatively, if we can figure out some way to pass
16188 @code{insert-rectangle} a list of the same length each time, then we
16189 can place point on the same line each time, but move it over one
16190 column to the right for each new column. If we do this, however, some
16191 of the entries in the list passed to @code{insert-rectangle} must be
16192 blanks rather than asterisks. For example, if the maximum height of
16193 the graph is 5, but the height of the column is 3, then
16194 @code{insert-rectangle} requires an argument that looks like this:
16197 (" " " " "*" "*" "*")
16200 This last proposal is not so difficult, so long as we can determine
16201 the column height. There are two ways for us to specify the column
16202 height: we can arbitrarily state what it will be, which would work
16203 fine for graphs of that height; or we can search through the list of
16204 numbers and use the maximum height of the list as the maximum height
16205 of the graph. If the latter operation were difficult, then the former
16206 procedure would be easiest, but there is a function built into Emacs
16207 that determines the maximum of its arguments. We can use that
16208 function. The function is called @code{max} and it returns the
16209 largest of all its arguments, which must be numbers. Thus, for
16217 returns 7. (A corresponding function called @code{min} returns the
16218 smallest of all its arguments.)
16222 However, we cannot simply call @code{max} on the @code{numbers-list};
16223 the @code{max} function expects numbers as its argument, not a list of
16224 numbers. Thus, the following expression,
16227 (max '(3 4 6 5 7 3))
16232 produces the following error message;
16235 Wrong type of argument: number-or-marker-p, (3 4 6 5 7 3)
16239 We need a function that passes a list of arguments to a function.
16240 This function is @code{apply}. This function `applies' its first
16241 argument (a function) to its remaining arguments, the last of which
16248 (apply 'max 3 4 7 3 '(4 8 5))
16254 (Incidentally, I don't know how you would learn of this function
16255 without a book such as this. It is possible to discover other
16256 functions, like @code{search-forward} or @code{insert-rectangle}, by
16257 guessing at a part of their names and then using @code{apropos}. Even
16258 though its base in metaphor is clear---`apply' its first argument to
16259 the rest---I doubt a novice would come up with that particular word
16260 when using @code{apropos} or other aid. Of course, I could be wrong;
16261 after all, the function was first named by someone who had to invent
16264 The second and subsequent arguments to @code{apply} are optional, so
16265 we can use @code{apply} to call a function and pass the elements of a
16266 list to it, like this, which also returns 8:
16269 (apply 'max '(4 8 5))
16272 This latter way is how we will use @code{apply}. The
16273 @code{recursive-lengths-list-many-files} function returns a numbers'
16274 list to which we can apply @code{max} (we could also apply @code{max} to
16275 the sorted numbers' list; it does not matter whether the list is
16279 Hence, the operation for finding the maximum height of the graph is this:
16282 (setq max-graph-height (apply 'max numbers-list))
16285 Now we can return to the question of how to create a list of strings
16286 for a column of the graph. Told the maximum height of the graph
16287 and the number of asterisks that should appear in the column, the
16288 function should return a list of strings for the
16289 @code{insert-rectangle} command to insert.
16291 Each column is made up of asterisks or blanks. Since the function is
16292 passed the value of the height of the column and the number of
16293 asterisks in the column, the number of blanks can be found by
16294 subtracting the number of asterisks from the height of the column.
16295 Given the number of blanks and the number of asterisks, two
16296 @code{while} loops can be used to construct the list:
16300 ;;; @r{First version.}
16301 (defun column-of-graph (max-graph-height actual-height)
16302 "Return list of strings that is one column of a graph."
16303 (let ((insert-list nil)
16304 (number-of-top-blanks
16305 (- max-graph-height actual-height)))
16309 ;; @r{Fill in asterisks.}
16310 (while (> actual-height 0)
16311 (setq insert-list (cons "*" insert-list))
16312 (setq actual-height (1- actual-height)))
16316 ;; @r{Fill in blanks.}
16317 (while (> number-of-top-blanks 0)
16318 (setq insert-list (cons " " insert-list))
16319 (setq number-of-top-blanks
16320 (1- number-of-top-blanks)))
16324 ;; @r{Return whole list.}
16329 If you install this function and then evaluate the following
16330 expression you will see that it returns the list as desired:
16333 (column-of-graph 5 3)
16341 (" " " " "*" "*" "*")
16344 As written, @code{column-of-graph} contains a major flaw: the symbols
16345 used for the blank and for the marked entries in the column are
16346 `hard-coded' as a space and asterisk. This is fine for a prototype,
16347 but you, or another user, may wish to use other symbols. For example,
16348 in testing the graph function, you many want to use a period in place
16349 of the space, to make sure the point is being repositioned properly
16350 each time the @code{insert-rectangle} function is called; or you might
16351 want to substitute a @samp{+} sign or other symbol for the asterisk.
16352 You might even want to make a graph-column that is more than one
16353 display column wide. The program should be more flexible. The way to
16354 do that is to replace the blank and the asterisk with two variables
16355 that we can call @code{graph-blank} and @code{graph-symbol} and define
16356 those variables separately.
16358 Also, the documentation is not well written. These considerations
16359 lead us to the second version of the function:
16363 (defvar graph-symbol "*"
16364 "String used as symbol in graph, usually an asterisk.")
16368 (defvar graph-blank " "
16369 "String used as blank in graph, usually a blank space.
16370 graph-blank must be the same number of columns wide
16376 (For an explanation of @code{defvar}, see
16377 @ref{defvar, , Initializing a Variable with @code{defvar}}.)
16381 ;;; @r{Second version.}
16382 (defun column-of-graph (max-graph-height actual-height)
16383 "Return MAX-GRAPH-HEIGHT strings; ACTUAL-HEIGHT are graph-symbols.
16387 The graph-symbols are contiguous entries at the end
16389 The list will be inserted as one column of a graph.
16390 The strings are either graph-blank or graph-symbol."
16394 (let ((insert-list nil)
16395 (number-of-top-blanks
16396 (- max-graph-height actual-height)))
16400 ;; @r{Fill in @code{graph-symbols}.}
16401 (while (> actual-height 0)
16402 (setq insert-list (cons graph-symbol insert-list))
16403 (setq actual-height (1- actual-height)))
16407 ;; @r{Fill in @code{graph-blanks}.}
16408 (while (> number-of-top-blanks 0)
16409 (setq insert-list (cons graph-blank insert-list))
16410 (setq number-of-top-blanks
16411 (1- number-of-top-blanks)))
16413 ;; @r{Return whole list.}
16418 If we wished, we could rewrite @code{column-of-graph} a third time to
16419 provide optionally for a line graph as well as for a bar graph. This
16420 would not be hard to do. One way to think of a line graph is that it
16421 is no more than a bar graph in which the part of each bar that is
16422 below the top is blank. To construct a column for a line graph, the
16423 function first constructs a list of blanks that is one shorter than
16424 the value, then it uses @code{cons} to attach a graph symbol to the
16425 list; then it uses @code{cons} again to attach the `top blanks' to
16428 It is easy to see how to write such a function, but since we don't
16429 need it, we will not do it. But the job could be done, and if it were
16430 done, it would be done with @code{column-of-graph}. Even more
16431 important, it is worth noting that few changes would have to be made
16432 anywhere else. The enhancement, if we ever wish to make it, is
16435 Now, finally, we come to our first actual graph printing function.
16436 This prints the body of a graph, not the labels for the vertical and
16437 horizontal axes, so we can call this @code{graph-body-print}.
16439 @node graph-body-print
16440 @section The @code{graph-body-print} Function
16441 @findex graph-body-print
16443 After our preparation in the preceding section, the
16444 @code{graph-body-print} function is straightforward. The function
16445 will print column after column of asterisks and blanks, using the
16446 elements of a numbers' list to specify the number of asterisks in each
16447 column. This is a repetitive act, which means we can use a
16448 decrementing @code{while} loop or recursive function for the job. In
16449 this section, we will write the definition using a @code{while} loop.
16451 The @code{column-of-graph} function requires the height of the graph
16452 as an argument, so we should determine and record that as a local variable.
16454 This leads us to the following template for the @code{while} loop
16455 version of this function:
16459 (defun graph-body-print (numbers-list)
16460 "@var{documentation}@dots{}"
16461 (let ((height @dots{}
16466 (while numbers-list
16467 @var{insert-columns-and-reposition-point}
16468 (setq numbers-list (cdr numbers-list)))))
16473 We need to fill in the slots of the template.
16475 Clearly, we can use the @code{(apply 'max numbers-list)} expression to
16476 determine the height of the graph.
16478 The @code{while} loop will cycle through the @code{numbers-list} one
16479 element at a time. As it is shortened by the @code{(setq numbers-list
16480 (cdr numbers-list))} expression, the @sc{car} of each instance of the
16481 list is the value of the argument for @code{column-of-graph}.
16483 At each cycle of the @code{while} loop, the @code{insert-rectangle}
16484 function inserts the list returned by @code{column-of-graph}. Since
16485 the @code{insert-rectangle} function moves point to the lower right of
16486 the inserted rectangle, we need to save the location of point at the
16487 time the rectangle is inserted, move back to that position after the
16488 rectangle is inserted, and then move horizontally to the next place
16489 from which @code{insert-rectangle} is called.
16491 If the inserted columns are one character wide, as they will be if
16492 single blanks and asterisks are used, the repositioning command is
16493 simply @code{(forward-char 1)}; however, the width of a column may be
16494 greater than one. This means that the repositioning command should be
16495 written @code{(forward-char symbol-width)}. The @code{symbol-width}
16496 itself is the length of a @code{graph-blank} and can be found using
16497 the expression @code{(length graph-blank)}. The best place to bind
16498 the @code{symbol-width} variable to the value of the width of graph
16499 column is in the varlist of the @code{let} expression.
16502 These considerations lead to the following function definition:
16506 (defun graph-body-print (numbers-list)
16507 "Print a bar graph of the NUMBERS-LIST.
16508 The numbers-list consists of the Y-axis values."
16510 (let ((height (apply 'max numbers-list))
16511 (symbol-width (length graph-blank))
16516 (while numbers-list
16517 (setq from-position (point))
16519 (column-of-graph height (car numbers-list)))
16520 (goto-char from-position)
16521 (forward-char symbol-width)
16524 ;; @r{Draw graph column by column.}
16526 (setq numbers-list (cdr numbers-list)))
16529 ;; @r{Place point for X axis labels.}
16530 (forward-line height)
16537 The one unexpected expression in this function is the
16538 @w{@code{(sit-for 0)}} expression in the @code{while} loop. This
16539 expression makes the graph printing operation more interesting to
16540 watch than it would be otherwise. The expression causes Emacs to
16541 `sit' or do nothing for a zero length of time and then redraw the
16542 screen. Placed here, it causes Emacs to redraw the screen column by
16543 column. Without it, Emacs would not redraw the screen until the
16546 We can test @code{graph-body-print} with a short list of numbers.
16550 Install @code{graph-symbol}, @code{graph-blank},
16551 @code{column-of-graph}, which are in
16553 @ref{Readying a Graph, , Readying a Graph},
16556 @ref{Columns of a graph},
16558 and @code{graph-body-print}.
16562 Copy the following expression:
16565 (graph-body-print '(1 2 3 4 6 4 3 5 7 6 5 2 3))
16569 Switch to the @file{*scratch*} buffer and place the cursor where you
16570 want the graph to start.
16573 Type @kbd{M-:} (@code{eval-expression}).
16576 Yank the @code{graph-body-print} expression into the minibuffer
16577 with @kbd{C-y} (@code{yank)}.
16580 Press @key{RET} to evaluate the @code{graph-body-print} expression.
16584 Emacs will print a graph like this:
16598 @node recursive-graph-body-print
16599 @section The @code{recursive-graph-body-print} Function
16600 @findex recursive-graph-body-print
16602 The @code{graph-body-print} function may also be written recursively.
16603 The recursive solution is divided into two parts: an outside `wrapper'
16604 that uses a @code{let} expression to determine the values of several
16605 variables that need only be found once, such as the maximum height of
16606 the graph, and an inside function that is called recursively to print
16610 The `wrapper' is uncomplicated:
16614 (defun recursive-graph-body-print (numbers-list)
16615 "Print a bar graph of the NUMBERS-LIST.
16616 The numbers-list consists of the Y-axis values."
16617 (let ((height (apply 'max numbers-list))
16618 (symbol-width (length graph-blank))
16620 (recursive-graph-body-print-internal
16627 The recursive function is a little more difficult. It has four parts:
16628 the `do-again-test', the printing code, the recursive call, and the
16629 `next-step-expression'. The `do-again-test' is a @code{when}
16630 expression that determines whether the @code{numbers-list} contains
16631 any remaining elements; if it does, the function prints one column of
16632 the graph using the printing code and calls itself again. The
16633 function calls itself again according to the value produced by the
16634 `next-step-expression' which causes the call to act on a shorter
16635 version of the @code{numbers-list}.
16639 (defun recursive-graph-body-print-internal
16640 (numbers-list height symbol-width)
16641 "Print a bar graph.
16642 Used within recursive-graph-body-print function."
16647 (setq from-position (point))
16649 (column-of-graph height (car numbers-list)))
16652 (goto-char from-position)
16653 (forward-char symbol-width)
16654 (sit-for 0) ; @r{Draw graph column by column.}
16655 (recursive-graph-body-print-internal
16656 (cdr numbers-list) height symbol-width)))
16661 After installation, this expression can be tested; here is a sample:
16664 (recursive-graph-body-print '(3 2 5 6 7 5 3 4 6 4 3 2 1))
16668 Here is what @code{recursive-graph-body-print} produces:
16682 Either of these two functions, @code{graph-body-print} or
16683 @code{recursive-graph-body-print}, create the body of a graph.
16686 @section Need for Printed Axes
16688 A graph needs printed axes, so you can orient yourself. For a do-once
16689 project, it may be reasonable to draw the axes by hand using Emacs's
16690 Picture mode; but a graph drawing function may be used more than once.
16692 For this reason, I have written enhancements to the basic
16693 @code{print-graph-body} function that automatically print labels for
16694 the horizontal and vertical axes. Since the label printing functions
16695 do not contain much new material, I have placed their description in
16696 an appendix. @xref{Full Graph, , A Graph with Labeled Axes}.
16698 @node Line Graph Exercise
16701 Write a line graph version of the graph printing functions.
16703 @node Emacs Initialization
16704 @chapter Your @file{.emacs} File
16705 @cindex @file{.emacs} file
16706 @cindex Customizing your @file{.emacs} file
16707 @cindex Initialization file
16709 ``You don't have to like Emacs to like it''---this seemingly
16710 paradoxical statement is the secret of GNU Emacs. The plain, `out of
16711 the box' Emacs is a generic tool. Most people who use it, customize
16712 it to suit themselves.
16714 GNU Emacs is mostly written in Emacs Lisp; this means that by writing
16715 expressions in Emacs Lisp you can change or extend Emacs.
16718 * Default Configuration::
16719 * Site-wide Init:: You can write site-wide init files.
16720 * defcustom:: Emacs will write code for you.
16721 * Beginning a .emacs File:: How to write a @code{.emacs file}.
16722 * Text and Auto-fill:: Automatically wrap lines.
16723 * Mail Aliases:: Use abbreviations for email addresses.
16724 * Indent Tabs Mode:: Don't use tabs with @TeX{}
16725 * Keybindings:: Create some personal keybindings.
16726 * Keymaps:: More about key binding.
16727 * Loading Files:: Load (i.e., evaluate) files automatically.
16728 * Autoload:: Make functions available.
16729 * Simple Extension:: Define a function; bind it to a key.
16730 * X11 Colors:: Colors in X.
16732 * Mode Line:: How to customize your mode line.
16736 @node Default Configuration
16737 @unnumberedsec Emacs's Default Configuration
16740 There are those who appreciate Emacs's default configuration. After
16741 all, Emacs starts you in C mode when you edit a C file, starts you in
16742 Fortran mode when you edit a Fortran file, and starts you in
16743 Fundamental mode when you edit an unadorned file. This all makes
16744 sense, if you do not know who is going to use Emacs. Who knows what a
16745 person hopes to do with an unadorned file? Fundamental mode is the
16746 right default for such a file, just as C mode is the right default for
16747 editing C code. (Enough programming languages have syntaxes
16748 that enable them to share or nearly share features, so C mode is
16749 now provided by CC mode, the `C Collection'.)
16751 But when you do know who is going to use Emacs---you,
16752 yourself---then it makes sense to customize Emacs.
16754 For example, I seldom want Fundamental mode when I edit an
16755 otherwise undistinguished file; I want Text mode. This is why I
16756 customize Emacs: so it suits me.
16758 You can customize and extend Emacs by writing or adapting a
16759 @file{~/.emacs} file. This is your personal initialization file; its
16760 contents, written in Emacs Lisp, tell Emacs what to do.@footnote{You
16761 may also add @file{.el} to @file{~/.emacs} and call it a
16762 @file{~/.emacs.el} file. In the past, you were forbidden to type the
16763 extra keystrokes that the name @file{~/.emacs.el} requires, but now
16764 you may. The new format is consistent with the Emacs Lisp file
16765 naming conventions; the old format saves typing.}
16767 A @file{~/.emacs} file contains Emacs Lisp code. You can write this
16768 code yourself; or you can use Emacs's @code{customize} feature to write
16769 the code for you. You can combine your own expressions and
16770 auto-written Customize expressions in your @file{.emacs} file.
16772 (I myself prefer to write my own expressions, except for those,
16773 particularly fonts, that I find easier to manipulate using the
16774 @code{customize} command. I combine the two methods.)
16776 Most of this chapter is about writing expressions yourself. It
16777 describes a simple @file{.emacs} file; for more information, see
16778 @ref{Init File, , The Init File, emacs, The GNU Emacs Manual}, and
16779 @ref{Init File, , The Init File, elisp, The GNU Emacs Lisp Reference
16782 @node Site-wide Init
16783 @section Site-wide Initialization Files
16785 @cindex @file{default.el} init file
16786 @cindex @file{site-init.el} init file
16787 @cindex @file{site-load.el} init file
16788 In addition to your personal initialization file, Emacs automatically
16789 loads various site-wide initialization files, if they exist. These
16790 have the same form as your @file{.emacs} file, but are loaded by
16793 Two site-wide initialization files, @file{site-load.el} and
16794 @file{site-init.el}, are loaded into Emacs and then `dumped' if a
16795 `dumped' version of Emacs is created, as is most common. (Dumped
16796 copies of Emacs load more quickly. However, once a file is loaded and
16797 dumped, a change to it does not lead to a change in Emacs unless you
16798 load it yourself or re-dump Emacs. @xref{Building Emacs, , Building
16799 Emacs, elisp, The GNU Emacs Lisp Reference Manual}, and the
16800 @file{INSTALL} file.)
16802 Three other site-wide initialization files are loaded automatically
16803 each time you start Emacs, if they exist. These are
16804 @file{site-start.el}, which is loaded @emph{before} your @file{.emacs}
16805 file, and @file{default.el}, and the terminal type file, which are both
16806 loaded @emph{after} your @file{.emacs} file.
16808 Settings and definitions in your @file{.emacs} file will overwrite
16809 conflicting settings and definitions in a @file{site-start.el} file,
16810 if it exists; but the settings and definitions in a @file{default.el}
16811 or terminal type file will overwrite those in your @file{.emacs} file.
16812 (You can prevent interference from a terminal type file by setting
16813 @code{term-file-prefix} to @code{nil}. @xref{Simple Extension, , A
16814 Simple Extension}.)
16816 @c Rewritten to avoid overfull hbox.
16817 The @file{INSTALL} file that comes in the distribution contains
16818 descriptions of the @file{site-init.el} and @file{site-load.el} files.
16820 The @file{loadup.el}, @file{startup.el}, and @file{loaddefs.el} files
16821 control loading. These files are in the @file{lisp} directory of the
16822 Emacs distribution and are worth perusing.
16824 The @file{loaddefs.el} file contains a good many suggestions as to
16825 what to put into your own @file{.emacs} file, or into a site-wide
16826 initialization file.
