1 \input texinfo @c -*-texinfo-*-
2 @comment %**start of header
3 @setfilename ../../info/eintr.info
4 @c setfilename emacs-lisp-intro.info
5 @c sethtmlfilename emacs-lisp-intro.html
6 @settitle Programming in Emacs Lisp
12 @include emacsver.texi
14 @c ================ How to Print a Book in Various Sizes ================
16 @c This book can be printed in any of three different sizes.
17 @c Set the following @-commands appropriately.
27 @c European A4 size paper:
32 @c (Note: if you edit the book so as to change the length of the
33 @c table of contents, you may have to change the value of 'pageno' below.)
35 @c <<<< For hard copy printing, this file is now
36 @c set for smallbook, which works for all sizes
37 @c of paper, and with PostScript figures >>>>
45 @c ================ Included Figures ================
47 @c If you clear this, the figures will be printed as ASCII diagrams
48 @c rather than PostScript/PDF.
49 @c (This is not relevant to Info, since Info only handles ASCII.)
50 @set print-postscript-figures
51 @c clear print-postscript-figures
53 @comment %**end of header
55 @c per rms and peterb, use 10pt fonts for the main text, mostly to
56 @c save on paper cost.
57 @c Do this inside @tex for now, so current makeinfo does not complain.
63 \global\hbadness=6666 % don't worry about not-too-underfull boxes
66 @c These refer to the printed book sold by the FSF.
67 @set edition-number 3.10
68 @set update-date 28 October 2009
70 @c For next or subsequent edition:
71 @c create function using with-output-to-temp-buffer
72 @c create a major mode, with keymaps
73 @c run an asynchronous process, like grep or diff
75 @c For 8.5 by 11 inch format: do not use such a small amount of
76 @c whitespace between paragraphs as smallbook format
79 \global\parskip 6pt plus 1pt
83 @c For all sized formats: print within-book cross
84 @c reference with ``...'' rather than [...]
86 @c This works with the texinfo.tex file, version 2003-05-04.08,
87 @c in the Texinfo version 4.6 of the 2003 Jun 13 distribution.
90 \if \xrefprintnodename
91 \global\def\xrefprintnodename#1{\unskip, ``#1''}
93 \global\def\xrefprintnodename#1{ ``#1''}
95 % \global\def\xrefprintnodename#1{, ``#1''}
98 @c ----------------------------------------------------
100 @dircategory Emacs lisp
102 * Emacs Lisp Intro: (eintr). 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}.
116 Copyright @copyright{} 1990--1995, 1997, 2001--2015 Free Software
123 GNU Press, @hfill @uref{http://www.fsf.org/licensing/gnu-press/}@*
124 a division of the @hfill email: @email{sales@@fsf.org}@*
125 Free Software Foundation, Inc. @hfill Tel: +1 (617) 542-5942@*
126 51 Franklin Street, Fifth Floor @hfill Fax: +1 (617) 542-2652@*
127 Boston, MA 02110-1301 USA
131 Printed copies available from @uref{http://shop.fsf.org/}. Published by:
134 GNU Press, http://www.fsf.org/licensing/gnu-press/
135 a division of the email: sales@@fsf.org
136 Free Software Foundation, Inc. Tel: +1 (617) 542-5942
137 51 Franklin Street, Fifth Floor Fax: +1 (617) 542-2652
138 Boston, MA 02110-1301 USA
146 Permission is granted to copy, distribute and/or modify this document
147 under the terms of the GNU Free Documentation License, Version 1.3 or
148 any later version published by the Free Software Foundation; there
149 being no Invariant Section, with the Front-Cover Texts being ``A GNU
150 Manual'', and with the Back-Cover Texts as in (a) below. A copy of
151 the license is included in the section entitled ``GNU Free
152 Documentation License''.
154 (a) The FSF's Back-Cover Text is: ``You have the freedom to
155 copy and modify this GNU manual. Buying copies from the FSF
156 supports it in developing GNU and promoting software freedom.''
160 @c half title; two lines here, so do not use 'shorttitlepage'
163 \hbox{}\vskip 1.5in \chaprm \centerline{An Introduction to}%
165 {\begingroup\hbox{}\vskip 0.25in \chaprm%
166 \centerline{Programming in Emacs Lisp}%
167 \endgroup\page\hbox{}\page}
172 @center @titlefont{An Introduction to}
174 @center @titlefont{Programming in Emacs Lisp}
176 @center Revised Third Edition
178 @center by Robert J. Chassell
181 @vskip 0pt plus 1filll
187 @evenheading @thispage @| @| @thischapter
188 @oddheading @thissection @| @| @thispage
192 @c Keep T.O.C. short by tightening up for largebook
195 \global\parskip 2pt plus 1pt
196 \global\advance\baselineskip by -1pt
206 @top An Introduction to Programming in Emacs Lisp
210 <p>The homepage for GNU Emacs is at
211 <a href="/software/emacs/">http://www.gnu.org/software/emacs/</a>.<br>
212 To view this manual in other formats, click
213 <a href="/software/emacs/manual/eintr.html">here</a>.
219 This master menu first lists each chapter and index; then it lists
220 every node in every chapter.
223 @c >>>> Set pageno appropriately <<<<
225 @c The first page of the Preface is a roman numeral; it is the first
226 @c right handed page after the Table of Contents; hence the following
227 @c setting must be for an odd negative number.
230 @c global@pageno = -11
233 @set COUNT-WORDS count-words-example
234 @c Length of variable name chosen so that things still line up when expanded.
237 * Preface:: What to look for.
238 * List Processing:: What is Lisp?
239 * Practicing Evaluation:: Running several programs.
240 * Writing Defuns:: How to write function definitions.
241 * Buffer Walk Through:: Exploring a few buffer-related functions.
242 * More Complex:: A few, even more complex functions.
243 * Narrowing & Widening:: Restricting your and Emacs attention to
245 * car cdr & cons:: Fundamental functions in Lisp.
246 * Cutting & Storing Text:: Removing text and saving it.
247 * List Implementation:: How lists are implemented in the computer.
248 * Yanking:: Pasting stored text.
249 * Loops & Recursion:: How to repeat a process.
250 * Regexp Search:: Regular expression searches.
251 * Counting Words:: A review of repetition and regexps.
252 * Words in a defun:: Counting words in a @code{defun}.
253 * Readying a Graph:: A prototype graph printing function.
254 * Emacs Initialization:: How to write a @file{.emacs} file.
255 * Debugging:: How to run the Emacs Lisp debuggers.
256 * Conclusion:: Now you have the basics.
257 * the-the:: An appendix: how to find reduplicated words.
258 * Kill Ring:: An appendix: how the kill ring works.
259 * Full Graph:: How to create a graph with labeled axes.
260 * Free Software and Free Manuals::
261 * GNU Free Documentation License::
266 --- The Detailed Node Listing ---
270 * Why:: Why learn Emacs Lisp?
271 * On Reading this Text:: Read, gain familiarity, pick up habits....
272 * Who You Are:: For whom this is written.
274 * Note for Novices:: You can read this as a novice.
279 * Lisp Lists:: What are lists?
280 * Run a Program:: Any list in Lisp is a program ready to run.
281 * Making Errors:: Generating an error message.
282 * Names & Definitions:: Names of symbols and function definitions.
283 * Lisp Interpreter:: What the Lisp interpreter does.
284 * Evaluation:: Running a program.
285 * Variables:: Returning a value from a variable.
286 * Arguments:: Passing information to a function.
287 * set & setq:: Setting the value of a variable.
288 * Summary:: The major points.
289 * Error Message Exercises::
293 * Numbers Lists:: List have numbers, other lists, in them.
294 * Lisp Atoms:: Elemental entities.
295 * Whitespace in Lists:: Formatting lists to be readable.
296 * Typing Lists:: How GNU Emacs helps you type lists.
300 * Complications:: Variables, Special forms, Lists within.
301 * Byte Compiling:: Specially processing code for speed.
305 * How the Interpreter Acts:: Returns and Side Effects...
306 * Evaluating Inner Lists:: Lists within lists...
310 * fill-column Example::
311 * Void Function:: The error message for a symbol
313 * Void Variable:: The error message for a symbol without a value.
317 * Data types:: Types of data passed to a function.
318 * Args as Variable or List:: An argument can be the value
319 of a variable or list.
320 * Variable Number of Arguments:: Some functions may take a
321 variable number of arguments.
322 * Wrong Type of Argument:: Passing an argument of the wrong type
324 * message:: A useful function for sending messages.
326 Setting the Value of a Variable
328 * Using set:: Setting values.
329 * Using setq:: Setting a quoted value.
330 * Counting:: Using @code{setq} to count.
332 Practicing Evaluation
334 * How to Evaluate:: Typing editing commands or @kbd{C-x C-e}
336 * Buffer Names:: Buffers and files are different.
337 * Getting Buffers:: Getting a buffer itself, not merely its name.
338 * Switching Buffers:: How to change to another buffer.
339 * Buffer Size & Locations:: Where point is located and the size of
341 * Evaluation Exercise::
343 How To Write Function Definitions
345 * Primitive Functions::
346 * defun:: The @code{defun} macro.
347 * Install:: Install a function definition.
348 * Interactive:: Making a function interactive.
349 * Interactive Options:: Different options for @code{interactive}.
350 * Permanent Installation:: Installing code permanently.
351 * let:: Creating and initializing local variables.
353 * else:: If--then--else expressions.
354 * Truth & Falsehood:: What Lisp considers false and true.
355 * save-excursion:: Keeping track of point and buffer.
359 Install a Function Definition
361 * Effect of installation::
362 * Change a defun:: How to change a function definition.
364 Make a Function Interactive
366 * Interactive multiply-by-seven:: An overview.
367 * multiply-by-seven in detail:: The interactive version.
371 * Prevent confusion::
372 * Parts of let Expression::
373 * Sample let Expression::
374 * Uninitialized let Variables::
376 The @code{if} Special Form
378 * if in more detail::
379 * type-of-animal in detail:: An example of an @code{if} expression.
381 Truth and Falsehood in Emacs Lisp
383 * nil explained:: @code{nil} has two meanings.
385 @code{save-excursion}
387 * Point and mark:: A review of various locations.