16829 @section Specifying Variables using @code{defcustom}
16832 You can specify variables using @code{defcustom} so that you and
16833 others can then use Emacs's @code{customize} feature to set their
16834 values. (You cannot use @code{customize} to write function
16835 definitions; but you can write @code{defuns} in your @file{.emacs}
16836 file. Indeed, you can write any Lisp expression in your @file{.emacs}
16839 The @code{customize} feature depends on the @code{defcustom} macro.
16840 Although you can use @code{defvar} or @code{setq} for variables that
16841 users set, the @code{defcustom} macro is designed for the job.
16843 You can use your knowledge of @code{defvar} for writing the
16844 first three arguments for @code{defcustom}. The first argument to
16845 @code{defcustom} is the name of the variable. The second argument is
16846 the variable's initial value, if any; and this value is set only if
16847 the value has not already been set. The third argument is the
16850 The fourth and subsequent arguments to @code{defcustom} specify types
16851 and options; these are not featured in @code{defvar}. (These
16852 arguments are optional.)
16854 Each of these arguments consists of a keyword followed by a value.
16855 Each keyword starts with the colon character @samp{:}.
16858 For example, the customizable user option variable
16859 @code{text-mode-hook} looks like this:
16863 (defcustom text-mode-hook nil
16864 "Normal hook run when entering Text mode and many related modes."
16866 :options '(turn-on-auto-fill flyspell-mode)
16872 The name of the variable is @code{text-mode-hook}; it has no default
16873 value; and its documentation string tells you what it does.
16875 The @code{:type} keyword tells Emacs the kind of data to which
16876 @code{text-mode-hook} should be set and how to display the value in a
16877 Customization buffer.
16879 The @code{:options} keyword specifies a suggested list of values for
16880 the variable. Usually, @code{:options} applies to a hook.
16881 The list is only a suggestion; it is not exclusive; a person who sets
16882 the variable may set it to other values; the list shown following the
16883 @code{:options} keyword is intended to offer convenient choices to a
16886 Finally, the @code{:group} keyword tells the Emacs Customization
16887 command in which group the variable is located. This tells where to
16890 The @code{defcustom} macro recognizes more than a dozen keywords.
16891 For more information, see @ref{Customization, , Writing Customization
16892 Definitions, elisp, The GNU Emacs Lisp Reference Manual}.
16894 Consider @code{text-mode-hook} as an example.
16896 There are two ways to customize this variable. You can use the
16897 customization command or write the appropriate expressions yourself.
16900 Using the customization command, you can type:
16907 and find that the group for editing files of data is called `data'.
16908 Enter that group. Text Mode Hook is the first member. You can click
16909 on its various options, such as @code{turn-on-auto-fill}, to set the
16910 values. After you click on the button to
16913 Save for Future Sessions
16917 Emacs will write an expression into your @file{.emacs} file.
16918 It will look like this:
16922 (custom-set-variables
16923 ;; custom-set-variables was added by Custom.
16924 ;; If you edit it by hand, you could mess it up, so be careful.
16925 ;; Your init file should contain only one such instance.
16926 ;; If there is more than one, they won't work right.
16927 '(text-mode-hook (quote (turn-on-auto-fill text-mode-hook-identify))))
16932 (The @code{text-mode-hook-identify} function tells
16933 @code{toggle-text-mode-auto-fill} which buffers are in Text mode.
16934 It comes on automatically.)
16936 The @code{custom-set-variables} function works somewhat differently
16937 than a @code{setq}. While I have never learned the differences, I
16938 modify the @code{custom-set-variables} expressions in my @file{.emacs}
16939 file by hand: I make the changes in what appears to me to be a
16940 reasonable manner and have not had any problems. Others prefer to use
16941 the Customization command and let Emacs do the work for them.
16943 Another @code{custom-set-@dots{}} function is @code{custom-set-faces}.
16944 This function sets the various font faces. Over time, I have set a
16945 considerable number of faces. Some of the time, I re-set them using
16946 @code{customize}; other times, I simply edit the
16947 @code{custom-set-faces} expression in my @file{.emacs} file itself.
16949 The second way to customize your @code{text-mode-hook} is to set it
16950 yourself in your @file{.emacs} file using code that has nothing to do
16951 with the @code{custom-set-@dots{}} functions.
16954 When you do this, and later use @code{customize}, you will see a
16958 CHANGED outside Customize; operating on it here may be unreliable.
16962 This message is only a warning. If you click on the button to
16965 Save for Future Sessions
16969 Emacs will write a @code{custom-set-@dots{}} expression near the end
16970 of your @file{.emacs} file that will be evaluated after your
16971 hand-written expression. It will, therefore, overrule your
16972 hand-written expression. No harm will be done. When you do this,
16973 however, be careful to remember which expression is active; if you
16974 forget, you may confuse yourself.
16976 So long as you remember where the values are set, you will have no
16977 trouble. In any event, the values are always set in your
16978 initialization file, which is usually called @file{.emacs}.
16980 I myself use @code{customize} for hardly anything. Mostly, I write
16981 expressions myself.
16985 Incidentally, to be more complete concerning defines: @code{defsubst}
16986 defines an inline function. The syntax is just like that of
16987 @code{defun}. @code{defconst} defines a symbol as a constant. The
16988 intent is that neither programs nor users should ever change a value
16989 set by @code{defconst}. (You can change it; the value set is a
16990 variable; but please do not.)
16992 @node Beginning a .emacs File
16993 @section Beginning a @file{.emacs} File
16994 @cindex @file{.emacs} file, beginning of
16996 When you start Emacs, it loads your @file{.emacs} file unless you tell
16997 it not to by specifying @samp{-q} on the command line. (The
16998 @code{emacs -q} command gives you a plain, out-of-the-box Emacs.)
17000 A @file{.emacs} file contains Lisp expressions. Often, these are no
17001 more than expressions to set values; sometimes they are function
17004 @xref{Init File, , The Init File @file{~/.emacs}, emacs, The GNU Emacs
17005 Manual}, for a short description of initialization files.
17007 This chapter goes over some of the same ground, but is a walk among
17008 extracts from a complete, long-used @file{.emacs} file---my own.
17010 The first part of the file consists of comments: reminders to myself.
17011 By now, of course, I remember these things, but when I started, I did
17017 ;;;; Bob's .emacs file
17018 ; Robert J. Chassell
17019 ; 26 September 1985
17024 Look at that date! I started this file a long time ago. I have been
17025 adding to it ever since.
17029 ; Each section in this file is introduced by a
17030 ; line beginning with four semicolons; and each
17031 ; entry is introduced by a line beginning with
17032 ; three semicolons.
17037 This describes the usual conventions for comments in Emacs Lisp.
17038 Everything on a line that follows a semicolon is a comment. Two,
17039 three, and four semicolons are used as subsection and section markers.
17040 (@xref{Comments, ,, elisp, The GNU Emacs Lisp Reference Manual}, for
17041 more about comments.)
17046 ; Control-h is the help key;
17047 ; after typing control-h, type a letter to
17048 ; indicate the subject about which you want help.
17049 ; For an explanation of the help facility,
17050 ; type control-h two times in a row.
17055 Just remember: type @kbd{C-h} two times for help.
17059 ; To find out about any mode, type control-h m
17060 ; while in that mode. For example, to find out
17061 ; about mail mode, enter mail mode and then type
17067 `Mode help', as I call this, is very helpful. Usually, it tells you
17068 all you need to know.
17070 Of course, you don't need to include comments like these in your
17071 @file{.emacs} file. I included them in mine because I kept forgetting
17072 about Mode help or the conventions for comments---but I was able to
17073 remember to look here to remind myself.
17075 @node Text and Auto-fill
17076 @section Text and Auto Fill Mode
17078 Now we come to the part that `turns on' Text mode and
17083 ;;; Text mode and Auto Fill mode
17084 ;; The next two lines put Emacs into Text mode
17085 ;; and Auto Fill mode, and are for writers who
17086 ;; want to start writing prose rather than code.
17087 (setq-default major-mode 'text-mode)
17088 (add-hook 'text-mode-hook 'turn-on-auto-fill)
17092 Here is the first part of this @file{.emacs} file that does something
17093 besides remind a forgetful human!
17095 The first of the two lines in parentheses tells Emacs to turn on Text
17096 mode when you find a file, @emph{unless} that file should go into some
17097 other mode, such as C mode.
17099 @cindex Per-buffer, local variables list
17100 @cindex Local variables list, per-buffer,
17101 @cindex Automatic mode selection
17102 @cindex Mode selection, automatic
17103 When Emacs reads a file, it looks at the extension to the file name,
17104 if any. (The extension is the part that comes after a @samp{.}.) If
17105 the file ends with a @samp{.c} or @samp{.h} extension then Emacs turns
17106 on C mode. Also, Emacs looks at first nonblank line of the file; if
17107 the line says @w{@samp{-*- C -*-}}, Emacs turns on C mode. Emacs
17108 possesses a list of extensions and specifications that it uses
17109 automatically. In addition, Emacs looks near the last page for a
17110 per-buffer, ``local variables list'', if any.
17113 @xref{Choosing Modes, , How Major Modes are Chosen, emacs, The GNU
17116 @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs
17120 See sections ``How Major Modes are Chosen'' and ``Local Variables in
17121 Files'' in @cite{The GNU Emacs Manual}.
17124 Now, back to the @file{.emacs} file.
17127 Here is the line again; how does it work?
17129 @cindex Text Mode turned on
17131 (setq major-mode 'text-mode)
17135 This line is a short, but complete Emacs Lisp expression.
17137 We are already familiar with @code{setq}. It sets the following variable,
17138 @code{major-mode}, to the subsequent value, which is @code{text-mode}.
17139 The single quote mark before @code{text-mode} tells Emacs to deal directly
17140 with the @code{text-mode} symbol, not with whatever it might stand for.
17141 @xref{set & setq, , Setting the Value of a Variable},
17142 for a reminder of how @code{setq} works.
17143 The main point is that there is no difference between the procedure you
17144 use to set a value in your @file{.emacs} file and the procedure you use
17145 anywhere else in Emacs.
17148 Here is the next line:
17150 @cindex Auto Fill mode turned on
17153 (add-hook 'text-mode-hook 'turn-on-auto-fill)
17157 In this line, the @code{add-hook} command adds
17158 @code{turn-on-auto-fill} to the variable.
17160 @code{turn-on-auto-fill} is the name of a program, that, you guessed
17161 it!, turns on Auto Fill mode.
17163 Every time Emacs turns on Text mode, Emacs runs the commands `hooked'
17164 onto Text mode. So every time Emacs turns on Text mode, Emacs also
17165 turns on Auto Fill mode.
17167 In brief, the first line causes Emacs to enter Text mode when you edit a
17168 file, unless the file name extension, a first non-blank line, or local
17169 variables to tell Emacs otherwise.
17171 Text mode among other actions, sets the syntax table to work
17172 conveniently for writers. In Text mode, Emacs considers an apostrophe
17173 as part of a word like a letter; but Emacs does not consider a period
17174 or a space as part of a word. Thus, @kbd{M-f} moves you over
17175 @samp{it's}. On the other hand, in C mode, @kbd{M-f} stops just after
17176 the @samp{t} of @samp{it's}.
17178 The second line causes Emacs to turn on Auto Fill mode when it turns
17179 on Text mode. In Auto Fill mode, Emacs automatically breaks a line
17180 that is too wide and brings the excessively wide part of the line down
17181 to the next line. Emacs breaks lines between words, not within them.
17183 When Auto Fill mode is turned off, lines continue to the right as you
17184 type them. Depending on how you set the value of
17185 @code{truncate-lines}, the words you type either disappear off the
17186 right side of the screen, or else are shown, in a rather ugly and
17187 unreadable manner, as a continuation line on the screen.
17190 In addition, in this part of my @file{.emacs} file, I tell the Emacs
17191 fill commands to insert two spaces after a colon:
17194 (setq colon-double-space t)
17198 @section Mail Aliases
17200 Here is a @code{setq} that `turns on' mail aliases, along with more
17206 ; To enter mail mode, type `C-x m'
17207 ; To enter RMAIL (for reading mail),
17209 (setq mail-aliases t)
17213 @cindex Mail aliases
17215 This @code{setq} command sets the value of the variable
17216 @code{mail-aliases} to @code{t}. Since @code{t} means true, the line
17217 says, in effect, ``Yes, use mail aliases.''
17219 Mail aliases are convenient short names for long email addresses or
17220 for lists of email addresses. The file where you keep your `aliases'
17221 is @file{~/.mailrc}. You write an alias like this:
17224 alias geo george@@foobar.wiz.edu
17228 When you write a message to George, address it to @samp{geo}; the
17229 mailer will automatically expand @samp{geo} to the full address.
17231 @node Indent Tabs Mode
17232 @section Indent Tabs Mode
17233 @cindex Tabs, preventing
17234 @findex indent-tabs-mode
17236 By default, Emacs inserts tabs in place of multiple spaces when it
17237 formats a region. (For example, you might indent many lines of text
17238 all at once with the @code{indent-region} command.) Tabs look fine on
17239 a terminal or with ordinary printing, but they produce badly indented
17240 output when you use @TeX{} or Texinfo since @TeX{} ignores tabs.
17243 The following turns off Indent Tabs mode:
17247 ;;; Prevent Extraneous Tabs
17248 (setq-default indent-tabs-mode nil)
17252 Note that this line uses @code{setq-default} rather than the
17253 @code{setq} command that we have seen before. The @code{setq-default}
17254 command sets values only in buffers that do not have their own local
17255 values for the variable.
17258 @xref{Just Spaces, , Tabs vs. Spaces, emacs, The GNU Emacs Manual}.
17260 @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs
17264 See sections ``Tabs vs.@: Spaces'' and ``Local Variables in
17265 Files'' in @cite{The GNU Emacs Manual}.
17270 @section Some Keybindings
17272 Now for some personal keybindings:
17276 ;;; Compare windows
17277 (global-set-key "\C-cw" 'compare-windows)
17281 @findex compare-windows
17282 @code{compare-windows} is a nifty command that compares the text in
17283 your current window with text in the next window. It makes the
17284 comparison by starting at point in each window, moving over text in
17285 each window as far as they match. I use this command all the time.
17287 This also shows how to set a key globally, for all modes.
17289 @cindex Setting a key globally
17290 @cindex Global set key
17291 @cindex Key setting globally
17292 @findex global-set-key
17293 The command is @code{global-set-key}. It is followed by the
17294 keybinding. In a @file{.emacs} file, the keybinding is written as
17295 shown: @code{\C-c} stands for `control-c', which means `press the
17296 control key and the @key{c} key at the same time'. The @code{w} means
17297 `press the @key{w} key'. The keybinding is surrounded by double
17298 quotation marks. In documentation, you would write this as
17299 @w{@kbd{C-c w}}. (If you were binding a @key{META} key, such as
17300 @kbd{M-c}, rather than a @key{CTRL} key, you would write
17301 @w{@code{\M-c}} in your @file{.emacs} file. @xref{Init Rebinding, ,
17302 Rebinding Keys in Your Init File, emacs, The GNU Emacs Manual}, for
17305 The command invoked by the keys is @code{compare-windows}. Note that
17306 @code{compare-windows} is preceded by a single quote; otherwise, Emacs
17307 would first try to evaluate the symbol to determine its value.
17309 These three things, the double quotation marks, the backslash before
17310 the @samp{C}, and the single quote mark are necessary parts of
17311 keybinding that I tend to forget. Fortunately, I have come to
17312 remember that I should look at my existing @file{.emacs} file, and
17313 adapt what is there.
17315 As for the keybinding itself: @kbd{C-c w}. This combines the prefix
17316 key, @kbd{C-c}, with a single character, in this case, @kbd{w}. This
17317 set of keys, @kbd{C-c} followed by a single character, is strictly
17318 reserved for individuals' own use. (I call these `own' keys, since
17319 these are for my own use.) You should always be able to create such a
17320 keybinding for your own use without stomping on someone else's
17321 keybinding. If you ever write an extension to Emacs, please avoid
17322 taking any of these keys for public use. Create a key like @kbd{C-c
17323 C-w} instead. Otherwise, we will run out of `own' keys.
17326 Here is another keybinding, with a comment:
17330 ;;; Keybinding for `occur'
17331 ; I use occur a lot, so let's bind it to a key:
17332 (global-set-key "\C-co" 'occur)
17337 The @code{occur} command shows all the lines in the current buffer
17338 that contain a match for a regular expression. Matching lines are
17339 shown in a buffer called @file{*Occur*}. That buffer serves as a menu
17340 to jump to occurrences.
17342 @findex global-unset-key
17343 @cindex Unbinding key
17344 @cindex Key unbinding
17346 Here is how to unbind a key, so it does not
17352 (global-unset-key "\C-xf")
17356 There is a reason for this unbinding: I found I inadvertently typed
17357 @w{@kbd{C-x f}} when I meant to type @kbd{C-x C-f}. Rather than find a
17358 file, as I intended, I accidentally set the width for filled text,
17359 almost always to a width I did not want. Since I hardly ever reset my
17360 default width, I simply unbound the key.
17362 @findex list-buffers, @r{rebound}
17363 @findex buffer-menu, @r{bound to key}
17365 The following rebinds an existing key:
17369 ;;; Rebind `C-x C-b' for `buffer-menu'
17370 (global-set-key "\C-x\C-b" 'buffer-menu)
17374 By default, @kbd{C-x C-b} runs the
17375 @code{list-buffers} command. This command lists
17376 your buffers in @emph{another} window. Since I
17377 almost always want to do something in that
17378 window, I prefer the @code{buffer-menu}
17379 command, which not only lists the buffers,
17380 but moves point into that window.
17385 @cindex Rebinding keys
17387 Emacs uses @dfn{keymaps} to record which keys call which commands.
17388 When you use @code{global-set-key} to set the keybinding for a single
17389 command in all parts of Emacs, you are specifying the keybinding in
17390 @code{current-global-map}.
17392 Specific modes, such as C mode or Text mode, have their own keymaps;
17393 the mode-specific keymaps override the global map that is shared by
17396 The @code{global-set-key} function binds, or rebinds, the global
17397 keymap. For example, the following binds the key @kbd{C-x C-b} to the
17398 function @code{buffer-menu}:
17401 (global-set-key "\C-x\C-b" 'buffer-menu)
17404 Mode-specific keymaps are bound using the @code{define-key} function,
17405 which takes a specific keymap as an argument, as well as the key and
17406 the command. For example, my @file{.emacs} file contains the
17407 following expression to bind the @code{texinfo-insert-@@group} command
17408 to @kbd{C-c C-c g}:
17412 (define-key texinfo-mode-map "\C-c\C-cg" 'texinfo-insert-@@group)
17417 The @code{texinfo-insert-@@group} function itself is a little extension
17418 to Texinfo mode that inserts @samp{@@group} into a Texinfo file. I
17419 use this command all the time and prefer to type the three strokes
17420 @kbd{C-c C-c g} rather than the six strokes @kbd{@@ g r o u p}.
17421 (@samp{@@group} and its matching @samp{@@end group} are commands that
17422 keep all enclosed text together on one page; many multi-line examples
17423 in this book are surrounded by @samp{@@group @dots{} @@end group}.)
17426 Here is the @code{texinfo-insert-@@group} function definition:
17430 (defun texinfo-insert-@@group ()
17431 "Insert the string @@group in a Texinfo buffer."
17433 (beginning-of-line)
17434 (insert "@@group\n"))
17438 (Of course, I could have used Abbrev mode to save typing, rather than
17439 write a function to insert a word; but I prefer key strokes consistent
17440 with other Texinfo mode key bindings.)
17442 You will see numerous @code{define-key} expressions in
17443 @file{loaddefs.el} as well as in the various mode libraries, such as
17444 @file{cc-mode.el} and @file{lisp-mode.el}.