388 * Template for save-excursion::
390 A Few Buffer-Related Functions
392 * Finding More:: How to find more information.
393 * simplified-beginning-of-buffer:: Shows @code{goto-char},
394 @code{point-min}, and @code{push-mark}.
395 * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}.
396 * append-to-buffer:: Uses @code{save-excursion} and
397 @code{insert-buffer-substring}.
398 * Buffer Related Review:: Review.
401 The Definition of @code{mark-whole-buffer}
403 * mark-whole-buffer overview::
404 * Body of mark-whole-buffer:: Only three lines of code.
406 The Definition of @code{append-to-buffer}
408 * append-to-buffer overview::
409 * append interactive:: A two part interactive expression.
410 * append-to-buffer body:: Incorporates a @code{let} expression.
411 * append save-excursion:: How the @code{save-excursion} works.
413 A Few More Complex Functions
415 * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}.
416 * insert-buffer:: Read-only, and with @code{or}.
417 * beginning-of-buffer:: Shows @code{goto-char},
418 @code{point-min}, and @code{push-mark}.
419 * Second Buffer Related Review::
420 * optional Exercise::
422 The Definition of @code{insert-buffer}
424 * insert-buffer code::
425 * insert-buffer interactive:: When you can read, but not write.
426 * insert-buffer body:: The body has an @code{or} and a @code{let}.
427 * if & or:: Using an @code{if} instead of an @code{or}.
428 * Insert or:: How the @code{or} expression works.
429 * Insert let:: Two @code{save-excursion} expressions.
430 * New insert-buffer::
432 The Interactive Expression in @code{insert-buffer}
434 * Read-only buffer:: When a buffer cannot be modified.
435 * b for interactive:: An existing buffer or else its name.
437 Complete Definition of @code{beginning-of-buffer}
439 * Optional Arguments::
440 * beginning-of-buffer opt arg:: Example with optional argument.
441 * beginning-of-buffer complete::
443 @code{beginning-of-buffer} with an Argument
445 * Disentangle beginning-of-buffer::
446 * Large buffer case::
447 * Small buffer case::
449 Narrowing and Widening
451 * Narrowing advantages:: The advantages of narrowing
452 * save-restriction:: The @code{save-restriction} special form.
453 * what-line:: The number of the line that point is on.
456 @code{car}, @code{cdr}, @code{cons}: Fundamental Functions
458 * Strange Names:: An historical aside: why the strange names?
459 * car & cdr:: Functions for extracting part of a list.
460 * cons:: Constructing a list.
461 * nthcdr:: Calling @code{cdr} repeatedly.
463 * setcar:: Changing the first element of a list.
464 * setcdr:: Changing the rest of a list.
470 * length:: How to find the length of a list.
472 Cutting and Storing Text
474 * Storing Text:: Text is stored in a list.
475 * zap-to-char:: Cutting out text up to a character.
476 * kill-region:: Cutting text out of a region.
477 * copy-region-as-kill:: A definition for copying text.
478 * Digression into C:: Minor note on C programming language macros.
479 * defvar:: How to give a variable an initial value.
480 * cons & search-fwd Review::
485 * Complete zap-to-char:: The complete implementation.
486 * zap-to-char interactive:: A three part interactive expression.
487 * zap-to-char body:: A short overview.
488 * search-forward:: How to search for a string.
489 * progn:: The @code{progn} special form.
490 * Summing up zap-to-char:: Using @code{point} and @code{search-forward}.
494 * Complete kill-region:: The function definition.
495 * condition-case:: Dealing with a problem.
498 @code{copy-region-as-kill}
500 * Complete copy-region-as-kill:: The complete function definition.
501 * copy-region-as-kill body:: The body of @code{copy-region-as-kill}.
503 The Body of @code{copy-region-as-kill}
505 * last-command & this-command::
506 * kill-append function::
507 * kill-new function::
509 Initializing a Variable with @code{defvar}
511 * See variable current value::
512 * defvar and asterisk::
514 How Lists are Implemented
517 * Symbols as Chest:: Exploring a powerful metaphor.
522 * Kill Ring Overview::
523 * kill-ring-yank-pointer:: The kill ring is a list.
524 * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable.
528 * while:: Causing a stretch of code to repeat.
530 * Recursion:: Causing a function to call itself.
535 * Looping with while:: Repeat so long as test returns true.
536 * Loop Example:: A @code{while} loop that uses a list.
537 * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}.
538 * Incrementing Loop:: A loop with an incrementing counter.
539 * Incrementing Loop Details::
540 * Decrementing Loop:: A loop with a decrementing counter.
542 Details of an Incrementing Loop
544 * Incrementing Example:: Counting pebbles in a triangle.
545 * Inc Example parts:: The parts of the function definition.
546 * Inc Example altogether:: Putting the function definition together.
548 Loop with a Decrementing Counter
550 * Decrementing Example:: More pebbles on the beach.
551 * Dec Example parts:: The parts of the function definition.
552 * Dec Example altogether:: Putting the function definition together.
554 Save your time: @code{dolist} and @code{dotimes}
561 * Building Robots:: Same model, different serial number ...
562 * Recursive Definition Parts:: Walk until you stop ...
563 * Recursion with list:: Using a list as the test whether to recurse.
564 * Recursive triangle function::
565 * Recursion with cond::
566 * Recursive Patterns:: Often used templates.
567 * No Deferment:: Don't store up work ...
568 * No deferment solution::
570 Recursion in Place of a Counter
572 * Recursive Example arg of 1 or 2::
573 * Recursive Example arg of 3 or 4::
581 Regular Expression Searches
583 * sentence-end:: The regular expression for @code{sentence-end}.
584 * re-search-forward:: Very similar to @code{search-forward}.
585 * forward-sentence:: A straightforward example of regexp search.
586 * forward-paragraph:: A somewhat complex example.
587 * etags:: How to create your own @file{TAGS} table.
589 * re-search Exercises::
591 @code{forward-sentence}
593 * Complete forward-sentence::
594 * fwd-sentence while loops:: Two @code{while} loops.
595 * fwd-sentence re-search:: A regular expression search.
597 @code{forward-paragraph}: a Goldmine of Functions
599 * forward-paragraph in brief:: Key parts of the function definition.
600 * fwd-para let:: The @code{let*} expression.
601 * fwd-para while:: The forward motion @code{while} loop.
603 Counting: Repetition and Regexps
606 * @value{COUNT-WORDS}:: Use a regexp, but find a problem.
607 * recursive-count-words:: Start with case of no words in region.
608 * Counting Exercise::
610 The @code{@value{COUNT-WORDS}} Function
612 * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop.
613 * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}.
615 Counting Words in a @code{defun}
617 * Divide and Conquer::
618 * Words and Symbols:: What to count?
619 * Syntax:: What constitutes a word or symbol?
620 * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}.
621 * Several defuns:: Counting several defuns in a file.
622 * Find a File:: Do you want to look at a file?
623 * lengths-list-file:: A list of the lengths of many definitions.
624 * Several files:: Counting in definitions in different files.
625 * Several files recursively:: Recursively counting in different files.
626 * Prepare the data:: Prepare the data for display in a graph.
628 Count Words in @code{defuns} in Different Files
630 * lengths-list-many-files:: Return a list of the lengths of defuns.
631 * append:: Attach one list to another.
633 Prepare the Data for Display in a Graph
635 * Data for Display in Detail::
636 * Sorting:: Sorting lists.
637 * Files List:: Making a list of files.
638 * Counting function definitions::
642 * Columns of a graph::
643 * graph-body-print:: How to print the body of a graph.
644 * recursive-graph-body-print::
646 * Line Graph Exercise::
648 Your @file{.emacs} File
650 * Default Configuration::
651 * Site-wide Init:: You can write site-wide init files.
652 * defcustom:: Emacs will write code for you.
653 * Beginning init File:: How to write a @file{.emacs} init file.
654 * Text and Auto-fill:: Automatically wrap lines.
655 * Mail Aliases:: Use abbreviations for email addresses.
656 * Indent Tabs Mode:: Don't use tabs with @TeX{}
657 * Keybindings:: Create some personal keybindings.
658 * Keymaps:: More about key binding.
659 * Loading Files:: Load (i.e., evaluate) files automatically.
660 * Autoload:: Make functions available.
661 * Simple Extension:: Define a function; bind it to a key.
662 * X11 Colors:: Colors in X.
664 * Mode Line:: How to customize your mode line.
668 * debug:: How to use the built-in debugger.
669 * debug-on-entry:: Start debugging when you call a function.
670 * debug-on-quit:: Start debugging when you quit with @kbd{C-g}.
671 * edebug:: How to use Edebug, a source level debugger.
672 * Debugging Exercises::
674 Handling the Kill Ring
676 * What the Kill Ring Does::
678 * yank:: Paste a copy of a clipped element.
679 * yank-pop:: Insert element pointed to.
682 The @code{current-kill} Function
684 * Code for current-kill::
685 * Understanding current-kill::
687 @code{current-kill} in Outline
689 * Body of current-kill::
690 * Digression concerning error:: How to mislead humans, but not computers.
691 * Determining the Element::
693 A Graph with Labeled Axes
696 * print-graph Varlist:: @code{let} expression in @code{print-graph}.
697 * print-Y-axis:: Print a label for the vertical axis.
698 * print-X-axis:: Print a horizontal label.
699 * Print Whole Graph:: The function to print a complete graph.
701 The @code{print-Y-axis} Function
703 * print-Y-axis in Detail::
704 * Height of label:: What height for the Y axis?
705 * Compute a Remainder:: How to compute the remainder of a division.
706 * Y Axis Element:: Construct a line for the Y axis.
707 * Y-axis-column:: Generate a list of Y axis labels.
708 * print-Y-axis Penultimate:: A not quite final version.
710 The @code{print-X-axis} Function
712 * Similarities differences:: Much like @code{print-Y-axis}, but not exactly.
713 * X Axis Tic Marks:: Create tic marks for the horizontal axis.
715 Printing the Whole Graph
717 * The final version:: A few changes.
718 * Test print-graph:: Run a short test.
719 * Graphing words in defuns:: Executing the final code.