17446 @xref{Key Bindings, , Customizing Key Bindings, emacs, The GNU Emacs
17447 Manual}, and @ref{Keymaps, , Keymaps, elisp, The GNU Emacs Lisp
17448 Reference Manual}, for more information about keymaps.
17450 @node Loading Files
17451 @section Loading Files
17452 @cindex Loading files
17455 Many people in the GNU Emacs community have written extensions to
17456 Emacs. As time goes by, these extensions are often included in new
17457 releases. For example, the Calendar and Diary packages are now part
17458 of the standard GNU Emacs, as is Calc.
17460 You can use a @code{load} command to evaluate a complete file and
17461 thereby install all the functions and variables in the file into Emacs.
17464 @c (auto-compression-mode t)
17467 (load "~/emacs/slowsplit")
17470 This evaluates, i.e., loads, the @file{slowsplit.el} file or if it
17471 exists, the faster, byte compiled @file{slowsplit.elc} file from the
17472 @file{emacs} sub-directory of your home directory. The file contains
17473 the function @code{split-window-quietly}, which John Robinson wrote in
17476 The @code{split-window-quietly} function splits a window with the
17477 minimum of redisplay. I installed it in 1989 because it worked well
17478 with the slow 1200 baud terminals I was then using. Nowadays, I only
17479 occasionally come across such a slow connection, but I continue to use
17480 the function because I like the way it leaves the bottom half of a
17481 buffer in the lower of the new windows and the top half in the upper
17485 To replace the key binding for the default
17486 @code{split-window-vertically}, you must also unset that key and bind
17487 the keys to @code{split-window-quietly}, like this:
17491 (global-unset-key "\C-x2")
17492 (global-set-key "\C-x2" 'split-window-quietly)
17497 If you load many extensions, as I do, then instead of specifying the
17498 exact location of the extension file, as shown above, you can specify
17499 that directory as part of Emacs's @code{load-path}. Then, when Emacs
17500 loads a file, it will search that directory as well as its default
17501 list of directories. (The default list is specified in @file{paths.h}
17502 when Emacs is built.)
17505 The following command adds your @file{~/emacs} directory to the
17506 existing load path:
17510 ;;; Emacs Load Path
17511 (setq load-path (cons "~/emacs" load-path))
17515 Incidentally, @code{load-library} is an interactive interface to the
17516 @code{load} function. The complete function looks like this:
17518 @findex load-library
17521 (defun load-library (library)
17522 "Load the library named LIBRARY.
17523 This is an interface to the function `load'."
17525 (list (completing-read "Load library: "
17526 (apply-partially 'locate-file-completion-table
17528 (get-load-suffixes)))))
17533 The name of the function, @code{load-library}, comes from the use of
17534 `library' as a conventional synonym for `file'. The source for the
17535 @code{load-library} command is in the @file{files.el} library.
17537 Another interactive command that does a slightly different job is
17538 @code{load-file}. @xref{Lisp Libraries, , Libraries of Lisp Code for
17539 Emacs, emacs, The GNU Emacs Manual}, for information on the
17540 distinction between @code{load-library} and this command.
17543 @section Autoloading
17546 Instead of installing a function by loading the file that contains it,
17547 or by evaluating the function definition, you can make the function
17548 available but not actually install it until it is first called. This
17549 is called @dfn{autoloading}.
17551 When you execute an autoloaded function, Emacs automatically evaluates
17552 the file that contains the definition, and then calls the function.
17554 Emacs starts quicker with autoloaded functions, since their libraries
17555 are not loaded right away; but you need to wait a moment when you
17556 first use such a function, while its containing file is evaluated.
17558 Rarely used functions are frequently autoloaded. The
17559 @file{loaddefs.el} library contains hundreds of autoloaded functions,
17560 from @code{bookmark-set} to @code{wordstar-mode}. Of course, you may
17561 come to use a `rare' function frequently. When you do, you should
17562 load that function's file with a @code{load} expression in your
17563 @file{.emacs} file.
17565 In my @file{.emacs} file, I load 14 libraries that contain functions
17566 that would otherwise be autoloaded. (Actually, it would have been
17567 better to include these files in my `dumped' Emacs, but I forgot.
17568 @xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp
17569 Reference Manual}, and the @file{INSTALL} file for more about
17572 You may also want to include autoloaded expressions in your @file{.emacs}
17573 file. @code{autoload} is a built-in function that takes up to five
17574 arguments, the final three of which are optional. The first argument
17575 is the name of the function to be autoloaded; the second is the name
17576 of the file to be loaded. The third argument is documentation for the
17577 function, and the fourth tells whether the function can be called
17578 interactively. The fifth argument tells what type of
17579 object---@code{autoload} can handle a keymap or macro as well as a
17580 function (the default is a function).
17583 Here is a typical example:
17587 (autoload 'html-helper-mode
17588 "html-helper-mode" "Edit HTML documents" t)
17593 (@code{html-helper-mode} is an older alternative to @code{html-mode},
17594 which is a standard part of the distribution.)
17597 This expression autoloads the @code{html-helper-mode} function. It
17598 takes it from the @file{html-helper-mode.el} file (or from the byte
17599 compiled version @file{html-helper-mode.elc}, if that exists.) The
17600 file must be located in a directory specified by @code{load-path}.
17601 The documentation says that this is a mode to help you edit documents
17602 written in the HyperText Markup Language. You can call this mode
17603 interactively by typing @kbd{M-x html-helper-mode}. (You need to
17604 duplicate the function's regular documentation in the autoload
17605 expression because the regular function is not yet loaded, so its
17606 documentation is not available.)
17608 @xref{Autoload, , Autoload, elisp, The GNU Emacs Lisp Reference
17609 Manual}, for more information.
17611 @node Simple Extension
17612 @section A Simple Extension: @code{line-to-top-of-window}
17613 @findex line-to-top-of-window
17614 @cindex Simple extension in @file{.emacs} file
17616 Here is a simple extension to Emacs that moves the line point is on to
17617 the top of the window. I use this all the time, to make text easier
17620 You can put the following code into a separate file and then load it
17621 from your @file{.emacs} file, or you can include it within your
17622 @file{.emacs} file.
17625 Here is the definition:
17629 ;;; Line to top of window;
17630 ;;; replace three keystroke sequence C-u 0 C-l
17631 (defun line-to-top-of-window ()
17632 "Move the line point is on to top of window."
17639 Now for the keybinding.
17641 Nowadays, function keys as well as mouse button events and
17642 non-@sc{ascii} characters are written within square brackets, without
17643 quotation marks. (In Emacs version 18 and before, you had to write
17644 different function key bindings for each different make of terminal.)
17646 I bind @code{line-to-top-of-window} to my @key{F6} function key like
17650 (global-set-key [f6] 'line-to-top-of-window)
17653 For more information, see @ref{Init Rebinding, , Rebinding Keys in
17654 Your Init File, emacs, The GNU Emacs Manual}.
17656 @cindex Conditional 'twixt two versions of Emacs
17657 @cindex Version of Emacs, choosing
17658 @cindex Emacs version, choosing
17659 If you run two versions of GNU Emacs, such as versions 22 and 23, and
17660 use one @file{.emacs} file, you can select which code to evaluate with
17661 the following conditional:
17666 ((= 22 emacs-major-version)
17667 ;; evaluate version 22 code
17669 ((= 23 emacs-major-version)
17670 ;; evaluate version 23 code
17675 For example, recent versions blink
17676 their cursors by default. I hate such blinking, as well as other
17677 features, so I placed the following in my @file{.emacs}
17678 file@footnote{When I start instances of Emacs that do not load my
17679 @file{.emacs} file or any site file, I also turn off blinking:
17682 emacs -q --no-site-file -eval '(blink-cursor-mode nil)'
17684 @exdent Or nowadays, using an even more sophisticated set of options,
17692 (when (>= emacs-major-version 21)
17693 (blink-cursor-mode 0)
17694 ;; Insert newline when you press `C-n' (next-line)
17695 ;; at the end of the buffer
17696 (setq next-line-add-newlines t)
17699 ;; Turn on image viewing
17700 (auto-image-file-mode t)
17703 ;; Turn on menu bar (this bar has text)
17704 ;; (Use numeric argument to turn on)
17708 ;; Turn off tool bar (this bar has icons)
17709 ;; (Use numeric argument to turn on)
17710 (tool-bar-mode nil)
17713 ;; Turn off tooltip mode for tool bar
17714 ;; (This mode causes icon explanations to pop up)
17715 ;; (Use numeric argument to turn on)
17717 ;; If tooltips turned on, make tips appear promptly
17718 (setq tooltip-delay 0.1) ; default is 0.7 second
17724 @section X11 Colors
17726 You can specify colors when you use Emacs with the MIT X Windowing
17729 I dislike the default colors and specify my own.
17732 Here are the expressions in my @file{.emacs}
17733 file that set values:
17737 ;; Set cursor color
17738 (set-cursor-color "white")
17741 (set-mouse-color "white")
17743 ;; Set foreground and background
17744 (set-foreground-color "white")
17745 (set-background-color "darkblue")
17749 ;;; Set highlighting colors for isearch and drag
17750 (set-face-foreground 'highlight "white")
17751 (set-face-background 'highlight "blue")
17755 (set-face-foreground 'region "cyan")
17756 (set-face-background 'region "blue")
17760 (set-face-foreground 'secondary-selection "skyblue")
17761 (set-face-background 'secondary-selection "darkblue")
17765 ;; Set calendar highlighting colors
17766 (setq calendar-load-hook
17768 (set-face-foreground 'diary-face "skyblue")
17769 (set-face-background 'holiday-face "slate blue")
17770 (set-face-foreground 'holiday-face "white")))
17774 The various shades of blue soothe my eye and prevent me from seeing
17775 the screen flicker.
17777 Alternatively, I could have set my specifications in various X
17778 initialization files. For example, I could set the foreground,
17779 background, cursor, and pointer (i.e., mouse) colors in my
17780 @file{~/.Xresources} file like this:
17784 Emacs*foreground: white
17785 Emacs*background: darkblue
17786 Emacs*cursorColor: white
17787 Emacs*pointerColor: white
17791 In any event, since it is not part of Emacs, I set the root color of
17792 my X window in my @file{~/.xinitrc} file, like this@footnote{I also
17793 run more modern window managers, such as Enlightenment, Gnome, or KDE;
17794 in those cases, I often specify an image rather than a plain color.}:
17797 xsetroot -solid Navy -fg white &
17801 @node Miscellaneous
17802 @section Miscellaneous Settings for a @file{.emacs} File
17805 Here are a few miscellaneous settings:
17810 Set the shape and color of the mouse cursor:
17814 ; Cursor shapes are defined in
17815 ; `/usr/include/X11/cursorfont.h';
17816 ; for example, the `target' cursor is number 128;
17817 ; the `top_left_arrow' cursor is number 132.
17821 (let ((mpointer (x-get-resource "*mpointer"
17822 "*emacs*mpointer")))
17823 ;; If you have not set your mouse pointer
17824 ;; then set it, otherwise leave as is:
17825 (if (eq mpointer nil)
17826 (setq mpointer "132")) ; top_left_arrow
17829 (setq x-pointer-shape (string-to-int mpointer))
17830 (set-mouse-color "white"))
17835 Or you can set the values of a variety of features in an alist, like
17841 default-frame-alist
17842 '((cursor-color . "white")
17843 (mouse-color . "white")
17844 (foreground-color . "white")
17845 (background-color . "DodgerBlue4")
17846 ;; (cursor-type . bar)
17847 (cursor-type . box)
17850 (tool-bar-lines . 0)
17851 (menu-bar-lines . 1)
17855 "-Misc-Fixed-Medium-R-Normal--20-200-75-75-C-100-ISO8859-1")
17861 Convert @kbd{@key{CTRL}-h} into @key{DEL} and @key{DEL}
17862 into @kbd{@key{CTRL}-h}.@*
17863 (Some older keyboards needed this, although I have not seen the
17868 ;; Translate `C-h' to <DEL>.
17869 ; (keyboard-translate ?\C-h ?\C-?)
17871 ;; Translate <DEL> to `C-h'.
17872 (keyboard-translate ?\C-? ?\C-h)
17876 @item Turn off a blinking cursor!
17880 (if (fboundp 'blink-cursor-mode)
17881 (blink-cursor-mode -1))
17886 or start GNU Emacs with the command @code{emacs -nbc}.
17889 @item When using `grep'@*
17890 @samp{-i}@w{ } Ignore case distinctions@*
17891 @samp{-n}@w{ } Prefix each line of output with line number@*
17892 @samp{-H}@w{ } Print the filename for each match.@*
17893 @samp{-e}@w{ } Protect patterns beginning with a hyphen character, @samp{-}
17896 (setq grep-command "grep -i -nH -e ")
17900 @c Evidently, no longer needed in GNU Emacs 22
17902 item Automatically uncompress compressed files when visiting them
17905 (load "uncompress")
17910 @item Find an existing buffer, even if it has a different name@*
17911 This avoids problems with symbolic links.
17914 (setq find-file-existing-other-name t)
17917 @item Set your language environment and default input method
17921 (set-language-environment "latin-1")
17922 ;; Remember you can enable or disable multilingual text input
17923 ;; with the @code{toggle-input-method'} (@kbd{C-\}) command
17924 (setq default-input-method "latin-1-prefix")
17928 If you want to write with Chinese `GB' characters, set this instead:
17932 (set-language-environment "Chinese-GB")
17933 (setq default-input-method "chinese-tonepy")
17938 @subsubheading Fixing Unpleasant Key Bindings
17939 @cindex Key bindings, fixing
17940 @cindex Bindings, key, fixing unpleasant
17942 Some systems bind keys unpleasantly. Sometimes, for example, the
17943 @key{CTRL} key appears in an awkward spot rather than at the far left
17946 Usually, when people fix these sorts of keybindings, they do not
17947 change their @file{~/.emacs} file. Instead, they bind the proper keys
17948 on their consoles with the @code{loadkeys} or @code{install-keymap}
17949 commands in their boot script and then include @code{xmodmap} commands
17950 in their @file{.xinitrc} or @file{.Xsession} file for X Windows.
17958 loadkeys /usr/share/keymaps/i386/qwerty/emacs2.kmap.gz
17960 install-keymap emacs2
17966 For a @file{.xinitrc} or @file{.Xsession} file when the @key{Caps
17967 Lock} key is at the far left of the home row:
17971 # Bind the key labeled `Caps Lock' to `Control'
17972 # (Such a broken user interface suggests that keyboard manufacturers
17973 # think that computers are typewriters from 1885.)
17975 xmodmap -e "clear Lock"
17976 xmodmap -e "add Control = Caps_Lock"
17982 In a @file{.xinitrc} or @file{.Xsession} file, to convert an @key{ALT}
17983 key to a @key{META} key:
17987 # Some ill designed keyboards have a key labeled ALT and no Meta
17988 xmodmap -e "keysym Alt_L = Meta_L Alt_L"
17994 @section A Modified Mode Line
17995 @vindex mode-line-format
17996 @cindex Mode line format
17998 Finally, a feature I really like: a modified mode line.
18000 When I work over a network, I forget which machine I am using. Also,
18001 I tend to I lose track of where I am, and which line point is on.
18003 So I reset my mode line to look like this:
18006 -:-- foo.texi rattlesnake:/home/bob/ Line 1 (Texinfo Fill) Top
18009 I am visiting a file called @file{foo.texi}, on my machine
18010 @file{rattlesnake} in my @file{/home/bob} buffer. I am on line 1, in
18011 Texinfo mode, and am at the top of the buffer.
18014 My @file{.emacs} file has a section that looks like this:
18018 ;; Set a Mode Line that tells me which machine, which directory,
18019 ;; and which line I am on, plus the other customary information.
18020 (setq-default mode-line-format
18024 "mouse-1: select window, mouse-2: delete others ..."))
18025 mode-line-mule-info
18027 mode-line-frame-identification
18031 mode-line-buffer-identification
18034 (system-name) 0 (string-match "\\..+" (system-name))))
18039 "mouse-1: select window, mouse-2: delete others ..."))
18040 (line-number-mode " Line %l ")
18046 "mouse-1: select window, mouse-2: delete others ..."))
18047 (:eval (mode-line-mode-name))
18050 #("%n" 0 2 (help-echo "mouse-2: widen" local-map (keymap ...)))
18059 Here, I redefine the default mode line. Most of the parts are from
18060 the original; but I make a few changes. I set the @emph{default} mode
18061 line format so as to permit various modes, such as Info, to override
18064 Many elements in the list are self-explanatory:
18065 @code{mode-line-modified} is a variable that tells whether the buffer
18066 has been modified, @code{mode-name} tells the name of the mode, and so
18067 on. However, the format looks complicated because of two features we
18068 have not discussed.
18070 @cindex Properties, in mode line example
18071 The first string in the mode line is a dash or hyphen, @samp{-}. In
18072 the old days, it would have been specified simply as @code{"-"}. But
18073 nowadays, Emacs can add properties to a string, such as highlighting
18074 or, as in this case, a help feature. If you place your mouse cursor
18075 over the hyphen, some help information appears (By default, you must
18076 wait seven-tenths of a second before the information appears. You can
18077 change that timing by changing the value of @code{tooltip-delay}.)
18080 The new string format has a special syntax:
18083 #("-" 0 1 (help-echo "mouse-1: select window, ..."))
18087 The @code{#(} begins a list. The first element of the list is the
18088 string itself, just one @samp{-}. The second and third
18089 elements specify the range over which the fourth element applies. A
18090 range starts @emph{after} a character, so a zero means the range
18091 starts just before the first character; a 1 means that the range ends
18092 just after the first character. The third element is the property for
18093 the range. It consists of a property list, a
18094 property name, in this case, @samp{help-echo}, followed by a value, in this
18095 case, a string. The second, third, and fourth elements of this new
18096 string format can be repeated.
18098 @xref{Text Properties, , Text Properties, elisp, The GNU Emacs Lisp
18099 Reference Manual}, and see @ref{Mode Line Format, , Mode Line Format,
18100 elisp, The GNU Emacs Lisp Reference Manual}, for more information.
18102 @code{mode-line-buffer-identification}
18103 displays the current buffer name. It is a list
18104 beginning @code{(#("%12b" 0 4 @dots{}}.
18105 The @code{#(} begins the list.
18107 The @samp{"%12b"} displays the current buffer name, using the
18108 @code{buffer-name} function with which we are familiar; the `12'
18109 specifies the maximum number of characters that will be displayed.
18110 When a name has fewer characters, whitespace is added to fill out to
18111 this number. (Buffer names can and often should be longer than 12
18112 characters; this length works well in a typical 80 column wide
18115 @code{:eval} says to evaluate the following form and use the result as
18116 a string to display. In this case, the expression displays the first
18117 component of the full system name. The end of the first component is
18118 a @samp{.} (`period'), so I use the @code{string-match} function to
18119 tell me the length of the first component. The substring from the
18120 zeroth character to that length is the name of the machine.
18123 This is the expression:
18128 (system-name) 0 (string-match "\\..+" (system-name))))
18132 @samp{%[} and @samp{%]} cause a pair of square brackets
18133 to appear for each recursive editing level. @samp{%n} says `Narrow'
18134 when narrowing is in effect. @samp{%P} tells you the percentage of
18135 the buffer that is above the bottom of the window, or `Top', `Bottom',
18136 or `All'. (A lower case @samp{p} tell you the percentage above the
18137 @emph{top} of the window.) @samp{%-} inserts enough dashes to fill
18140 Remember, ``You don't have to like Emacs to like it''---your own
18141 Emacs can have different colors, different commands, and different
18142 keys than a default Emacs.
18144 On the other hand, if you want to bring up a plain `out of the box'
18145 Emacs, with no customization, type:
18152 This will start an Emacs that does @emph{not} load your
18153 @file{~/.emacs} initialization file. A plain, default Emacs. Nothing
18160 GNU Emacs has two debuggers, @code{debug} and @code{edebug}. The
18161 first is built into the internals of Emacs and is always with you;
18162 the second requires that you instrument a function before you can use it.