720 * lambda:: How to write an anonymous function.
721 * mapcar:: Apply a function to elements of a list.
722 * Another Bug:: Yet another bug @dots{} most insidious.
723 * Final printed graph:: The graph itself!
731 Most of the GNU Emacs integrated environment is written in the programming
732 language called Emacs Lisp. The code written in this programming
733 language is the software---the sets of instructions---that tell the
734 computer what to do when you give it commands. Emacs is designed so
735 that you can write new code in Emacs Lisp and easily install it as an
736 extension to the editor.
738 (GNU Emacs is sometimes called an ``extensible editor'', but it does
739 much more than provide editing capabilities. It is better to refer to
740 Emacs as an ``extensible computing environment''. However, that
741 phrase is quite a mouthful. It is easier to refer to Emacs simply as
742 an editor. Moreover, everything you do in Emacs---find the Mayan date
743 and phases of the moon, simplify polynomials, debug code, manage
744 files, read letters, write books---all these activities are kinds of
745 editing in the most general sense of the word.)
748 * Why:: Why learn Emacs Lisp?
749 * On Reading this Text:: Read, gain familiarity, pick up habits....
750 * Who You Are:: For whom this is written.
752 * Note for Novices:: You can read this as a novice.
758 @unnumberedsec Why Study Emacs Lisp?
761 Although Emacs Lisp is usually thought of in association only with Emacs,
762 it is a full computer programming language. You can use Emacs Lisp as
763 you would any other programming language.
765 Perhaps you want to understand programming; perhaps you want to extend
766 Emacs; or perhaps you want to become a programmer. This introduction to
767 Emacs Lisp is designed to get you started: to guide you in learning the
768 fundamentals of programming, and more importantly, to show you how you
769 can teach yourself to go further.
771 @node On Reading this Text
772 @unnumberedsec On Reading this Text
774 All through this document, you will see little sample programs you can
775 run inside of Emacs. If you read this document in Info inside of GNU
776 Emacs, you can run the programs as they appear. (This is easy to do and
777 is explained when the examples are presented.) Alternatively, you can
778 read this introduction as a printed book while sitting beside a computer
779 running Emacs. (This is what I like to do; I like printed books.) If
780 you don't have a running Emacs beside you, you can still read this book,
781 but in this case, it is best to treat it as a novel or as a travel guide
782 to a country not yet visited: interesting, but not the same as being
785 Much of this introduction is dedicated to walkthroughs or guided tours
786 of code used in GNU Emacs. These tours are designed for two purposes:
787 first, to give you familiarity with real, working code (code you use
788 every day); and, second, to give you familiarity with the way Emacs
789 works. It is interesting to see how a working environment is
792 hope that you will pick up the habit of browsing through source code.
793 You can learn from it and mine it for ideas. Having GNU Emacs is like
794 having a dragon's cave of treasures.
796 In addition to learning about Emacs as an editor and Emacs Lisp as a
797 programming language, the examples and guided tours will give you an
798 opportunity to get acquainted with Emacs as a Lisp programming
799 environment. GNU Emacs supports programming and provides tools that
800 you will want to become comfortable using, such as @kbd{M-.} (the key
801 which invokes the @code{find-tag} command). You will also learn about
802 buffers and other objects that are part of the environment.
803 Learning about these features of Emacs is like learning new routes
804 around your home town.
807 In addition, I have written several programs as extended examples.
808 Although these are examples, the programs are real. I use them.
809 Other people use them. You may use them. Beyond the fragments of
810 programs used for illustrations, there is very little in here that is
811 ``just for teaching purposes''; what you see is used. This is a great
812 advantage of Emacs Lisp: it is easy to learn to use it for work.
815 Finally, I hope to convey some of the skills for using Emacs to
816 learn aspects of programming that you don't know. You can often use
817 Emacs to help you understand what puzzles you or to find out how to do
818 something new. This self-reliance is not only a pleasure, but an
822 @unnumberedsec For Whom This is Written
824 This text is written as an elementary introduction for people who are
825 not programmers. If you are a programmer, you may not be satisfied with
826 this primer. The reason is that you may have become expert at reading
827 reference manuals and be put off by the way this text is organized.
829 An expert programmer who reviewed this text said to me:
832 @i{I prefer to learn from reference manuals. I ``dive into'' each
833 paragraph, and ``come up for air'' between paragraphs.}
835 @i{When I get to the end of a paragraph, I assume that that subject is
836 done, finished, that I know everything I need (with the
837 possible exception of the case when the next paragraph starts talking
838 about it in more detail). I expect that a well written reference manual
839 will not have a lot of redundancy, and that it will have excellent
840 pointers to the (one) place where the information I want is.}
843 This introduction is not written for this person!
845 Firstly, I try to say everything at least three times: first, to
846 introduce it; second, to show it in context; and third, to show it in a
847 different context, or to review it.
849 Secondly, I hardly ever put all the information about a subject in one
850 place, much less in one paragraph. To my way of thinking, that imposes
851 too heavy a burden on the reader. Instead I try to explain only what
852 you need to know at the time. (Sometimes I include a little extra
853 information so you won't be surprised later when the additional
854 information is formally introduced.)
856 When you read this text, you are not expected to learn everything the
857 first time. Frequently, you need only make, as it were, a ``nodding
858 acquaintance'' with some of the items mentioned. My hope is that I have
859 structured the text and given you enough hints that you will be alert to
860 what is important, and concentrate on it.
862 You will need to ``dive into'' some paragraphs; there is no other way
863 to read them. But I have tried to keep down the number of such
864 paragraphs. This book is intended as an approachable hill, rather than
865 as a daunting mountain.
867 This introduction to @cite{Programming in Emacs Lisp} has a companion
870 @cite{The GNU Emacs Lisp Reference Manual}.
873 @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU
874 Emacs Lisp Reference Manual}.
876 The reference manual has more detail than this introduction. In the
877 reference manual, all the information about one topic is concentrated
878 in one place. You should turn to it if you are like the programmer
879 quoted above. And, of course, after you have read this
880 @cite{Introduction}, you will find the @cite{Reference Manual} useful
881 when you are writing your own programs.
884 @unnumberedsec Lisp History
887 Lisp was first developed in the late 1950s at the Massachusetts
888 Institute of Technology for research in artificial intelligence. The
889 great power of the Lisp language makes it superior for other purposes as
890 well, such as writing editor commands and integrated environments.
894 GNU Emacs Lisp is largely inspired by Maclisp, which was written at MIT
895 in the 1960s. It is somewhat inspired by Common Lisp, which became a
896 standard in the 1980s. However, Emacs Lisp is much simpler than Common
897 Lisp. (The standard Emacs distribution contains an optional extensions
898 file, @file{cl.el}, that adds many Common Lisp features to Emacs Lisp.)
900 @node Note for Novices
901 @unnumberedsec A Note for Novices
903 If you don't know GNU Emacs, you can still read this document
904 profitably. However, I recommend you learn Emacs, if only to learn to
905 move around your computer screen. You can teach yourself how to use
906 Emacs with the built-in tutorial. To use it, type @kbd{C-h t}. (This
907 means you press and release the @key{CTRL} key and the @kbd{h} at the
908 same time, and then press and release @kbd{t}.)
910 Also, I often refer to one of Emacs's standard commands by listing the
911 keys which you press to invoke the command and then giving the name of
912 the command in parentheses, like this: @kbd{M-C-\}
913 (@code{indent-region}). What this means is that the
914 @code{indent-region} command is customarily invoked by typing
915 @kbd{M-C-\}. (You can, if you wish, change the keys that are typed to
916 invoke the command; this is called @dfn{rebinding}. @xref{Keymaps, ,
917 Keymaps}.) The abbreviation @kbd{M-C-\} means that you type your
918 @key{META} key, @key{CTRL} key and @key{\} key all at the same time.
919 (On many modern keyboards the @key{META} key is labeled
921 Sometimes a combination like this is called a keychord, since it is
922 similar to the way you play a chord on a piano. If your keyboard does
923 not have a @key{META} key, the @key{ESC} key prefix is used in place
924 of it. In this case, @kbd{M-C-\} means that you press and release your
925 @key{ESC} key and then type the @key{CTRL} key and the @key{\} key at
926 the same time. But usually @kbd{M-C-\} means press the @key{CTRL} key
927 along with the key that is labeled @key{ALT} and, at the same time,
928 press the @key{\} key.
930 In addition to typing a lone keychord, you can prefix what you type
931 with @kbd{C-u}, which is called the ``universal argument''. The
932 @kbd{C-u} keychord passes an argument to the subsequent command.
933 Thus, to indent a region of plain text by 6 spaces, mark the region,
934 and then type @w{@kbd{C-u 6 M-C-\}}. (If you do not specify a number,
935 Emacs either passes the number 4 to the command or otherwise runs the
936 command differently than it would otherwise.) @xref{Arguments, ,
937 Numeric Arguments, emacs, The GNU Emacs Manual}.
939 If you are reading this in Info using GNU Emacs, you can read through
940 this whole document just by pressing the space bar, @key{SPC}.
941 (To learn about Info, type @kbd{C-h i} and then select Info.)
943 A note on terminology: when I use the word Lisp alone, I often am
944 referring to the various dialects of Lisp in general, but when I speak
945 of Emacs Lisp, I am referring to GNU Emacs Lisp in particular.
948 @unnumberedsec Thank You
950 My thanks to all who helped me with this book. My especial thanks to
951 @r{Jim Blandy}, @r{Noah Friedman}, @w{Jim Kingdon}, @r{Roland
952 McGrath}, @w{Frank Ritter}, @w{Randy Smith}, @w{Richard M.
953 Stallman}, and @w{Melissa Weisshaus}. My thanks also go to both
954 @w{Philip Johnson} and @w{David Stampe} for their patient
955 encouragement. My mistakes are my own.
967 @c ================ Beginning of main text ================
969 @c Start main text on right-hand (verso) page
972 \par\vfill\supereject
975 \par\vfill\supereject
977 \par\vfill\supereject
979 \par\vfill\supereject
983 @c Note: this resetting of the page number back to 1 causes TeX to gripe
984 @c about already having seen page numbers 1-4 before (in the preface):
985 @c pdfTeX warning (ext4): destination with the same identifier (name{1})
986 @c has been already used, duplicate ignored
987 @c I guess that is harmless (what happens if a later part of the text
988 @c makes a link to something in the first 4 pages though?).