18164 Both debuggers are described extensively in @ref{Debugging, ,
18165 Debugging Lisp Programs, elisp, The GNU Emacs Lisp Reference Manual}.
18166 In this chapter, I will walk through a short example of each.
18169 * debug:: How to use the built-in debugger.
18170 * debug-on-entry:: Start debugging when you call a function.
18171 * debug-on-quit:: Start debugging when you quit with @kbd{C-g}.
18172 * edebug:: How to use Edebug, a source level debugger.
18173 * Debugging Exercises::
18177 @section @code{debug}
18180 Suppose you have written a function definition that is intended to
18181 return the sum of the numbers 1 through a given number. (This is the
18182 @code{triangle} function discussed earlier. @xref{Decrementing
18183 Example, , Example with Decrementing Counter}, for a discussion.)
18184 @c xref{Decrementing Loop,, Loop with a Decrementing Counter}, for a discussion.)
18186 However, your function definition has a bug. You have mistyped
18187 @samp{1=} for @samp{1-}. Here is the broken definition:
18189 @findex triangle-bugged
18192 (defun triangle-bugged (number)
18193 "Return sum of numbers 1 through NUMBER inclusive."
18195 (while (> number 0)
18196 (setq total (+ total number))
18197 (setq number (1= number))) ; @r{Error here.}
18202 If you are reading this in Info, you can evaluate this definition in
18203 the normal fashion. You will see @code{triangle-bugged} appear in the
18207 Now evaluate the @code{triangle-bugged} function with an
18211 (triangle-bugged 4)
18215 In a recent GNU Emacs, you will create and enter a @file{*Backtrace*}
18221 ---------- Buffer: *Backtrace* ----------
18222 Debugger entered--Lisp error: (void-function 1=)
18224 (setq number (1= number))
18225 (while (> number 0) (setq total (+ total number))
18226 (setq number (1= number)))
18227 (let ((total 0)) (while (> number 0) (setq total ...)
18228 (setq number ...)) total)
18232 eval((triangle-bugged 4))
18233 eval-last-sexp-1(nil)
18234 eval-last-sexp(nil)
18235 call-interactively(eval-last-sexp)
18236 ---------- Buffer: *Backtrace* ----------
18241 (I have reformatted this example slightly; the debugger does not fold
18242 long lines. As usual, you can quit the debugger by typing @kbd{q} in
18243 the @file{*Backtrace*} buffer.)
18245 In practice, for a bug as simple as this, the `Lisp error' line will
18246 tell you what you need to know to correct the definition. The
18247 function @code{1=} is `void'.
18251 In GNU Emacs 20 and before, you will see:
18254 Symbol's function definition is void:@: 1=
18258 which has the same meaning as the @file{*Backtrace*} buffer line in
18262 However, suppose you are not quite certain what is going on?
18263 You can read the complete backtrace.
18265 In this case, you need to run a recent GNU Emacs, which automatically
18266 starts the debugger that puts you in the @file{*Backtrace*} buffer; or
18267 else, you need to start the debugger manually as described below.
18269 Read the @file{*Backtrace*} buffer from the bottom up; it tells you
18270 what Emacs did that led to the error. Emacs made an interactive call
18271 to @kbd{C-x C-e} (@code{eval-last-sexp}), which led to the evaluation
18272 of the @code{triangle-bugged} expression. Each line above tells you
18273 what the Lisp interpreter evaluated next.
18276 The third line from the top of the buffer is
18279 (setq number (1= number))
18283 Emacs tried to evaluate this expression; in order to do so, it tried
18284 to evaluate the inner expression shown on the second line from the
18293 This is where the error occurred; as the top line says:
18296 Debugger entered--Lisp error: (void-function 1=)
18300 You can correct the mistake, re-evaluate the function definition, and
18301 then run your test again.
18303 @node debug-on-entry
18304 @section @code{debug-on-entry}
18305 @findex debug-on-entry
18307 A recent GNU Emacs starts the debugger automatically when your
18308 function has an error.
18311 GNU Emacs version 20 and before did not; it simply
18312 presented you with an error message. You had to start the debugger
18316 Incidentally, you can start the debugger manually for all versions of
18317 Emacs; the advantage is that the debugger runs even if you do not have
18318 a bug in your code. Sometimes your code will be free of bugs!
18320 You can enter the debugger when you call the function by calling
18321 @code{debug-on-entry}.
18328 M-x debug-on-entry RET triangle-bugged RET
18333 Now, evaluate the following:
18336 (triangle-bugged 5)
18340 All versions of Emacs will create a @file{*Backtrace*} buffer and tell
18341 you that it is beginning to evaluate the @code{triangle-bugged}
18346 ---------- Buffer: *Backtrace* ----------
18347 Debugger entered--entering a function:
18348 * triangle-bugged(5)
18349 eval((triangle-bugged 5))
18352 eval-last-sexp-1(nil)
18353 eval-last-sexp(nil)
18354 call-interactively(eval-last-sexp)
18355 ---------- Buffer: *Backtrace* ----------
18359 In the @file{*Backtrace*} buffer, type @kbd{d}. Emacs will evaluate
18360 the first expression in @code{triangle-bugged}; the buffer will look
18365 ---------- Buffer: *Backtrace* ----------
18366 Debugger entered--beginning evaluation of function call form:
18367 * (let ((total 0)) (while (> number 0) (setq total ...)
18368 (setq number ...)) total)
18369 * triangle-bugged(5)
18370 eval((triangle-bugged 5))
18373 eval-last-sexp-1(nil)
18374 eval-last-sexp(nil)
18375 call-interactively(eval-last-sexp)
18376 ---------- Buffer: *Backtrace* ----------
18381 Now, type @kbd{d} again, eight times, slowly. Each time you type
18382 @kbd{d}, Emacs will evaluate another expression in the function
18386 Eventually, the buffer will look like this:
18390 ---------- Buffer: *Backtrace* ----------
18391 Debugger entered--beginning evaluation of function call form:
18392 * (setq number (1= number))
18393 * (while (> number 0) (setq total (+ total number))
18394 (setq number (1= number)))
18397 * (let ((total 0)) (while (> number 0) (setq total ...)
18398 (setq number ...)) total)
18399 * triangle-bugged(5)
18400 eval((triangle-bugged 5))
18403 eval-last-sexp-1(nil)
18404 eval-last-sexp(nil)
18405 call-interactively(eval-last-sexp)
18406 ---------- Buffer: *Backtrace* ----------
18412 Finally, after you type @kbd{d} two more times, Emacs will reach the
18413 error, and the top two lines of the @file{*Backtrace*} buffer will look
18418 ---------- Buffer: *Backtrace* ----------
18419 Debugger entered--Lisp error: (void-function 1=)
18422 ---------- Buffer: *Backtrace* ----------
18426 By typing @kbd{d}, you were able to step through the function.
18428 You can quit a @file{*Backtrace*} buffer by typing @kbd{q} in it; this
18429 quits the trace, but does not cancel @code{debug-on-entry}.
18431 @findex cancel-debug-on-entry
18432 To cancel the effect of @code{debug-on-entry}, call
18433 @code{cancel-debug-on-entry} and the name of the function, like this:
18436 M-x cancel-debug-on-entry RET triangle-bugged RET
18440 (If you are reading this in Info, cancel @code{debug-on-entry} now.)
18442 @node debug-on-quit
18443 @section @code{debug-on-quit} and @code{(debug)}
18445 In addition to setting @code{debug-on-error} or calling @code{debug-on-entry},
18446 there are two other ways to start @code{debug}.
18448 @findex debug-on-quit
18449 You can start @code{debug} whenever you type @kbd{C-g}
18450 (@code{keyboard-quit}) by setting the variable @code{debug-on-quit} to
18451 @code{t}. This is useful for debugging infinite loops.
18454 @cindex @code{(debug)} in code
18455 Or, you can insert a line that says @code{(debug)} into your code
18456 where you want the debugger to start, like this:
18460 (defun triangle-bugged (number)
18461 "Return sum of numbers 1 through NUMBER inclusive."
18463 (while (> number 0)
18464 (setq total (+ total number))
18465 (debug) ; @r{Start debugger.}
18466 (setq number (1= number))) ; @r{Error here.}
18471 The @code{debug} function is described in detail in @ref{Debugger, ,
18472 The Lisp Debugger, elisp, The GNU Emacs Lisp Reference Manual}.
18475 @section The @code{edebug} Source Level Debugger
18476 @cindex Source level debugger
18479 Edebug is a source level debugger. Edebug normally displays the
18480 source of the code you are debugging, with an arrow at the left that
18481 shows which line you are currently executing.
18483 You can walk through the execution of a function, line by line, or run
18484 quickly until reaching a @dfn{breakpoint} where execution stops.
18486 Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs
18487 Lisp Reference Manual}.
18490 Here is a bugged function definition for @code{triangle-recursively}.
18491 @xref{Recursive triangle function, , Recursion in place of a counter},
18492 for a review of it.
18496 (defun triangle-recursively-bugged (number)
18497 "Return sum of numbers 1 through NUMBER inclusive.
18502 (triangle-recursively-bugged
18503 (1= number))))) ; @r{Error here.}
18508 Normally, you would install this definition by positioning your cursor
18509 after the function's closing parenthesis and typing @kbd{C-x C-e}
18510 (@code{eval-last-sexp}) or else by positioning your cursor within the
18511 definition and typing @kbd{C-M-x} (@code{eval-defun}). (By default,
18512 the @code{eval-defun} command works only in Emacs Lisp mode or in Lisp
18516 However, to prepare this function definition for Edebug, you must
18517 first @dfn{instrument} the code using a different command. You can do
18518 this by positioning your cursor within or just after the definition
18522 M-x edebug-defun RET
18526 This will cause Emacs to load Edebug automatically if it is not
18527 already loaded, and properly instrument the function.
18529 After instrumenting the function, place your cursor after the
18530 following expression and type @kbd{C-x C-e} (@code{eval-last-sexp}):
18533 (triangle-recursively-bugged 3)
18537 You will be jumped back to the source for
18538 @code{triangle-recursively-bugged} and the cursor positioned at the
18539 beginning of the @code{if} line of the function. Also, you will see
18540 an arrowhead at the left hand side of that line. The arrowhead marks
18541 the line where the function is executing. (In the following examples,
18542 we show the arrowhead with @samp{=>}; in a windowing system, you may
18543 see the arrowhead as a solid triangle in the window `fringe'.)
18546 =>@point{}(if (= number 1)
18551 In the example, the location of point is displayed with a star,
18552 @samp{@point{}} (in Info, it is displayed as @samp{-!-}).
18555 In the example, the location of point is displayed as @samp{@point{}}
18556 (in a printed book, it is displayed with a five pointed star).
18559 If you now press @key{SPC}, point will move to the next expression to
18560 be executed; the line will look like this:
18563 =>(if @point{}(= number 1)
18567 As you continue to press @key{SPC}, point will move from expression to
18568 expression. At the same time, whenever an expression returns a value,
18569 that value will be displayed in the echo area. For example, after you
18570 move point past @code{number}, you will see the following:
18573 Result: 3 (#o3, #x3, ?\C-c)
18577 This means the value of @code{number} is 3, which is octal three,
18578 hexadecimal three, and @sc{ascii} `control-c' (the third letter of the
18579 alphabet, in case you need to know this information).
18581 You can continue moving through the code until you reach the line with
18582 the error. Before evaluation, that line looks like this:
18585 => @point{}(1= number))))) ; @r{Error here.}
18590 When you press @key{SPC} once again, you will produce an error message
18594 Symbol's function definition is void:@: 1=
18600 Press @kbd{q} to quit Edebug.
18602 To remove instrumentation from a function definition, simply
18603 re-evaluate it with a command that does not instrument it.
18604 For example, you could place your cursor after the definition's
18605 closing parenthesis and type @kbd{C-x C-e}.
18607 Edebug does a great deal more than walk with you through a function.
18608 You can set it so it races through on its own, stopping only at an
18609 error or at specified stopping points; you can cause it to display the
18610 changing values of various expressions; you can find out how many
18611 times a function is called, and more.
18613 Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs
18614 Lisp Reference Manual}.
18617 @node Debugging Exercises
18618 @section Debugging Exercises
18622 Install the @code{@value{COUNT-WORDS}} function and then cause it to
18623 enter the built-in debugger when you call it. Run the command on a
18624 region containing two words. You will need to press @kbd{d} a
18625 remarkable number of times. On your system, is a `hook' called after
18626 the command finishes? (For information on hooks, see @ref{Command
18627 Overview, , Command Loop Overview, elisp, The GNU Emacs Lisp Reference
18631 Copy @code{@value{COUNT-WORDS}} into the @file{*scratch*} buffer,
18632 instrument the function for Edebug, and walk through its execution.
18633 The function does not need to have a bug, although you can introduce
18634 one if you wish. If the function lacks a bug, the walk-through
18635 completes without problems.
18638 While running Edebug, type @kbd{?} to see a list of all the Edebug commands.
18639 (The @code{global-edebug-prefix} is usually @kbd{C-x X}, i.e.,
18640 @kbd{@key{CTRL}-x} followed by an upper case @kbd{X}; use this prefix
18641 for commands made outside of the Edebug debugging buffer.)
18644 In the Edebug debugging buffer, use the @kbd{p}
18645 (@code{edebug-bounce-point}) command to see where in the region the
18646 @code{@value{COUNT-WORDS}} is working.
18649 Move point to some spot further down the function and then type the
18650 @kbd{h} (@code{edebug-goto-here}) command to jump to that location.
18653 Use the @kbd{t} (@code{edebug-trace-mode}) command to cause Edebug to
18654 walk through the function on its own; use an upper case @kbd{T} for
18655 @code{edebug-Trace-fast-mode}.
18658 Set a breakpoint, then run Edebug in Trace mode until it reaches the
18663 @chapter Conclusion
18665 We have now reached the end of this Introduction. You have now
18666 learned enough about programming in Emacs Lisp to set values, to write
18667 simple @file{.emacs} files for yourself and your friends, and write
18668 simple customizations and extensions to Emacs.
18670 This is a place to stop. Or, if you wish, you can now go onward, and
18673 You have learned some of the basic nuts and bolts of programming. But
18674 only some. There are a great many more brackets and hinges that are
18675 easy to use that we have not touched.
18677 A path you can follow right now lies among the sources to GNU Emacs
18680 @cite{The GNU Emacs Lisp Reference Manual}.
18683 @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU
18684 Emacs Lisp Reference Manual}.
18687 The Emacs Lisp sources are an adventure. When you read the sources and
18688 come across a function or expression that is unfamiliar, you need to
18689 figure out or find out what it does.
18691 Go to the Reference Manual. It is a thorough, complete, and fairly
18692 easy-to-read description of Emacs Lisp. It is written not only for
18693 experts, but for people who know what you know. (The @cite{Reference
18694 Manual} comes with the standard GNU Emacs distribution. Like this
18695 introduction, it comes as a Texinfo source file, so you can read it
18696 on-line and as a typeset, printed book.)
18698 Go to the other on-line help that is part of GNU Emacs: the on-line
18699 documentation for all functions and variables, and @code{find-tag},
18700 the program that takes you to sources.
18702 Here is an example of how I explore the sources. Because of its name,
18703 @file{simple.el} is the file I looked at first, a long time ago. As
18704 it happens some of the functions in @file{simple.el} are complicated,
18705 or at least look complicated at first sight. The @code{open-line}
18706 function, for example, looks complicated.
18708 You may want to walk through this function slowly, as we did with the
18709 @code{forward-sentence} function. (@xref{forward-sentence, The
18710 @code{forward-sentence} function}.) Or you may want to skip that
18711 function and look at another, such as @code{split-line}. You don't
18712 need to read all the functions. According to
18713 @code{count-words-in-defun}, the @code{split-line} function contains
18714 102 words and symbols.
18716 Even though it is short, @code{split-line} contains expressions
18717 we have not studied: @code{skip-chars-forward}, @code{indent-to},
18718 @code{current-column} and @code{insert-and-inherit}.
18720 Consider the @code{skip-chars-forward} function. (It is part of the
18721 function definition for @code{back-to-indentation}, which is shown in
18722 @ref{Review, , Review}.)
18724 In GNU Emacs, you can find out more about @code{skip-chars-forward} by
18725 typing @kbd{C-h f} (@code{describe-function}) and the name of the
18726 function. This gives you the function documentation.
18728 You may be able to guess what is done by a well named function such as
18729 @code{indent-to}; or you can look it up, too. Incidentally, the
18730 @code{describe-function} function itself is in @file{help.el}; it is
18731 one of those long, but decipherable functions. You can look up
18732 @code{describe-function} using the @kbd{C-h f} command!
18734 In this instance, since the code is Lisp, the @file{*Help*} buffer
18735 contains the name of the library containing the function's source.
18736 You can put point over the name of the library and press the RET key,
18737 which in this situation is bound to @code{help-follow}, and be taken
18738 directly to the source, in the same way as @kbd{M-.}
18741 The definition for @code{describe-function} illustrates how to
18742 customize the @code{interactive} expression without using the standard
18743 character codes; and it shows how to create a temporary buffer.
18745 (The @code{indent-to} function is written in C rather than Emacs Lisp;
18746 it is a `built-in' function. @code{help-follow} takes you to its
18747 source as does @code{find-tag}, when properly set up.)
18749 You can look at a function's source using @code{find-tag}, which is
18750 bound to @kbd{M-.} Finally, you can find out what the Reference
18751 Manual has to say by visiting the manual in Info, and typing @kbd{i}
18752 (@code{Info-index}) and the name of the function, or by looking up the
18753 function in the index to a printed copy of the manual.
18755 Similarly, you can find out what is meant by
18756 @code{insert-and-inherit}.
18758 Other interesting source files include @file{paragraphs.el},
18759 @file{loaddefs.el}, and @file{loadup.el}. The @file{paragraphs.el}
18760 file includes short, easily understood functions as well as longer
18761 ones. The @file{loaddefs.el} file contains the many standard
18762 autoloads and many keymaps. I have never looked at it all; only at
18763 parts. @file{loadup.el} is the file that loads the standard parts of
18764 Emacs; it tells you a great deal about how Emacs is built.
18765 (@xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp
18766 Reference Manual}, for more about building.)
18768 As I said, you have learned some nuts and bolts; however, and very
18769 importantly, we have hardly touched major aspects of programming; I
18770 have said nothing about how to sort information, except to use the
18771 predefined @code{sort} function; I have said nothing about how to store
18772 information, except to use variables and lists; I have said nothing
18773 about how to write programs that write programs. These are topics for
18774 another, and different kind of book, a different kind of learning.
18776 What you have done is learn enough for much practical work with GNU
18777 Emacs. What you have done is get started. This is the end of a
18780 @c ================ Appendix ================
18783 @appendix The @code{the-the} Function
18785 @cindex Duplicated words function
18786 @cindex Words, duplicated
18788 Sometimes when you you write text, you duplicate words---as with ``you
18789 you'' near the beginning of this sentence. I find that most
18790 frequently, I duplicate ``the''; hence, I call the function for
18791 detecting duplicated words, @code{the-the}.
18794 As a first step, you could use the following regular expression to
18795 search for duplicates:
18798 \\(\\w+[ \t\n]+\\)\\1
18802 This regexp matches one or more word-constituent characters followed
18803 by one or more spaces, tabs, or newlines. However, it does not detect
18804 duplicated words on different lines, since the ending of the first
18805 word, the end of the line, is different from the ending of the second
18806 word, a space. (For more information about regular expressions, see
18807 @ref{Regexp Search, , Regular Expression Searches}, as well as
18808 @ref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs
18809 Manual}, and @ref{Regular Expressions, , Regular Expressions, elisp,
18810 The GNU Emacs Lisp Reference Manual}.)
18812 You might try searching just for duplicated word-constituent
18813 characters but that does not work since the pattern detects doubles
18814 such as the two occurrences of `th' in `with the'.
18816 Another possible regexp searches for word-constituent characters
18817 followed by non-word-constituent characters, reduplicated. Here,
18818 @w{@samp{\\w+}} matches one or more word-constituent characters and
18819 @w{@samp{\\W*}} matches zero or more non-word-constituent characters.
18822 \\(\\(\\w+\\)\\W*\\)\\1
18828 Here is the pattern that I use. It is not perfect, but good enough.
18829 @w{@samp{\\b}} matches the empty string, provided it is at the beginning
18830 or end of a word; @w{@samp{[^@@ \n\t]+}} matches one or more occurrences of
18831 any characters that are @emph{not} an @@-sign, space, newline, or tab.
18834 \\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b
18837 One can write more complicated expressions, but I found that this
18838 expression is good enough, so I use it.
18840 Here is the @code{the-the} function, as I include it in my
18841 @file{.emacs} file, along with a handy global key binding:
18846 "Search forward for for a duplicated word."
18848 (message "Searching for for duplicated words ...")
18852 ;; This regexp is not perfect
18853 ;; but is fairly good over all:
18854 (if (re-search-forward
18855 "\\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b" nil 'move)
18856 (message "Found duplicated word.")
18857 (message "End of buffer")))
18861 ;; Bind `the-the' to C-c \
18862 (global-set-key "\C-c\\" 'the-the)
18871 one two two three four five
18876 You can substitute the other regular expressions shown above in the
18877 function definition and try each of them on this list.
18880 @appendix Handling the Kill Ring
18881 @cindex Kill ring handling
18882 @cindex Handling the kill ring
18883 @cindex Ring, making a list like a
18885 The kill ring is a list that is transformed into a ring by the
18886 workings of the @code{current-kill} function. The @code{yank} and
18887 @code{yank-pop} commands use the @code{current-kill} function.
18889 This appendix describes the @code{current-kill} function as well as
18890 both the @code{yank} and the @code{yank-pop} commands, but first,
18891 consider the workings of the kill ring.
18894 * What the Kill Ring Does::
18896 * yank:: Paste a copy of a clipped element.
18897 * yank-pop:: Insert element pointed to.
18902 @node What the Kill Ring Does
18903 @unnumberedsec What the Kill Ring Does
18907 The kill ring has a default maximum length of sixty items; this number
18908 is too large for an explanation. Instead, set it to four. Please
18909 evaluate the following:
18913 (setq old-kill-ring-max kill-ring-max)
18914 (setq kill-ring-max 4)
18919 Then, please copy each line of the following indented example into the
18920 kill ring. You may kill each line with @kbd{C-k} or mark it and copy
18924 (In a read-only buffer, such as the @file{*info*} buffer, the kill
18925 command, @kbd{C-k} (@code{kill-line}), will not remove the text,
18926 merely copy it to the kill ring. However, your machine may beep at
18927 you. Alternatively, for silence, you may copy the region of each line
18928 with the @kbd{M-w} (@code{kill-ring-save}) command. You must mark
18929 each line for this command to succeed, but it does not matter at which
18930 end you put point or mark.)
18934 Please invoke the calls in order, so that five elements attempt to
18935 fill the kill ring:
18940 second piece of text
18942 fourth line of text
18949 Then find the value of @code{kill-ring} by evaluating
18961 ("fifth bit of text" "fourth line of text"
18962 "third line" "second piece of text")
18967 The first element, @samp{first some text}, was dropped.
18970 To return to the old value for the length of the kill ring, evaluate:
18973 (setq kill-ring-max old-kill-ring-max)
18977 @appendixsec The @code{current-kill} Function
18978 @findex current-kill
18980 The @code{current-kill} function changes the element in the kill ring
18981 to which @code{kill-ring-yank-pointer} points. (Also, the
18982 @code{kill-new} function sets @code{kill-ring-yank-pointer} to point
18983 to the latest element of the kill ring. The @code{kill-new}
18984 function is used directly or indirectly by @code{kill-append},
18985 @code{copy-region-as-kill}, @code{kill-ring-save}, @code{kill-line},
18986 and @code{kill-region}.)
18989 * Code for current-kill::
18990 * Understanding current-kill::
18994 @node Code for current-kill
18995 @unnumberedsubsec The code for @code{current-kill}
19000 The @code{current-kill} function is used by @code{yank} and by
19001 @code{yank-pop}. Here is the code for @code{current-kill}:
19005 (defun current-kill (n &optional do-not-move)
19006 "Rotate the yanking point by N places, and then return that kill.
19007 If N is zero, `interprogram-paste-function' is set, and calling it
19008 returns a string, then that string is added to the front of the
19009 kill ring and returned as the latest kill.
19012 If optional arg DO-NOT-MOVE is non-nil, then don't actually move the
19013 yanking point; just return the Nth kill forward."
19014 (let ((interprogram-paste (and (= n 0)
19015 interprogram-paste-function
19016 (funcall interprogram-paste-function))))
19019 (if interprogram-paste
19021 ;; Disable the interprogram cut function when we add the new
19022 ;; text to the kill ring, so Emacs doesn't try to own the
19023 ;; selection, with identical text.
19024 (let ((interprogram-cut-function nil))
19025 (kill-new interprogram-paste))
19026 interprogram-paste)
19029 (or kill-ring (error "Kill ring is empty"))
19030 (let ((ARGth-kill-element
19031 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19032 (length kill-ring))
19035 (setq kill-ring-yank-pointer ARGth-kill-element))
19036 (car ARGth-kill-element)))))
19040 Remember also that the @code{kill-new} function sets
19041 @code{kill-ring-yank-pointer} to the latest element of the kill
19042 ring, which means that all the functions that call it set the value
19043 indirectly: @code{kill-append}, @code{copy-region-as-kill},
19044 @code{kill-ring-save}, @code{kill-line}, and @code{kill-region}.
19047 Here is the line in @code{kill-new}, which is explained in
19048 @ref{kill-new function, , The @code{kill-new} function}.
19051 (setq kill-ring-yank-pointer kill-ring)
19055 @node Understanding current-kill
19056 @unnumberedsubsec @code{current-kill} in Outline
19059 The @code{current-kill} function looks complex, but as usual, it can
19060 be understood by taking it apart piece by piece. First look at it in
19065 (defun current-kill (n &optional do-not-move)
19066 "Rotate the yanking point by N places, and then return that kill."
19072 This function takes two arguments, one of which is optional. It has a
19073 documentation string. It is @emph{not} interactive.
19076 * Body of current-kill::
19077 * Digression concerning error:: How to mislead humans, but not computers.
19078 * Determining the Element::
19082 @node Body of current-kill
19083 @unnumberedsubsubsec The Body of @code{current-kill}
19086 The body of the function definition is a @code{let} expression, which
19087 itself has a body as well as a @var{varlist}.
19089 The @code{let} expression declares a variable that will be only usable
19090 within the bounds of this function. This variable is called
19091 @code{interprogram-paste} and is for copying to another program. It
19092 is not for copying within this instance of GNU Emacs. Most window
19093 systems provide a facility for interprogram pasting. Sadly, that
19094 facility usually provides only for the last element. Most windowing
19095 systems have not adopted a ring of many possibilities, even though
19096 Emacs has provided it for decades.
19098 The @code{if} expression has two parts, one if there exists
19099 @code{interprogram-paste} and one if not.
19102 Let us consider the `if not' or else-part of the @code{current-kill}
19103 function. (The then-part uses the @code{kill-new} function, which
19104 we have already described. @xref{kill-new function, , The
19105 @code{kill-new} function}.)
19109 (or kill-ring (error "Kill ring is empty"))
19110 (let ((ARGth-kill-element
19111 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19112 (length kill-ring))
19115 (setq kill-ring-yank-pointer ARGth-kill-element))
19116 (car ARGth-kill-element))
19121 The code first checks whether the kill ring has content; otherwise it
19125 Note that the @code{or} expression is very similar to testing length
19132 (if (zerop (length kill-ring)) ; @r{if-part}
19133 (error "Kill ring is empty")) ; @r{then-part}
19139 If there is not anything in the kill ring, its length must be zero and
19140 an error message sent to the user: @samp{Kill ring is empty}. The
19141 @code{current-kill} function uses an @code{or} expression which is
19142 simpler. But an @code{if} expression reminds us what goes on.
19144 This @code{if} expression uses the function @code{zerop} which returns
19145 true if the value it is testing is zero. When @code{zerop} tests
19146 true, the then-part of the @code{if} is evaluated. The then-part is a
19147 list starting with the function @code{error}, which is a function that
19148 is similar to the @code{message} function
19149 (@pxref{message, , The @code{message} Function}) in that
19150 it prints a one-line message in the echo area. However, in addition
19151 to printing a message, @code{error} also stops evaluation of the
19152 function within which it is embedded. This means that the rest of the
19153 function will not be evaluated if the length of the kill ring is zero.
19155 Then the @code{current-kill} function selects the element to return.
19156 The selection depends on the number of places that @code{current-kill}
19157 rotates and on where @code{kill-ring-yank-pointer} points.
19159 Next, either the optional @code{do-not-move} argument is true or the
19160 current value of @code{kill-ring-yank-pointer} is set to point to the
19161 list. Finally, another expression returns the first element of the
19162 list even if the @code{do-not-move} argument is true.
19165 @node Digression concerning error
19166 @unnumberedsubsubsec Digression about the word `error'
19169 In my opinion, it is slightly misleading, at least to humans, to use
19170 the term `error' as the name of the @code{error} function. A better
19171 term would be `cancel'. Strictly speaking, of course, you cannot
19172 point to, much less rotate a pointer to a list that has no length, so
19173 from the point of view of the computer, the word `error' is correct.
19174 But a human expects to attempt this sort of thing, if only to find out
19175 whether the kill ring is full or empty. This is an act of
19178 From the human point of view, the act of exploration and discovery is
19179 not necessarily an error, and therefore should not be labeled as one,
19180 even in the bowels of a computer. As it is, the code in Emacs implies
19181 that a human who is acting virtuously, by exploring his or her
19182 environment, is making an error. This is bad. Even though the computer
19183 takes the same steps as it does when there is an `error', a term such as
19184 `cancel' would have a clearer connotation.
19187 @node Determining the Element
19188 @unnumberedsubsubsec Determining the Element
19191 Among other actions, the else-part of the @code{if} expression sets
19192 the value of @code{kill-ring-yank-pointer} to
19193 @code{ARGth-kill-element} when the kill ring has something in it and
19194 the value of @code{do-not-move} is @code{nil}.
19197 The code looks like this:
19201 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19202 (length kill-ring))
19207 This needs some examination. Unless it is not supposed to move the
19208 pointer, the @code{current-kill} function changes where
19209 @code{kill-ring-yank-pointer} points.
19211 @w{@code{(setq kill-ring-yank-pointer ARGth-kill-element))}}
19212 expression does. Also, clearly, @code{ARGth-kill-element} is being
19213 set to be equal to some @sc{cdr} of the kill ring, using the
19214 @code{nthcdr} function that is described in an earlier section.
19215 (@xref{copy-region-as-kill}.) How does it do this?
19217 As we have seen before (@pxref{nthcdr}), the @code{nthcdr} function
19218 works by repeatedly taking the @sc{cdr} of a list---it takes the
19219 @sc{cdr} of the @sc{cdr} of the @sc{cdr} @dots{}
19222 The two following expressions produce the same result:
19226 (setq kill-ring-yank-pointer (cdr kill-ring))
19228 (setq kill-ring-yank-pointer (nthcdr 1 kill-ring))
19232 However, the @code{nthcdr} expression is more complicated. It uses
19233 the @code{mod} function to determine which @sc{cdr} to select.
19235 (You will remember to look at inner functions first; indeed, we will
19236 have to go inside the @code{mod}.)
19238 The @code{mod} function returns the value of its first argument modulo
19239 the second; that is to say, it returns the remainder after dividing
19240 the first argument by the second. The value returned has the same
19241 sign as the second argument.
19249 @result{} 0 ;; @r{because there is no remainder}
19256 In this case, the first argument is often smaller than the second.
19268 We can guess what the @code{-} function does. It is like @code{+} but
19269 subtracts instead of adds; the @code{-} function subtracts its second
19270 argument from its first. Also, we already know what the @code{length}
19271 function does (@pxref{length}). It returns the length of a list.
19273 And @code{n} is the name of the required argument to the
19274 @code{current-kill} function.
19277 So when the first argument to @code{nthcdr} is zero, the @code{nthcdr}
19278 expression returns the whole list, as you can see by evaluating the
19283 ;; kill-ring-yank-pointer @r{and} kill-ring @r{have a length of four}
19284 ;; @r{and} (mod (- 0 4) 4) @result{} 0
19285 (nthcdr (mod (- 0 4) 4)
19286 '("fourth line of text"
19288 "second piece of text"
19289 "first some text"))
19294 When the first argument to the @code{current-kill} function is one,
19295 the @code{nthcdr} expression returns the list without its first
19300 (nthcdr (mod (- 1 4) 4)
19301 '("fourth line of text"
19303 "second piece of text"
19304 "first some text"))
19308 @cindex @samp{global variable} defined
19309 @cindex @samp{variable, global}, defined
19310 Incidentally, both @code{kill-ring} and @code{kill-ring-yank-pointer}
19311 are @dfn{global variables}. That means that any expression in Emacs
19312 Lisp can access them. They are not like the local variables set by
19313 @code{let} or like the symbols in an argument list.
19314 Local variables can only be accessed
19315 within the @code{let} that defines them or the function that specifies
19316 them in an argument list (and within expressions called by them).
19319 @c texi2dvi fails when the name of the section is within ifnottex ...
19320 (@xref{Prevent confusion, , @code{let} Prevents Confusion}, and
19321 @ref{defun, , The @code{defun} Macro}.)
19325 @appendixsec @code{yank}
19328 After learning about @code{current-kill}, the code for the
19329 @code{yank} function is almost easy.
19331 The @code{yank} function does not use the
19332 @code{kill-ring-yank-pointer} variable directly. It calls
19333 @code{insert-for-yank} which calls @code{current-kill} which sets the
19334 @code{kill-ring-yank-pointer} variable.
19337 The code looks like this:
19342 (defun yank (&optional arg)
19343 "Reinsert (\"paste\") the last stretch of killed text.
19344 More precisely, reinsert the stretch of killed text most recently
19345 killed OR yanked. Put point at end, and set mark at beginning.
19346 With just \\[universal-argument] as argument, same but put point at
19347 beginning (and mark at end). With argument N, reinsert the Nth most
19348 recently killed stretch of killed text.
19350 When this command inserts killed text into the buffer, it honors
19351 `yank-excluded-properties' and `yank-handler' as described in the
19352 doc string for `insert-for-yank-1', which see.
19354 See also the command \\[yank-pop]."
19358 (setq yank-window-start (window-start))
19359 ;; If we don't get all the way thru, make last-command indicate that
19360 ;; for the following command.
19361 (setq this-command t)
19362 (push-mark (point))
19365 (insert-for-yank (current-kill (cond
19370 ;; This is like exchange-point-and-mark,
19371 ;; but doesn't activate the mark.
19372 ;; It is cleaner to avoid activation, even though the command
19373 ;; loop would deactivate the mark because we inserted text.
19374 (goto-char (prog1 (mark t)
19375 (set-marker (mark-marker) (point) (current-buffer)))))
19378 ;; If we do get all the way thru, make this-command indicate that.
19379 (if (eq this-command t)
19380 (setq this-command 'yank))
19385 The key expression is @code{insert-for-yank}, which inserts the string
19386 returned by @code{current-kill}, but removes some text properties from
19389 However, before getting to that expression, the function sets the value
19390 of @code{yank-window-start} to the position returned by the
19391 @code{(window-start)} expression, the position at which the display
19392 currently starts. The @code{yank} function also sets
19393 @code{this-command} and pushes the mark.
19395 After it yanks the appropriate element, if the optional argument is a
19396 @sc{cons} rather than a number or nothing, it puts point at beginning
19397 of the yanked text and mark at its end.
19399 (The @code{prog1} function is like @code{progn} but returns the value
19400 of its first argument rather than the value of its last argument. Its
19401 first argument is forced to return the buffer's mark as an integer.
19402 You can see the documentation for these functions by placing point
19403 over them in this buffer and then typing @kbd{C-h f}
19404 (@code{describe-function}) followed by a @kbd{RET}; the default is the
19407 The last part of the function tells what to do when it succeeds.
19410 @appendixsec @code{yank-pop}
19413 After understanding @code{yank} and @code{current-kill}, you know how
19414 to approach the @code{yank-pop} function. Leaving out the
19415 documentation to save space, it looks like this:
19420 (defun yank-pop (&optional arg)
19423 (if (not (eq last-command 'yank))
19424 (error "Previous command was not a yank"))
19427 (setq this-command 'yank)
19428 (unless arg (setq arg 1))
19429 (let ((inhibit-read-only t)
19430 (before (< (point) (mark t))))
19434 (funcall (or yank-undo-function 'delete-region) (point) (mark t))
19435 (funcall (or yank-undo-function 'delete-region) (mark t) (point)))
19436 (setq yank-undo-function nil)
19439 (set-marker (mark-marker) (point) (current-buffer))
19440 (insert-for-yank (current-kill arg))
19441 ;; Set the window start back where it was in the yank command,
19443 (set-window-start (selected-window) yank-window-start t)
19447 ;; This is like exchange-point-and-mark,
19448 ;; but doesn't activate the mark.
19449 ;; It is cleaner to avoid activation, even though the command
19450 ;; loop would deactivate the mark because we inserted text.
19451 (goto-char (prog1 (mark t)
19452 (set-marker (mark-marker)
19454 (current-buffer))))))
19459 The function is interactive with a small @samp{p} so the prefix
19460 argument is processed and passed to the function. The command can
19461 only be used after a previous yank; otherwise an error message is
19462 sent. This check uses the variable @code{last-command} which is set
19463 by @code{yank} and is discussed elsewhere.
19464 (@xref{copy-region-as-kill}.)
19466 The @code{let} clause sets the variable @code{before} to true or false
19467 depending whether point is before or after mark and then the region
19468 between point and mark is deleted. This is the region that was just
19469 inserted by the previous yank and it is this text that will be
19472 @code{funcall} calls its first argument as a function, passing
19473 remaining arguments to it. The first argument is whatever the
19474 @code{or} expression returns. The two remaining arguments are the
19475 positions of point and mark set by the preceding @code{yank} command.
19477 There is more, but that is the hardest part.
19480 @appendixsec The @file{ring.el} File
19481 @cindex @file{ring.el} file
19483 Interestingly, GNU Emacs posses a file called @file{ring.el} that
19484 provides many of the features we just discussed. But functions such
19485 as @code{kill-ring-yank-pointer} do not use this library, possibly
19486 because they were written earlier.
19489 @appendix A Graph with Labeled Axes
19491 Printed axes help you understand a graph. They convey scale. In an
19492 earlier chapter (@pxref{Readying a Graph, , Readying a Graph}), we
19493 wrote the code to print the body of a graph. Here we write the code
19494 for printing and labeling vertical and horizontal axes, along with the
19498 * Labeled Example::
19499 * print-graph Varlist:: @code{let} expression in @code{print-graph}.
19500 * print-Y-axis:: Print a label for the vertical axis.
19501 * print-X-axis:: Print a horizontal label.
19502 * Print Whole Graph:: The function to print a complete graph.
19506 @node Labeled Example
19507 @unnumberedsec Labeled Example Graph
19510 Since insertions fill a buffer to the right and below point, the new
19511 graph printing function should first print the Y or vertical axis,
19512 then the body of the graph, and finally the X or horizontal axis.
19513 This sequence lays out for us the contents of the function:
19523 Print body of graph.
19530 Here is an example of how a finished graph should look:
19543 1 - ****************
19550 In this graph, both the vertical and the horizontal axes are labeled
19551 with numbers. However, in some graphs, the horizontal axis is time
19552 and would be better labeled with months, like this:
19566 Indeed, with a little thought, we can easily come up with a variety of
19567 vertical and horizontal labeling schemes. Our task could become
19568 complicated. But complications breed confusion. Rather than permit
19569 this, it is better choose a simple labeling scheme for our first
19570 effort, and to modify or replace it later.