989 @c E.g., note that the Emacs manual has a preface, but does not bother
990 @c resetting the page numbers back to 1 after that.
993 @evenheading @thispage @| @| @thischapter
994 @oddheading @thissection @| @| @thispage
998 @node List Processing
999 @chapter List Processing
1001 To the untutored eye, Lisp is a strange programming language. In Lisp
1002 code there are parentheses everywhere. Some people even claim that
1003 the name stands for ``Lots of Isolated Silly Parentheses''. But the
1004 claim is unwarranted. Lisp stands for LISt Processing, and the
1005 programming language handles @emph{lists} (and lists of lists) by
1006 putting them between parentheses. The parentheses mark the boundaries
1007 of the list. Sometimes a list is preceded by a single apostrophe or
1008 quotation mark, @samp{'}@footnote{The single apostrophe or quotation
1009 mark is an abbreviation for the function @code{quote}; you need not
1010 think about functions now; functions are defined in @ref{Making
1011 Errors, , Generate an Error Message}.} Lists are the basis of Lisp.
1014 * Lisp Lists:: What are lists?
1015 * Run a Program:: Any list in Lisp is a program ready to run.
1016 * Making Errors:: Generating an error message.
1017 * Names & Definitions:: Names of symbols and function definitions.
1018 * Lisp Interpreter:: What the Lisp interpreter does.
1019 * Evaluation:: Running a program.
1020 * Variables:: Returning a value from a variable.
1021 * Arguments:: Passing information to a function.
1022 * set & setq:: Setting the value of a variable.
1023 * Summary:: The major points.
1024 * Error Message Exercises::
1031 In Lisp, a list looks like this: @code{'(rose violet daisy buttercup)}.
1032 This list is preceded by a single apostrophe. It could just as well be
1033 written as follows, which looks more like the kind of list you are likely
1034 to be familiar with:
1046 The elements of this list are the names of the four different flowers,
1047 separated from each other by whitespace and surrounded by parentheses,
1048 like flowers in a field with a stone wall around them.
1049 @cindex Flowers in a field
1052 * Numbers Lists:: List have numbers, other lists, in them.
1053 * Lisp Atoms:: Elemental entities.
1054 * Whitespace in Lists:: Formatting lists to be readable.
1055 * Typing Lists:: How GNU Emacs helps you type lists.
1060 @unnumberedsubsec Numbers, Lists inside of Lists
1063 Lists can also have numbers in them, as in this list: @code{(+ 2 2)}.
1064 This list has a plus-sign, @samp{+}, followed by two @samp{2}s, each
1065 separated by whitespace.
1067 In Lisp, both data and programs are represented the same way; that is,
1068 they are both lists of words, numbers, or other lists, separated by
1069 whitespace and surrounded by parentheses. (Since a program looks like
1070 data, one program may easily serve as data for another; this is a very
1071 powerful feature of Lisp.) (Incidentally, these two parenthetical
1072 remarks are @emph{not} Lisp lists, because they contain @samp{;} and
1073 @samp{.} as punctuation marks.)
1076 Here is another list, this time with a list inside of it:
1079 '(this list has (a list inside of it))
1082 The components of this list are the words @samp{this}, @samp{list},
1083 @samp{has}, and the list @samp{(a list inside of it)}. The interior
1084 list is made up of the words @samp{a}, @samp{list}, @samp{inside},
1085 @samp{of}, @samp{it}.
1088 @subsection Lisp Atoms
1091 In Lisp, what we have been calling words are called @dfn{atoms}. This
1092 term comes from the historical meaning of the word atom, which means
1093 ``indivisible''. As far as Lisp is concerned, the words we have been
1094 using in the lists cannot be divided into any smaller parts and still
1095 mean the same thing as part of a program; likewise with numbers and
1096 single character symbols like @samp{+}. On the other hand, unlike an
1097 ancient atom, a list can be split into parts. (@xref{car cdr & cons,
1098 , @code{car} @code{cdr} & @code{cons} Fundamental Functions}.)
1100 In a list, atoms are separated from each other by whitespace. They can be
1101 right next to a parenthesis.
1103 @cindex @samp{empty list} defined
1104 Technically speaking, a list in Lisp consists of parentheses surrounding
1105 atoms separated by whitespace or surrounding other lists or surrounding
1106 both atoms and other lists. A list can have just one atom in it or
1107 have nothing in it at all. A list with nothing in it looks like this:
1108 @code{()}, and is called the @dfn{empty list}. Unlike anything else, an
1109 empty list is considered both an atom and a list at the same time.
1111 @cindex Symbolic expressions, introduced
1112 @cindex @samp{expression} defined
1113 @cindex @samp{form} defined
1114 The printed representation of both atoms and lists are called
1115 @dfn{symbolic expressions} or, more concisely, @dfn{s-expressions}.