19573 These considerations suggest the following outline for the
19574 @code{print-graph} function:
19578 (defun print-graph (numbers-list)
19579 "@var{documentation}@dots{}"
19580 (let ((height @dots{}
19584 (print-Y-axis height @dots{} )
19585 (graph-body-print numbers-list)
19586 (print-X-axis @dots{} )))
19590 We can work on each part of the @code{print-graph} function definition
19593 @node print-graph Varlist
19594 @appendixsec The @code{print-graph} Varlist
19595 @cindex @code{print-graph} varlist
19597 In writing the @code{print-graph} function, the first task is to write
19598 the varlist in the @code{let} expression. (We will leave aside for the
19599 moment any thoughts about making the function interactive or about the
19600 contents of its documentation string.)
19602 The varlist should set several values. Clearly, the top of the label
19603 for the vertical axis must be at least the height of the graph, which
19604 means that we must obtain this information here. Note that the
19605 @code{print-graph-body} function also requires this information. There
19606 is no reason to calculate the height of the graph in two different
19607 places, so we should change @code{print-graph-body} from the way we
19608 defined it earlier to take advantage of the calculation.
19610 Similarly, both the function for printing the X axis labels and the
19611 @code{print-graph-body} function need to learn the value of the width of
19612 each symbol. We can perform the calculation here and change the
19613 definition for @code{print-graph-body} from the way we defined it in the
19616 The length of the label for the horizontal axis must be at least as long
19617 as the graph. However, this information is used only in the function
19618 that prints the horizontal axis, so it does not need to be calculated here.
19620 These thoughts lead us directly to the following form for the varlist
19621 in the @code{let} for @code{print-graph}:
19625 (let ((height (apply 'max numbers-list)) ; @r{First version.}
19626 (symbol-width (length graph-blank)))
19631 As we shall see, this expression is not quite right.
19635 @appendixsec The @code{print-Y-axis} Function
19636 @cindex Axis, print vertical
19637 @cindex Y axis printing
19638 @cindex Vertical axis printing
19639 @cindex Print vertical axis
19641 The job of the @code{print-Y-axis} function is to print a label for
19642 the vertical axis that looks like this:
19660 The function should be passed the height of the graph, and then should
19661 construct and insert the appropriate numbers and marks.
19664 * print-Y-axis in Detail::
19665 * Height of label:: What height for the Y axis?
19666 * Compute a Remainder:: How to compute the remainder of a division.
19667 * Y Axis Element:: Construct a line for the Y axis.
19668 * Y-axis-column:: Generate a list of Y axis labels.
19669 * print-Y-axis Penultimate:: A not quite final version.
19673 @node print-Y-axis in Detail
19674 @unnumberedsubsec The @code{print-Y-axis} Function in Detail
19677 It is easy enough to see in the figure what the Y axis label should
19678 look like; but to say in words, and then to write a function
19679 definition to do the job is another matter. It is not quite true to
19680 say that we want a number and a tic every five lines: there are only
19681 three lines between the @samp{1} and the @samp{5} (lines 2, 3, and 4),
19682 but four lines between the @samp{5} and the @samp{10} (lines 6, 7, 8,
19683 and 9). It is better to say that we want a number and a tic mark on
19684 the base line (number 1) and then that we want a number and a tic on
19685 the fifth line from the bottom and on every line that is a multiple of
19689 @node Height of label
19690 @unnumberedsubsec What height should the label be?
19693 The next issue is what height the label should be? Suppose the maximum
19694 height of tallest column of the graph is seven. Should the highest
19695 label on the Y axis be @samp{5 -}, and should the graph stick up above
19696 the label? Or should the highest label be @samp{7 -}, and mark the peak
19697 of the graph? Or should the highest label be @code{10 -}, which is a
19698 multiple of five, and be higher than the topmost value of the graph?
19700 The latter form is preferred. Most graphs are drawn within rectangles
19701 whose sides are an integral number of steps long---5, 10, 15, and so
19702 on for a step distance of five. But as soon as we decide to use a
19703 step height for the vertical axis, we discover that the simple
19704 expression in the varlist for computing the height is wrong. The
19705 expression is @code{(apply 'max numbers-list)}. This returns the
19706 precise height, not the maximum height plus whatever is necessary to
19707 round up to the nearest multiple of five. A more complex expression
19710 As usual in cases like this, a complex problem becomes simpler if it is
19711 divided into several smaller problems.
19713 First, consider the case when the highest value of the graph is an
19714 integral multiple of five---when it is 5, 10, 15, or some higher
19715 multiple of five. We can use this value as the Y axis height.
19717 A fairly simply way to determine whether a number is a multiple of
19718 five is to divide it by five and see if the division results in a
19719 remainder. If there is no remainder, the number is a multiple of
19720 five. Thus, seven divided by five has a remainder of two, and seven
19721 is not an integral multiple of five. Put in slightly different
19722 language, more reminiscent of the classroom, five goes into seven
19723 once, with a remainder of two. However, five goes into ten twice,
19724 with no remainder: ten is an integral multiple of five.
19726 @node Compute a Remainder
19727 @appendixsubsec Side Trip: Compute a Remainder
19729 @findex % @r{(remainder function)}
19730 @cindex Remainder function, @code{%}
19731 In Lisp, the function for computing a remainder is @code{%}. The
19732 function returns the remainder of its first argument divided by its
19733 second argument. As it happens, @code{%} is a function in Emacs Lisp
19734 that you cannot discover using @code{apropos}: you find nothing if you
19735 type @kbd{M-x apropos @key{RET} remainder @key{RET}}. The only way to
19736 learn of the existence of @code{%} is to read about it in a book such
19737 as this or in the Emacs Lisp sources.
19739 You can try the @code{%} function by evaluating the following two
19751 The first expression returns 2 and the second expression returns 0.
19753 To test whether the returned value is zero or some other number, we
19754 can use the @code{zerop} function. This function returns @code{t} if
19755 its argument, which must be a number, is zero.
19767 Thus, the following expression will return @code{t} if the height
19768 of the graph is evenly divisible by five:
19771 (zerop (% height 5))
19775 (The value of @code{height}, of course, can be found from @code{(apply
19776 'max numbers-list)}.)
19778 On the other hand, if the value of @code{height} is not a multiple of
19779 five, we want to reset the value to the next higher multiple of five.
19780 This is straightforward arithmetic using functions with which we are
19781 already familiar. First, we divide the value of @code{height} by five
19782 to determine how many times five goes into the number. Thus, five
19783 goes into twelve twice. If we add one to this quotient and multiply by
19784 five, we will obtain the value of the next multiple of five that is
19785 larger than the height. Five goes into twelve twice. Add one to two,
19786 and multiply by five; the result is fifteen, which is the next multiple
19787 of five that is higher than twelve. The Lisp expression for this is:
19790 (* (1+ (/ height 5)) 5)
19794 For example, if you evaluate the following, the result is 15:
19797 (* (1+ (/ 12 5)) 5)
19800 All through this discussion, we have been using `five' as the value
19801 for spacing labels on the Y axis; but we may want to use some other
19802 value. For generality, we should replace `five' with a variable to
19803 which we can assign a value. The best name I can think of for this
19804 variable is @code{Y-axis-label-spacing}.
19807 Using this term, and an @code{if} expression, we produce the
19812 (if (zerop (% height Y-axis-label-spacing))
19815 (* (1+ (/ height Y-axis-label-spacing))
19816 Y-axis-label-spacing))
19821 This expression returns the value of @code{height} itself if the height
19822 is an even multiple of the value of the @code{Y-axis-label-spacing} or
19823 else it computes and returns a value of @code{height} that is equal to
19824 the next higher multiple of the value of the @code{Y-axis-label-spacing}.
19826 We can now include this expression in the @code{let} expression of the
19827 @code{print-graph} function (after first setting the value of
19828 @code{Y-axis-label-spacing}):
19829 @vindex Y-axis-label-spacing
19833 (defvar Y-axis-label-spacing 5
19834 "Number of lines from one Y axis label to next.")
19839 (let* ((height (apply 'max numbers-list))
19840 (height-of-top-line
19841 (if (zerop (% height Y-axis-label-spacing))
19846 (* (1+ (/ height Y-axis-label-spacing))
19847 Y-axis-label-spacing)))
19848 (symbol-width (length graph-blank))))
19854 (Note use of the @code{let*} function: the initial value of height is
19855 computed once by the @code{(apply 'max numbers-list)} expression and
19856 then the resulting value of @code{height} is used to compute its
19857 final value. @xref{fwd-para let, , The @code{let*} expression}, for
19858 more about @code{let*}.)
19860 @node Y Axis Element
19861 @appendixsubsec Construct a Y Axis Element
19863 When we print the vertical axis, we want to insert strings such as
19864 @w{@samp{5 -}} and @w{@samp{10 - }} every five lines.
19865 Moreover, we want the numbers and dashes to line up, so shorter
19866 numbers must be padded with leading spaces. If some of the strings
19867 use two digit numbers, the strings with single digit numbers must
19868 include a leading blank space before the number.
19870 @findex number-to-string
19871 To figure out the length of the number, the @code{length} function is
19872 used. But the @code{length} function works only with a string, not with
19873 a number. So the number has to be converted from being a number to
19874 being a string. This is done with the @code{number-to-string} function.
19879 (length (number-to-string 35))
19882 (length (number-to-string 100))
19888 (@code{number-to-string} is also called @code{int-to-string}; you will
19889 see this alternative name in various sources.)
19891 In addition, in each label, each number is followed by a string such
19892 as @w{@samp{ - }}, which we will call the @code{Y-axis-tic} marker.
19893 This variable is defined with @code{defvar}:
19898 (defvar Y-axis-tic " - "
19899 "String that follows number in a Y axis label.")
19903 The length of the Y label is the sum of the length of the Y axis tic
19904 mark and the length of the number of the top of the graph.
19907 (length (concat (number-to-string height) Y-axis-tic)))
19910 This value will be calculated by the @code{print-graph} function in
19911 its varlist as @code{full-Y-label-width} and passed on. (Note that we
19912 did not think to include this in the varlist when we first proposed it.)
19914 To make a complete vertical axis label, a tic mark is concatenated
19915 with a number; and the two together may be preceded by one or more
19916 spaces depending on how long the number is. The label consists of
19917 three parts: the (optional) leading spaces, the number, and the tic
19918 mark. The function is passed the value of the number for the specific
19919 row, and the value of the width of the top line, which is calculated
19920 (just once) by @code{print-graph}.
19924 (defun Y-axis-element (number full-Y-label-width)
19925 "Construct a NUMBERed label element.
19926 A numbered element looks like this ` 5 - ',
19927 and is padded as needed so all line up with
19928 the element for the largest number."
19931 (let* ((leading-spaces
19932 (- full-Y-label-width
19934 (concat (number-to-string number)
19939 (make-string leading-spaces ? )
19940 (number-to-string number)
19945 The @code{Y-axis-element} function concatenates together the leading
19946 spaces, if any; the number, as a string; and the tic mark.
19948 To figure out how many leading spaces the label will need, the
19949 function subtracts the actual length of the label---the length of the
19950 number plus the length of the tic mark---from the desired label width.
19952 @findex make-string
19953 Blank spaces are inserted using the @code{make-string} function. This
19954 function takes two arguments: the first tells it how long the string
19955 will be and the second is a symbol for the character to insert, in a
19956 special format. The format is a question mark followed by a blank
19957 space, like this, @samp{? }. @xref{Character Type, , Character Type,
19958 elisp, The GNU Emacs Lisp Reference Manual}, for a description of the
19959 syntax for characters. (Of course, you might want to replace the
19960 blank space by some other character @dots{} You know what to do.)
19962 The @code{number-to-string} function is used in the concatenation
19963 expression, to convert the number to a string that is concatenated
19964 with the leading spaces and the tic mark.
19966 @node Y-axis-column
19967 @appendixsubsec Create a Y Axis Column
19969 The preceding functions provide all the tools needed to construct a
19970 function that generates a list of numbered and blank strings to insert
19971 as the label for the vertical axis:
19973 @findex Y-axis-column
19976 (defun Y-axis-column (height width-of-label)
19977 "Construct list of Y axis labels and blank strings.
19978 For HEIGHT of line above base and WIDTH-OF-LABEL."
19982 (while (> height 1)
19983 (if (zerop (% height Y-axis-label-spacing))
19984 ;; @r{Insert label.}
19987 (Y-axis-element height width-of-label)
19991 ;; @r{Else, insert blanks.}
19994 (make-string width-of-label ? )
19996 (setq height (1- height)))
19997 ;; @r{Insert base line.}
19999 (cons (Y-axis-element 1 width-of-label) Y-axis))
20000 (nreverse Y-axis)))
20004 In this function, we start with the value of @code{height} and
20005 repetitively subtract one from its value. After each subtraction, we
20006 test to see whether the value is an integral multiple of the
20007 @code{Y-axis-label-spacing}. If it is, we construct a numbered label
20008 using the @code{Y-axis-element} function; if not, we construct a
20009 blank label using the @code{make-string} function. The base line
20010 consists of the number one followed by a tic mark.
20013 @node print-Y-axis Penultimate
20014 @appendixsubsec The Not Quite Final Version of @code{print-Y-axis}
20016 The list constructed by the @code{Y-axis-column} function is passed to
20017 the @code{print-Y-axis} function, which inserts the list as a column.
20019 @findex print-Y-axis
20022 (defun print-Y-axis (height full-Y-label-width)
20023 "Insert Y axis using HEIGHT and FULL-Y-LABEL-WIDTH.
20024 Height must be the maximum height of the graph.
20025 Full width is the width of the highest label element."
20026 ;; Value of height and full-Y-label-width
20027 ;; are passed by `print-graph'.
20030 (let ((start (point)))
20032 (Y-axis-column height full-Y-label-width))
20033 ;; @r{Place point ready for inserting graph.}
20035 ;; @r{Move point forward by value of} full-Y-label-width
20036 (forward-char full-Y-label-width)))
20040 The @code{print-Y-axis} uses the @code{insert-rectangle} function to
20041 insert the Y axis labels created by the @code{Y-axis-column} function.
20042 In addition, it places point at the correct position for printing the body of
20045 You can test @code{print-Y-axis}:
20053 Y-axis-label-spacing
20062 Copy the following expression:
20065 (print-Y-axis 12 5)
20069 Switch to the @file{*scratch*} buffer and place the cursor where you
20070 want the axis labels to start.
20073 Type @kbd{M-:} (@code{eval-expression}).
20076 Yank the @code{graph-body-print} expression into the minibuffer
20077 with @kbd{C-y} (@code{yank)}.
20080 Press @key{RET} to evaluate the expression.
20083 Emacs will print labels vertically, the top one being @w{@samp{10 -@w{
20084 }}}. (The @code{print-graph} function will pass the value of
20085 @code{height-of-top-line}, which in this case will end up as 15,
20086 thereby getting rid of what might appear as a bug.)
20090 @appendixsec The @code{print-X-axis} Function
20091 @cindex Axis, print horizontal
20092 @cindex X axis printing
20093 @cindex Print horizontal axis
20094 @cindex Horizontal axis printing
20096 X axis labels are much like Y axis labels, except that the ticks are on a
20097 line above the numbers. Labels should look like this:
20106 The first tic is under the first column of the graph and is preceded by
20107 several blank spaces. These spaces provide room in rows above for the Y
20108 axis labels. The second, third, fourth, and subsequent ticks are all
20109 spaced equally, according to the value of @code{X-axis-label-spacing}.
20111 The second row of the X axis consists of numbers, preceded by several
20112 blank spaces and also separated according to the value of the variable
20113 @code{X-axis-label-spacing}.
20115 The value of the variable @code{X-axis-label-spacing} should itself be
20116 measured in units of @code{symbol-width}, since you may want to change
20117 the width of the symbols that you are using to print the body of the
20118 graph without changing the ways the graph is labeled.
20121 * Similarities differences:: Much like @code{print-Y-axis}, but not exactly.
20122 * X Axis Tic Marks:: Create tic marks for the horizontal axis.
20126 @node Similarities differences
20127 @unnumberedsubsec Similarities and differences
20130 The @code{print-X-axis} function is constructed in more or less the
20131 same fashion as the @code{print-Y-axis} function except that it has
20132 two lines: the line of tic marks and the numbers. We will write a
20133 separate function to print each line and then combine them within the
20134 @code{print-X-axis} function.
20136 This is a three step process:
20140 Write a function to print the X axis tic marks, @code{print-X-axis-tic-line}.
20143 Write a function to print the X numbers, @code{print-X-axis-numbered-line}.
20146 Write a function to print both lines, the @code{print-X-axis} function,
20147 using @code{print-X-axis-tic-line} and
20148 @code{print-X-axis-numbered-line}.
20151 @node X Axis Tic Marks
20152 @appendixsubsec X Axis Tic Marks
20154 The first function should print the X axis tic marks. We must specify
20155 the tic marks themselves and their spacing:
20159 (defvar X-axis-label-spacing
20160 (if (boundp 'graph-blank)
20161 (* 5 (length graph-blank)) 5)
20162 "Number of units from one X axis label to next.")
20167 (Note that the value of @code{graph-blank} is set by another
20168 @code{defvar}. The @code{boundp} predicate checks whether it has
20169 already been set; @code{boundp} returns @code{nil} if it has not. If
20170 @code{graph-blank} were unbound and we did not use this conditional
20171 construction, in a recent GNU Emacs, we would enter the debugger and
20172 see an error message saying @samp{@w{Debugger entered--Lisp error:}
20173 @w{(void-variable graph-blank)}}.)
20176 Here is the @code{defvar} for @code{X-axis-tic-symbol}:
20180 (defvar X-axis-tic-symbol "|"
20181 "String to insert to point to a column in X axis.")
20186 The goal is to make a line that looks like this:
20192 The first tic is indented so that it is under the first column, which is
20193 indented to provide space for the Y axis labels.
20195 A tic element consists of the blank spaces that stretch from one tic to
20196 the next plus a tic symbol. The number of blanks is determined by the
20197 width of the tic symbol and the @code{X-axis-label-spacing}.
20200 The code looks like this:
20204 ;;; X-axis-tic-element
20208 ;; @r{Make a string of blanks.}
20209 (- (* symbol-width X-axis-label-spacing)
20210 (length X-axis-tic-symbol))
20212 ;; @r{Concatenate blanks with tic symbol.}
20218 Next, we determine how many blanks are needed to indent the first tic
20219 mark to the first column of the graph. This uses the value of
20220 @code{full-Y-label-width} passed it by the @code{print-graph} function.
20223 The code to make @code{X-axis-leading-spaces}
20228 ;; X-axis-leading-spaces
20230 (make-string full-Y-label-width ? )
20235 We also need to determine the length of the horizontal axis, which is
20236 the length of the numbers list, and the number of ticks in the horizontal
20243 (length numbers-list)
20249 (* symbol-width X-axis-label-spacing)
20253 ;; number-of-X-ticks
20254 (if (zerop (% (X-length tic-width)))
20255 (/ (X-length tic-width))
20256 (1+ (/ (X-length tic-width))))
20261 All this leads us directly to the function for printing the X axis tic line:
20263 @findex print-X-axis-tic-line
20266 (defun print-X-axis-tic-line
20267 (number-of-X-tics X-axis-leading-spaces X-axis-tic-element)
20268 "Print ticks for X axis."
20269 (insert X-axis-leading-spaces)
20270 (insert X-axis-tic-symbol) ; @r{Under first column.}
20273 ;; @r{Insert second tic in the right spot.}
20276 (- (* symbol-width X-axis-label-spacing)
20277 ;; @r{Insert white space up to second tic symbol.}
20278 (* 2 (length X-axis-tic-symbol)))
20280 X-axis-tic-symbol))
20283 ;; @r{Insert remaining ticks.}
20284 (while (> number-of-X-tics 1)
20285 (insert X-axis-tic-element)
20286 (setq number-of-X-tics (1- number-of-X-tics))))
20290 The line of numbers is equally straightforward:
20293 First, we create a numbered element with blank spaces before each number:
20295 @findex X-axis-element
20298 (defun X-axis-element (number)
20299 "Construct a numbered X axis element."
20300 (let ((leading-spaces
20301 (- (* symbol-width X-axis-label-spacing)
20302 (length (number-to-string number)))))
20303 (concat (make-string leading-spaces ? )
20304 (number-to-string number))))
20308 Next, we create the function to print the numbered line, starting with
20309 the number ``1'' under the first column:
20311 @findex print-X-axis-numbered-line
20314 (defun print-X-axis-numbered-line
20315 (number-of-X-tics X-axis-leading-spaces)
20316 "Print line of X-axis numbers"
20317 (let ((number X-axis-label-spacing))
20318 (insert X-axis-leading-spaces)
20324 ;; @r{Insert white space up to next number.}
20325 (- (* symbol-width X-axis-label-spacing) 2)
20327 (number-to-string number)))
20330 ;; @r{Insert remaining numbers.}
20331 (setq number (+ number X-axis-label-spacing))
20332 (while (> number-of-X-tics 1)
20333 (insert (X-axis-element number))
20334 (setq number (+ number X-axis-label-spacing))
20335 (setq number-of-X-tics (1- number-of-X-tics)))))
20339 Finally, we need to write the @code{print-X-axis} that uses
20340 @code{print-X-axis-tic-line} and
20341 @code{print-X-axis-numbered-line}.
20343 The function must determine the local values of the variables used by both
20344 @code{print-X-axis-tic-line} and @code{print-X-axis-numbered-line}, and
20345 then it must call them. Also, it must print the carriage return that
20346 separates the two lines.
20348 The function consists of a varlist that specifies five local variables,
20349 and calls to each of the two line printing functions:
20351 @findex print-X-axis
20354 (defun print-X-axis (numbers-list)
20355 "Print X axis labels to length of NUMBERS-LIST."
20356 (let* ((leading-spaces
20357 (make-string full-Y-label-width ? ))
20360 ;; symbol-width @r{is provided by} graph-body-print
20361 (tic-width (* symbol-width X-axis-label-spacing))
20362 (X-length (length numbers-list))
20370 ;; @r{Make a string of blanks.}
20371 (- (* symbol-width X-axis-label-spacing)
20372 (length X-axis-tic-symbol))
20376 ;; @r{Concatenate blanks with tic symbol.}
20377 X-axis-tic-symbol))
20381 (if (zerop (% X-length tic-width))
20382 (/ X-length tic-width)
20383 (1+ (/ X-length tic-width)))))
20386 (print-X-axis-tic-line tic-number leading-spaces X-tic)
20388 (print-X-axis-numbered-line tic-number leading-spaces)))
20393 You can test @code{print-X-axis}:
20397 Install @code{X-axis-tic-symbol}, @code{X-axis-label-spacing},
20398 @code{print-X-axis-tic-line}, as well as @code{X-axis-element},
20399 @code{print-X-axis-numbered-line}, and @code{print-X-axis}.
20402 Copy the following expression:
20407 (let ((full-Y-label-width 5)
20410 '(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16))))
20415 Switch to the @file{*scratch*} buffer and place the cursor where you
20416 want the axis labels to start.
20419 Type @kbd{M-:} (@code{eval-expression}).
20422 Yank the test expression into the minibuffer
20423 with @kbd{C-y} (@code{yank)}.
20426 Press @key{RET} to evaluate the expression.
20430 Emacs will print the horizontal axis like this:
20440 @node Print Whole Graph
20441 @appendixsec Printing the Whole Graph
20442 @cindex Printing the whole graph
20443 @cindex Whole graph printing
20444 @cindex Graph, printing all
20446 Now we are nearly ready to print the whole graph.
20448 The function to print the graph with the proper labels follows the
20449 outline we created earlier (@pxref{Full Graph, , A Graph with Labeled
20450 Axes}), but with additions.
20453 Here is the outline:
20457 (defun print-graph (numbers-list)
20458 "@var{documentation}@dots{}"
20459 (let ((height @dots{}
20463 (print-Y-axis height @dots{} )
20464 (graph-body-print numbers-list)
20465 (print-X-axis @dots{} )))
20470 * The final version:: A few changes.
20471 * Test print-graph:: Run a short test.
20472 * Graphing words in defuns:: Executing the final code.
20473 * lambda:: How to write an anonymous function.
20474 * mapcar:: Apply a function to elements of a list.
20475 * Another Bug:: Yet another bug @dots{} most insidious.
20476 * Final printed graph:: The graph itself!
20480 @node The final version
20481 @unnumberedsubsec Changes for the Final Version
20484 The final version is different from what we planned in two ways:
20485 first, it contains additional values calculated once in the varlist;
20486 second, it carries an option to specify the labels' increment per row.
20487 This latter feature turns out to be essential; otherwise, a graph may
20488 have more rows than fit on a display or on a sheet of paper.
20491 This new feature requires a change to the @code{Y-axis-column}
20492 function, to add @code{vertical-step} to it. The function looks like
20495 @findex Y-axis-column @r{Final version.}
20498 ;;; @r{Final version.}
20499 (defun Y-axis-column
20500 (height width-of-label &optional vertical-step)
20501 "Construct list of labels for Y axis.
20502 HEIGHT is maximum height of graph.
20503 WIDTH-OF-LABEL is maximum width of label.
20504 VERTICAL-STEP, an option, is a positive integer
20505 that specifies how much a Y axis label increments
20506 for each line. For example, a step of 5 means
20507 that each line is five units of the graph."
20511 (number-per-line (or vertical-step 1)))
20512 (while (> height 1)
20513 (if (zerop (% height Y-axis-label-spacing))
20516 ;; @r{Insert label.}
20520 (* height number-per-line)
20525 ;; @r{Else, insert blanks.}
20528 (make-string width-of-label ? )
20530 (setq height (1- height)))
20533 ;; @r{Insert base line.}
20534 (setq Y-axis (cons (Y-axis-element
20535 (or vertical-step 1)
20538 (nreverse Y-axis)))
20542 The values for the maximum height of graph and the width of a symbol
20543 are computed by @code{print-graph} in its @code{let} expression; so
20544 @code{graph-body-print} must be changed to accept them.
20546 @findex graph-body-print @r{Final version.}
20549 ;;; @r{Final version.}
20550 (defun graph-body-print (numbers-list height symbol-width)
20551 "Print a bar graph of the NUMBERS-LIST.
20552 The numbers-list consists of the Y-axis values.
20553 HEIGHT is maximum height of graph.
20554 SYMBOL-WIDTH is number of each column."
20557 (let (from-position)
20558 (while numbers-list
20559 (setq from-position (point))
20561 (column-of-graph height (car numbers-list)))
20562 (goto-char from-position)
20563 (forward-char symbol-width)
20566 ;; @r{Draw graph column by column.}
20568 (setq numbers-list (cdr numbers-list)))
20569 ;; @r{Place point for X axis labels.}
20570 (forward-line height)
20576 Finally, the code for the @code{print-graph} function:
20578 @findex print-graph @r{Final version.}
20581 ;;; @r{Final version.}
20583 (numbers-list &optional vertical-step)
20584 "Print labeled bar graph of the NUMBERS-LIST.
20585 The numbers-list consists of the Y-axis values.
20589 Optionally, VERTICAL-STEP, a positive integer,
20590 specifies how much a Y axis label increments for
20591 each line. For example, a step of 5 means that
20592 each row is five units."
20595 (let* ((symbol-width (length graph-blank))
20596 ;; @code{height} @r{is both the largest number}
20597 ;; @r{and the number with the most digits.}
20598 (height (apply 'max numbers-list))
20601 (height-of-top-line
20602 (if (zerop (% height Y-axis-label-spacing))
20605 (* (1+ (/ height Y-axis-label-spacing))
20606 Y-axis-label-spacing)))
20609 (vertical-step (or vertical-step 1))
20610 (full-Y-label-width
20616 (* height-of-top-line vertical-step))
20622 height-of-top-line full-Y-label-width vertical-step)
20626 numbers-list height-of-top-line symbol-width)
20627 (print-X-axis numbers-list)))
20631 @node Test print-graph
20632 @appendixsubsec Testing @code{print-graph}
20635 We can test the @code{print-graph} function with a short list of numbers:
20639 Install the final versions of @code{Y-axis-column},
20640 @code{graph-body-print}, and @code{print-graph} (in addition to the
20644 Copy the following expression:
20647 (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1))
20651 Switch to the @file{*scratch*} buffer and place the cursor where you
20652 want the axis labels to start.
20655 Type @kbd{M-:} (@code{eval-expression}).
20658 Yank the test expression into the minibuffer
20659 with @kbd{C-y} (@code{yank)}.
20662 Press @key{RET} to evaluate the expression.
20666 Emacs will print a graph that looks like this:
20687 On the other hand, if you pass @code{print-graph} a
20688 @code{vertical-step} value of 2, by evaluating this expression:
20691 (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1) 2)
20696 The graph looks like this:
20717 (A question: is the `2' on the bottom of the vertical axis a bug or a
20718 feature? If you think it is a bug, and should be a `1' instead, (or
20719 even a `0'), you can modify the sources.)
20721 @node Graphing words in defuns
20722 @appendixsubsec Graphing Numbers of Words and Symbols
20724 Now for the graph for which all this code was written: a graph that
20725 shows how many function definitions contain fewer than 10 words and
20726 symbols, how many contain between 10 and 19 words and symbols, how
20727 many contain between 20 and 29 words and symbols, and so on.
20729 This is a multi-step process. First make sure you have loaded all the
20733 It is a good idea to reset the value of @code{top-of-ranges} in case
20734 you have set it to some different value. You can evaluate the
20739 (setq top-of-ranges
20742 110 120 130 140 150
20743 160 170 180 190 200
20744 210 220 230 240 250
20745 260 270 280 290 300)
20750 Next create a list of the number of words and symbols in each range.
20754 Evaluate the following:
20758 (setq list-for-graph
20761 (recursive-lengths-list-many-files
20762 (directory-files "/usr/local/emacs/lisp"
20770 On my old machine, this took about an hour. It looked though 303 Lisp
20771 files in my copy of Emacs version 19.23. After all that computing,
20772 the @code{list-for-graph} had this value:
20776 (537 1027 955 785 594 483 349 292 224 199 166 120 116 99
20777 90 80 67 48 52 45 41 33 28 26 25 20 12 28 11 13 220)
20782 This means that my copy of Emacs had 537 function definitions with
20783 fewer than 10 words or symbols in them, 1,027 function definitions
20784 with 10 to 19 words or symbols in them, 955 function definitions with
20785 20 to 29 words or symbols in them, and so on.
20787 Clearly, just by looking at this list we can see that most function
20788 definitions contain ten to thirty words and symbols.
20790 Now for printing. We do @emph{not} want to print a graph that is
20791 1,030 lines high @dots{} Instead, we should print a graph that is
20792 fewer than twenty-five lines high. A graph that height can be
20793 displayed on almost any monitor, and easily printed on a sheet of paper.
20795 This means that each value in @code{list-for-graph} must be reduced to
20796 one-fiftieth its present value.
20798 Here is a short function to do just that, using two functions we have
20799 not yet seen, @code{mapcar} and @code{lambda}.
20803 (defun one-fiftieth (full-range)
20804 "Return list, each number one-fiftieth of previous."
20805 (mapcar (lambda (arg) (/ arg 50)) full-range))
20810 @appendixsubsec A @code{lambda} Expression: Useful Anonymity
20811 @cindex Anonymous function
20814 @code{lambda} is the symbol for an anonymous function, a function
20815 without a name. Every time you use an anonymous function, you need to
20816 include its whole body.
20823 (lambda (arg) (/ arg 50))
20827 is a function definition that says `return the value resulting from
20828 dividing whatever is passed to me as @code{arg} by 50'.
20831 Earlier, for example, we had a function @code{multiply-by-seven}; it
20832 multiplied its argument by 7. This function is similar, except it
20833 divides its argument by 50; and, it has no name. The anonymous
20834 equivalent of @code{multiply-by-seven} is:
20837 (lambda (number) (* 7 number))
20841 (@xref{defun, , The @code{defun} Macro}.)
20845 If we want to multiply 3 by 7, we can write:
20847 @c clear print-postscript-figures
20848 @c lambda example diagram #1
20852 (multiply-by-seven 3)
20853 \_______________/ ^
20859 @ifset print-postscript-figures
20862 @center @image{lambda-1}
20866 @ifclear print-postscript-figures
20870 (multiply-by-seven 3)
20871 \_______________/ ^
20880 This expression returns 21.
20884 Similarly, we can write:
20886 @c lambda example diagram #2
20890 ((lambda (number) (* 7 number)) 3)
20891 \____________________________/ ^
20893 anonymous function argument
20897 @ifset print-postscript-figures
20900 @center @image{lambda-2}
20904 @ifclear print-postscript-figures
20908 ((lambda (number) (* 7 number)) 3)
20909 \____________________________/ ^
20911 anonymous function argument
20919 If we want to divide 100 by 50, we can write:
20921 @c lambda example diagram #3
20925 ((lambda (arg) (/ arg 50)) 100)
20926 \______________________/ \_/
20928 anonymous function argument
20932 @ifset print-postscript-figures
20935 @center @image{lambda-3}
20939 @ifclear print-postscript-figures
20943 ((lambda (arg) (/ arg 50)) 100)
20944 \______________________/ \_/
20946 anonymous function argument
20953 This expression returns 2. The 100 is passed to the function, which
20954 divides that number by 50.
20956 @xref{Lambda Expressions, , Lambda Expressions, elisp, The GNU Emacs
20957 Lisp Reference Manual}, for more about @code{lambda}. Lisp and lambda
20958 expressions derive from the Lambda Calculus.
20961 @appendixsubsec The @code{mapcar} Function
20964 @code{mapcar} is a function that calls its first argument with each
20965 element of its second argument, in turn. The second argument must be
20968 The @samp{map} part of the name comes from the mathematical phrase,
20969 `mapping over a domain', meaning to apply a function to each of the
20970 elements in a domain. The mathematical phrase is based on the
20971 metaphor of a surveyor walking, one step at a time, over an area he is
20972 mapping. And @samp{car}, of course, comes from the Lisp notion of the
20981 (mapcar '1+ '(2 4 6))
20987 The function @code{1+} which adds one to its argument, is executed on
20988 @emph{each} element of the list, and a new list is returned.
20990 Contrast this with @code{apply}, which applies its first argument to
20992 (@xref{Readying a Graph, , Readying a Graph}, for a explanation of
20996 In the definition of @code{one-fiftieth}, the first argument is the
20997 anonymous function:
21000 (lambda (arg) (/ arg 50))
21004 and the second argument is @code{full-range}, which will be bound to
21005 @code{list-for-graph}.
21008 The whole expression looks like this:
21011 (mapcar (lambda (arg) (/ arg 50)) full-range))
21014 @xref{Mapping Functions, , Mapping Functions, elisp, The GNU Emacs
21015 Lisp Reference Manual}, for more about @code{mapcar}.
21017 Using the @code{one-fiftieth} function, we can generate a list in
21018 which each element is one-fiftieth the size of the corresponding
21019 element in @code{list-for-graph}.
21023 (setq fiftieth-list-for-graph
21024 (one-fiftieth list-for-graph))
21029 The resulting list looks like this:
21033 (10 20 19 15 11 9 6 5 4 3 3 2 2
21034 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 4)
21039 This, we are almost ready to print! (We also notice the loss of
21040 information: many of the higher ranges are 0, meaning that fewer than
21041 50 defuns had that many words or symbols---but not necessarily meaning
21042 that none had that many words or symbols.)
21045 @appendixsubsec Another Bug @dots{} Most Insidious
21046 @cindex Bug, most insidious type
21047 @cindex Insidious type of bug
21049 I said `almost ready to print'! Of course, there is a bug in the
21050 @code{print-graph} function @dots{} It has a @code{vertical-step}
21051 option, but not a @code{horizontal-step} option. The
21052 @code{top-of-range} scale goes from 10 to 300 by tens. But the
21053 @code{print-graph} function will print only by ones.
21055 This is a classic example of what some consider the most insidious
21056 type of bug, the bug of omission. This is not the kind of bug you can
21057 find by studying the code, for it is not in the code; it is an omitted
21058 feature. Your best actions are to try your program early and often;
21059 and try to arrange, as much as you can, to write code that is easy to
21060 understand and easy to change. Try to be aware, whenever you can,
21061 that whatever you have written, @emph{will} be rewritten, if not soon,
21062 eventually. A hard maxim to follow.
21064 It is the @code{print-X-axis-numbered-line} function that needs the
21065 work; and then the @code{print-X-axis} and the @code{print-graph}
21066 functions need to be adapted. Not much needs to be done; there is one
21067 nicety: the numbers ought to line up under the tic marks. This takes
21071 Here is the corrected @code{print-X-axis-numbered-line}:
21075 (defun print-X-axis-numbered-line
21076 (number-of-X-tics X-axis-leading-spaces
21077 &optional horizontal-step)
21078 "Print line of X-axis numbers"
21079 (let ((number X-axis-label-spacing)
21080 (horizontal-step (or horizontal-step 1)))
21083 (insert X-axis-leading-spaces)
21084 ;; @r{Delete extra leading spaces.}
21087 (length (number-to-string horizontal-step)))))
21092 ;; @r{Insert white space.}
21094 X-axis-label-spacing)
21097 (number-to-string horizontal-step)))
21101 (* number horizontal-step))))
21104 ;; @r{Insert remaining numbers.}
21105 (setq number (+ number X-axis-label-spacing))
21106 (while (> number-of-X-tics 1)
21107 (insert (X-axis-element
21108 (* number horizontal-step)))
21109 (setq number (+ number X-axis-label-spacing))
21110 (setq number-of-X-tics (1- number-of-X-tics)))))
21115 If you are reading this in Info, you can see the new versions of
21116 @code{print-X-axis} @code{print-graph} and evaluate them. If you are
21117 reading this in a printed book, you can see the changed lines here
21118 (the full text is too much to print).
21123 (defun print-X-axis (numbers-list horizontal-step)
21125 (print-X-axis-numbered-line
21126 tic-number leading-spaces horizontal-step))
21134 &optional vertical-step horizontal-step)
21136 (print-X-axis numbers-list horizontal-step))
21144 (defun print-X-axis (numbers-list horizontal-step)
21145 "Print X axis labels to length of NUMBERS-LIST.
21146 Optionally, HORIZONTAL-STEP, a positive integer,
21147 specifies how much an X axis label increments for
21151 ;; Value of symbol-width and full-Y-label-width
21152 ;; are passed by `print-graph'.
21153 (let* ((leading-spaces
21154 (make-string full-Y-label-width ? ))
21155 ;; symbol-width @r{is provided by} graph-body-print
21156 (tic-width (* symbol-width X-axis-label-spacing))
21157 (X-length (length numbers-list))
21163 ;; @r{Make a string of blanks.}
21164 (- (* symbol-width X-axis-label-spacing)
21165 (length X-axis-tic-symbol))
21169 ;; @r{Concatenate blanks with tic symbol.}
21170 X-axis-tic-symbol))
21172 (if (zerop (% X-length tic-width))
21173 (/ X-length tic-width)
21174 (1+ (/ X-length tic-width)))))
21178 (print-X-axis-tic-line
21179 tic-number leading-spaces X-tic)
21181 (print-X-axis-numbered-line
21182 tic-number leading-spaces horizontal-step)))
21189 (numbers-list &optional vertical-step horizontal-step)
21190 "Print labeled bar graph of the NUMBERS-LIST.
21191 The numbers-list consists of the Y-axis values.
21195 Optionally, VERTICAL-STEP, a positive integer,
21196 specifies how much a Y axis label increments for
21197 each line. For example, a step of 5 means that
21198 each row is five units.
21202 Optionally, HORIZONTAL-STEP, a positive integer,
21203 specifies how much an X axis label increments for
21205 (let* ((symbol-width (length graph-blank))
21206 ;; @code{height} @r{is both the largest number}
21207 ;; @r{and the number with the most digits.}
21208 (height (apply 'max numbers-list))
21211 (height-of-top-line
21212 (if (zerop (% height Y-axis-label-spacing))
21215 (* (1+ (/ height Y-axis-label-spacing))
21216 Y-axis-label-spacing)))
21219 (vertical-step (or vertical-step 1))
21220 (full-Y-label-width
21224 (* height-of-top-line vertical-step))
21229 height-of-top-line full-Y-label-width vertical-step)
21231 numbers-list height-of-top-line symbol-width)
21232 (print-X-axis numbers-list horizontal-step)))
21239 Graphing Definitions Re-listed
21242 Here are all the graphing definitions in their final form:
21246 (defvar top-of-ranges
21249 110 120 130 140 150
21250 160 170 180 190 200
21251 210 220 230 240 250)
21252 "List specifying ranges for `defuns-per-range'.")
21256 (defvar graph-symbol "*"
21257 "String used as symbol in graph, usually an asterisk.")
21261 (defvar graph-blank " "
21262 "String used as blank in graph, usually a blank space.
21263 graph-blank must be the same number of columns wide
21268 (defvar Y-axis-tic " - "
21269 "String that follows number in a Y axis label.")
21273 (defvar Y-axis-label-spacing 5
21274 "Number of lines from one Y axis label to next.")
21278 (defvar X-axis-tic-symbol "|"
21279 "String to insert to point to a column in X axis.")
21283 (defvar X-axis-label-spacing
21284 (if (boundp 'graph-blank)
21285 (* 5 (length graph-blank)) 5)
21286 "Number of units from one X axis label to next.")
21292 (defun count-words-in-defun ()
21293 "Return the number of words and symbols in a defun."
21294 (beginning-of-defun)
21296 (end (save-excursion (end-of-defun) (point))))
21301 (and (< (point) end)
21303 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
21305 (setq count (1+ count)))
21312 (defun lengths-list-file (filename)
21313 "Return list of definitions' lengths within FILE.
21314 The returned list is a list of numbers.
21315 Each number is the number of words or
21316 symbols in one function definition."
21320 (message "Working on `%s' ... " filename)
21322 (let ((buffer (find-file-noselect filename))
21324 (set-buffer buffer)
21325 (setq buffer-read-only t)
21327 (goto-char (point-min))
21331 (while (re-search-forward "^(defun" nil t)
21333 (cons (count-words-in-defun) lengths-list)))
21334 (kill-buffer buffer)
21341 (defun lengths-list-many-files (list-of-files)
21342 "Return list of lengths of defuns in LIST-OF-FILES."
21343 (let (lengths-list)
21344 ;;; @r{true-or-false-test}
21345 (while list-of-files
21351 ;;; @r{Generate a lengths' list.}
21353 (expand-file-name (car list-of-files)))))
21354 ;;; @r{Make files' list shorter.}
21355 (setq list-of-files (cdr list-of-files)))
21356 ;;; @r{Return final value of lengths' list.}
21363 (defun defuns-per-range (sorted-lengths top-of-ranges)
21364 "SORTED-LENGTHS defuns in each TOP-OF-RANGES range."
21365 (let ((top-of-range (car top-of-ranges))
21366 (number-within-range 0)
21367 defuns-per-range-list)
21372 (while top-of-ranges
21376 ;; @r{Need number for numeric test.}
21377 (car sorted-lengths)
21378 (< (car sorted-lengths) top-of-range))
21380 ;; @r{Count number of definitions within current range.}
21381 (setq number-within-range (1+ number-within-range))
21382 (setq sorted-lengths (cdr sorted-lengths)))
21386 ;; @r{Exit inner loop but remain within outer loop.}
21388 (setq defuns-per-range-list
21389 (cons number-within-range defuns-per-range-list))
21390 (setq number-within-range 0) ; @r{Reset count to zero.}
21392 ;; @r{Move to next range.}
21393 (setq top-of-ranges (cdr top-of-ranges))
21394 ;; @r{Specify next top of range value.}
21395 (setq top-of-range (car top-of-ranges)))
21399 ;; @r{Exit outer loop and count the number of defuns larger than}
21400 ;; @r{ the largest top-of-range value.}
21401 (setq defuns-per-range-list
21403 (length sorted-lengths)
21404 defuns-per-range-list))
21406 ;; @r{Return a list of the number of definitions within each range,}
21407 ;; @r{ smallest to largest.}
21408 (nreverse defuns-per-range-list)))
21414 (defun column-of-graph (max-graph-height actual-height)
21415 "Return list of MAX-GRAPH-HEIGHT strings;
21416 ACTUAL-HEIGHT are graph-symbols.
21417 The graph-symbols are contiguous entries at the end
21419 The list will be inserted as one column of a graph.
21420 The strings are either graph-blank or graph-symbol."
21424 (let ((insert-list nil)
21425 (number-of-top-blanks
21426 (- max-graph-height actual-height)))
21428 ;; @r{Fill in @code{graph-symbols}.}
21429 (while (> actual-height 0)
21430 (setq insert-list (cons graph-symbol insert-list))
21431 (setq actual-height (1- actual-height)))
21435 ;; @r{Fill in @code{graph-blanks}.}
21436 (while (> number-of-top-blanks 0)
21437 (setq insert-list (cons graph-blank insert-list))
21438 (setq number-of-top-blanks
21439 (1- number-of-top-blanks)))
21441 ;; @r{Return whole list.}
21448 (defun Y-axis-element (number full-Y-label-width)
21449 "Construct a NUMBERed label element.
21450 A numbered element looks like this ` 5 - ',
21451 and is padded as needed so all line up with
21452 the element for the largest number."
21455 (let* ((leading-spaces
21456 (- full-Y-label-width
21458 (concat (number-to-string number)
21463 (make-string leading-spaces ? )
21464 (number-to-string number)
21471 (defun print-Y-axis
21472 (height full-Y-label-width &optional vertical-step)
21473 "Insert Y axis by HEIGHT and FULL-Y-LABEL-WIDTH.
21474 Height must be the maximum height of the graph.
21475 Full width is the width of the highest label element.
21476 Optionally, print according to VERTICAL-STEP."
21479 ;; Value of height and full-Y-label-width
21480 ;; are passed by `print-graph'.
21481 (let ((start (point)))
21483 (Y-axis-column height full-Y-label-width vertical-step))
21486 ;; @r{Place point ready for inserting graph.}
21488 ;; @r{Move point forward by value of} full-Y-label-width
21489 (forward-char full-Y-label-width)))
21495 (defun print-X-axis-tic-line
21496 (number-of-X-tics X-axis-leading-spaces X-axis-tic-element)
21497 "Print ticks for X axis."
21498 (insert X-axis-leading-spaces)
21499 (insert X-axis-tic-symbol) ; @r{Under first column.}
21502 ;; @r{Insert second tic in the right spot.}
21505 (- (* symbol-width X-axis-label-spacing)
21506 ;; @r{Insert white space up to second tic symbol.}
21507 (* 2 (length X-axis-tic-symbol)))
21509 X-axis-tic-symbol))
21512 ;; @r{Insert remaining ticks.}
21513 (while (> number-of-X-tics 1)
21514 (insert X-axis-tic-element)
21515 (setq number-of-X-tics (1- number-of-X-tics))))
21521 (defun X-axis-element (number)
21522 "Construct a numbered X axis element."
21523 (let ((leading-spaces
21524 (- (* symbol-width X-axis-label-spacing)
21525 (length (number-to-string number)))))
21526 (concat (make-string leading-spaces ? )
21527 (number-to-string number))))
21533 (defun graph-body-print (numbers-list height symbol-width)
21534 "Print a bar graph of the NUMBERS-LIST.
21535 The numbers-list consists of the Y-axis values.
21536 HEIGHT is maximum height of graph.
21537 SYMBOL-WIDTH is number of each column."
21540 (let (from-position)
21541 (while numbers-list
21542 (setq from-position (point))
21544 (column-of-graph height (car numbers-list)))
21545 (goto-char from-position)
21546 (forward-char symbol-width)
21549 ;; @r{Draw graph column by column.}
21551 (setq numbers-list (cdr numbers-list)))
21552 ;; @r{Place point for X axis labels.}
21553 (forward-line height)
21560 (defun Y-axis-column
21561 (height width-of-label &optional vertical-step)
21562 "Construct list of labels for Y axis.
21563 HEIGHT is maximum height of graph.
21564 WIDTH-OF-LABEL is maximum width of label.
21567 VERTICAL-STEP, an option, is a positive integer
21568 that specifies how much a Y axis label increments
21569 for each line. For example, a step of 5 means
21570 that each line is five units of the graph."
21572 (number-per-line (or vertical-step 1)))
21575 (while (> height 1)
21576 (if (zerop (% height Y-axis-label-spacing))
21577 ;; @r{Insert label.}
21581 (* height number-per-line)
21586 ;; @r{Else, insert blanks.}
21589 (make-string width-of-label ? )
21591 (setq height (1- height)))
21594 ;; @r{Insert base line.}
21595 (setq Y-axis (cons (Y-axis-element
21596 (or vertical-step 1)
21599 (nreverse Y-axis)))
21605 (defun print-X-axis-numbered-line
21606 (number-of-X-tics X-axis-leading-spaces
21607 &optional horizontal-step)
21608 "Print line of X-axis numbers"
21609 (let ((number X-axis-label-spacing)
21610 (horizontal-step (or horizontal-step 1)))
21613 (insert X-axis-leading-spaces)
21615 (delete-char (- (1- (length (number-to-string horizontal-step)))))
21618 ;; @r{Insert white space up to next number.}
21619 (- (* symbol-width X-axis-label-spacing)
21620 (1- (length (number-to-string horizontal-step)))
21623 (number-to-string (* number horizontal-step))))
21626 ;; @r{Insert remaining numbers.}
21627 (setq number (+ number X-axis-label-spacing))
21628 (while (> number-of-X-tics 1)
21629 (insert (X-axis-element (* number horizontal-step)))
21630 (setq number (+ number X-axis-label-spacing))
21631 (setq number-of-X-tics (1- number-of-X-tics)))))
21637 (defun print-X-axis (numbers-list horizontal-step)
21638 "Print X axis labels to length of NUMBERS-LIST.
21639 Optionally, HORIZONTAL-STEP, a positive integer,
21640 specifies how much an X axis label increments for
21644 ;; Value of symbol-width and full-Y-label-width
21645 ;; are passed by `print-graph'.
21646 (let* ((leading-spaces
21647 (make-string full-Y-label-width ? ))
21648 ;; symbol-width @r{is provided by} graph-body-print
21649 (tic-width (* symbol-width X-axis-label-spacing))
21650 (X-length (length numbers-list))
21656 ;; @r{Make a string of blanks.}
21657 (- (* symbol-width X-axis-label-spacing)
21658 (length X-axis-tic-symbol))
21662 ;; @r{Concatenate blanks with tic symbol.}
21663 X-axis-tic-symbol))
21665 (if (zerop (% X-length tic-width))
21666 (/ X-length tic-width)
21667 (1+ (/ X-length tic-width)))))
21671 (print-X-axis-tic-line
21672 tic-number leading-spaces X-tic)
21674 (print-X-axis-numbered-line
21675 tic-number leading-spaces horizontal-step)))
21681 (defun one-fiftieth (full-range)
21682 "Return list, each number of which is 1/50th previous."
21683 (mapcar (lambda (arg) (/ arg 50)) full-range))
21690 (numbers-list &optional vertical-step horizontal-step)
21691 "Print labeled bar graph of the NUMBERS-LIST.
21692 The numbers-list consists of the Y-axis values.
21696 Optionally, VERTICAL-STEP, a positive integer,
21697 specifies how much a Y axis label increments for
21698 each line. For example, a step of 5 means that
21699 each row is five units.
21703 Optionally, HORIZONTAL-STEP, a positive integer,
21704 specifies how much an X axis label increments for
21706 (let* ((symbol-width (length graph-blank))
21707 ;; @code{height} @r{is both the largest number}
21708 ;; @r{and the number with the most digits.}
21709 (height (apply 'max numbers-list))
21712 (height-of-top-line
21713 (if (zerop (% height Y-axis-label-spacing))
21716 (* (1+ (/ height Y-axis-label-spacing))
21717 Y-axis-label-spacing)))
21720 (vertical-step (or vertical-step 1))
21721 (full-Y-label-width
21725 (* height-of-top-line vertical-step))
21731 height-of-top-line full-Y-label-width vertical-step)
21733 numbers-list height-of-top-line symbol-width)
21734 (print-X-axis numbers-list horizontal-step)))
21741 @node Final printed graph
21742 @appendixsubsec The Printed Graph
21744 When made and installed, you can call the @code{print-graph} command
21750 (print-graph fiftieth-list-for-graph 50 10)
21780 50 - ***************** * *
21782 10 50 100 150 200 250 300 350
21789 The largest group of functions contain 10--19 words and symbols each.
21791 @node Free Software and Free Manuals
21792 @appendix Free Software and Free Manuals
21794 @strong{by Richard M. Stallman}
21797 The biggest deficiency in free operating systems is not in the
21798 software---it is the lack of good free manuals that we can include in
21799 these systems. Many of our most important programs do not come with
21800 full manuals. Documentation is an essential part of any software
21801 package; when an important free software package does not come with a
21802 free manual, that is a major gap. We have many such gaps today.
21804 Once upon a time, many years ago, I thought I would learn Perl. I got
21805 a copy of a free manual, but I found it hard to read. When I asked
21806 Perl users about alternatives, they told me that there were better
21807 introductory manuals---but those were not free.
21809 Why was this? The authors of the good manuals had written them for
21810 O'Reilly Associates, which published them with restrictive terms---no
21811 copying, no modification, source files not available---which exclude
21812 them from the free software community.
21814 That wasn't the first time this sort of thing has happened, and (to
21815 our community's great loss) it was far from the last. Proprietary
21816 manual publishers have enticed a great many authors to restrict their
21817 manuals since then. Many times I have heard a GNU user eagerly tell me
21818 about a manual that he is writing, with which he expects to help the
21819 GNU project---and then had my hopes dashed, as he proceeded to explain
21820 that he had signed a contract with a publisher that would restrict it
21821 so that we cannot use it.
21823 Given that writing good English is a rare skill among programmers, we
21824 can ill afford to lose manuals this way.
21826 Free documentation, like free software, is a matter of freedom, not
21827 price. The problem with these manuals was not that O'Reilly Associates
21828 charged a price for printed copies---that in itself is fine. The Free
21829 Software Foundation @uref{http://shop.fsf.org, sells printed copies} of
21830 free @uref{http://www.gnu.org/doc/doc.html, GNU manuals}, too.
21831 But GNU manuals are available in source code form, while these manuals
21832 are available only on paper. GNU manuals come with permission to copy
21833 and modify; the Perl manuals do not. These restrictions are the
21836 The criterion for a free manual is pretty much the same as for free
21837 software: it is a matter of giving all users certain
21838 freedoms. Redistribution (including commercial redistribution) must be
21839 permitted, so that the manual can accompany every copy of the program,
21840 on-line or on paper. Permission for modification is crucial too.
21842 As a general rule, I don't believe that it is essential for people to
21843 have permission to modify all sorts of articles and books. The issues
21844 for writings are not necessarily the same as those for software. For
21845 example, I don't think you or I are obliged to give permission to
21846 modify articles like this one, which describe our actions and our
21849 But there is a particular reason why the freedom to modify is crucial
21850 for documentation for free software. When people exercise their right
21851 to modify the software, and add or change its features, if they are
21852 conscientious they will change the manual too---so they can provide
21853 accurate and usable documentation with the modified program. A manual
21854 which forbids programmers to be conscientious and finish the job, or
21855 more precisely requires them to write a new manual from scratch if
21856 they change the program, does not fill our community's needs.
21858 While a blanket prohibition on modification is unacceptable, some
21859 kinds of limits on the method of modification pose no problem. For
21860 example, requirements to preserve the original author's copyright
21861 notice, the distribution terms, or the list of authors, are ok. It is
21862 also no problem to require modified versions to include notice that
21863 they were modified, even to have entire sections that may not be
21864 deleted or changed, as long as these sections deal with nontechnical
21865 topics. (Some GNU manuals have them.)
21867 These kinds of restrictions are not a problem because, as a practical
21868 matter, they don't stop the conscientious programmer from adapting the
21869 manual to fit the modified program. In other words, they don't block
21870 the free software community from making full use of the manual.
21872 However, it must be possible to modify all the technical content of
21873 the manual, and then distribute the result in all the usual media,
21874 through all the usual channels; otherwise, the restrictions do block
21875 the community, the manual is not free, and so we need another manual.
21877 Unfortunately, it is often hard to find someone to write another
21878 manual when a proprietary manual exists. The obstacle is that many
21879 users think that a proprietary manual is good enough---so they don't
21880 see the need to write a free manual. They do not see that the free
21881 operating system has a gap that needs filling.
21883 Why do users think that proprietary manuals are good enough? Some have
21884 not considered the issue. I hope this article will do something to
21887 Other users consider proprietary manuals acceptable for the same
21888 reason so many people consider proprietary software acceptable: they
21889 judge in purely practical terms, not using freedom as a
21890 criterion. These people are entitled to their opinions, but since
21891 those opinions spring from values which do not include freedom, they
21892 are no guide for those of us who do value freedom.
21894 Please spread the word about this issue. We continue to lose manuals
21895 to proprietary publishing. If we spread the word that proprietary
21896 manuals are not sufficient, perhaps the next person who wants to help
21897 GNU by writing documentation will realize, before it is too late, that
21898 he must above all make it free.
21900 We can also encourage commercial publishers to sell free, copylefted
21901 manuals instead of proprietary ones. One way you can help this is to
21902 check the distribution terms of a manual before you buy it, and prefer
21903 copylefted manuals to non-copylefted ones.
21907 Note: The Free Software Foundation maintains a page on its Web site
21908 that lists free books available from other publishers:@*
21909 @uref{http://www.gnu.org/doc/other-free-books.html}
21911 @node GNU Free Documentation License
21912 @appendix GNU Free Documentation License
21914 @cindex FDL, GNU Free Documentation License
21915 @include doclicense.texi
21921 MENU ENTRY: NODE NAME.
21927 @c Place biographical information on right-hand (verso) page
21930 \par\vfill\supereject
21932 \global\evenheadline={\hfil} \global\evenfootline={\hfil}
21933 \global\oddheadline={\hfil} \global\oddfootline={\hfil}
21936 % \par\vfill\supereject
21937 \global\evenheadline={\hfil} \global\evenfootline={\hfil}
21938 \global\oddheadline={\hfil} \global\oddfootline={\hfil}
21939 %\page\hbox{}%\page
21940 %\page\hbox{}%\page
21947 @c ================ Biographical information ================
21951 @center About the Author
21956 @node About the Author
21957 @unnumbered About the Author
21961 Robert J. Chassell has worked with GNU Emacs since 1985. He writes
21962 and edits, teaches Emacs and Emacs Lisp, and speaks throughout the
21963 world on software freedom. Chassell was a founding Director and
21964 Treasurer of the Free Software Foundation, Inc. He is co-author of
21965 the @cite{Texinfo} manual, and has edited more than a dozen other
21966 books. He graduated from Cambridge University, in England. He has an
21967 abiding interest in social and economic history and flies his own
21974 @c @c Prevent page number on blank verso, so eject it first.
21976 @c \par\vfill\supereject
21981 @c @evenheading @thispage @| @| @thistitle
21982 @c @oddheading @| @| @thispage