1116 The word @dfn{expression} by itself can refer to either the printed
1117 representation, or to the atom or list as it is held internally in the
1118 computer. Often, people use the term @dfn{expression}
1119 indiscriminately. (Also, in many texts, the word @dfn{form} is used
1120 as a synonym for expression.)
1122 Incidentally, the atoms that make up our universe were named such when
1123 they were thought to be indivisible; but it has been found that physical
1124 atoms are not indivisible. Parts can split off an atom or it can
1125 fission into two parts of roughly equal size. Physical atoms were named
1126 prematurely, before their truer nature was found. In Lisp, certain
1127 kinds of atom, such as an array, can be separated into parts; but the
1128 mechanism for doing this is different from the mechanism for splitting a
1129 list. As far as list operations are concerned, the atoms of a list are
1132 As in English, the meanings of the component letters of a Lisp atom
1133 are different from the meaning the letters make as a word. For
1134 example, the word for the South American sloth, the @samp{ai}, is
1135 completely different from the two words, @samp{a}, and @samp{i}.
1137 There are many kinds of atom in nature but only a few in Lisp: for
1138 example, @dfn{numbers}, such as 37, 511, or 1729, and @dfn{symbols}, such
1139 as @samp{+}, @samp{foo}, or @samp{forward-line}. The words we have
1140 listed in the examples above are all symbols. In everyday Lisp
1141 conversation, the word ``atom'' is not often used, because programmers
1142 usually try to be more specific about what kind of atom they are dealing
1143 with. Lisp programming is mostly about symbols (and sometimes numbers)
1144 within lists. (Incidentally, the preceding three word parenthetical
1145 remark is a proper list in Lisp, since it consists of atoms, which in
1146 this case are symbols, separated by whitespace and enclosed by
1147 parentheses, without any non-Lisp punctuation.)
1150 Text between double quotation marks---even sentences or
1151 paragraphs---is also an atom. Here is an example:
1152 @cindex Text between double quotation marks
1155 '(this list includes "text between quotation marks.")
1158 @cindex @samp{string} defined
1160 In Lisp, all of the quoted text including the punctuation mark and the
1161 blank spaces is a single atom. This kind of atom is called a
1162 @dfn{string} (for ``string of characters'') and is the sort of thing that
1163 is used for messages that a computer can print for a human to read.
1164 Strings are a different kind of atom than numbers or symbols and are
1167 @node Whitespace in Lists
1168 @subsection Whitespace in Lists
1169 @cindex Whitespace in lists
1172 The amount of whitespace in a list does not matter. From the point of view
1173 of the Lisp language,
1184 is exactly the same as this:
1187 '(this list looks like this)
1190 Both examples show what to Lisp is the same list, the list made up of
1191 the symbols @samp{this}, @samp{list}, @samp{looks}, @samp{like}, and
1192 @samp{this} in that order.
1194 Extra whitespace and newlines are designed to make a list more readable
1195 by humans. When Lisp reads the expression, it gets rid of all the extra
1196 whitespace (but it needs to have at least one space between atoms in
1197 order to tell them apart.)
1199 Odd as it seems, the examples we have seen cover almost all of what Lisp
1200 lists look like! Every other list in Lisp looks more or less like one
1201 of these examples, except that the list may be longer and more complex.
1202 In brief, a list is between parentheses, a string is between quotation
1203 marks, a symbol looks like a word, and a number looks like a number.
1204 (For certain situations, square brackets, dots and a few other special
1205 characters may be used; however, we will go quite far without them.)
1208 @subsection GNU Emacs Helps You Type Lists
1209 @cindex Help typing lists
1210 @cindex Formatting help
1212 When you type a Lisp expression in GNU Emacs using either Lisp
1213 Interaction mode or Emacs Lisp mode, you have available to you several
1214 commands to format the Lisp expression so it is easy to read. For
1215 example, pressing the @key{TAB} key automatically indents the line the
1216 cursor is on by the right amount. A command to properly indent the
1217 code in a region is customarily bound to @kbd{M-C-\}. Indentation is
1218 designed so that you can see which elements of a list belong to which
1219 list---elements of a sub-list are indented more than the elements of
1222 In addition, when you type a closing parenthesis, Emacs momentarily
1223 jumps the cursor back to the matching opening parenthesis, so you can
1224 see which one it is. This is very useful, since every list you type
1225 in Lisp must have its closing parenthesis match its opening
1226 parenthesis. (@xref{Major Modes, , Major Modes, emacs, The GNU Emacs
1227 Manual}, for more information about Emacs's modes.)
1230 @section Run a Program
1231 @cindex Run a program
1232 @cindex Program, running one
1234 @cindex @samp{evaluate} defined
1235 A list in Lisp---any list---is a program ready to run. If you run it
1236 (for which the Lisp jargon is @dfn{evaluate}), the computer will do one
1237 of three things: do nothing except return to you the list itself; send
1238 you an error message; or, treat the first symbol in the list as a
1239 command to do something. (Usually, of course, it is the last of these
1240 three things that you really want!)
1242 @c use code for the single apostrophe, not samp.
1243 The single apostrophe, @code{'}, that I put in front of some of the
1244 example lists in preceding sections is called a @dfn{quote}; when it
1245 precedes a list, it tells Lisp to do nothing with the list, other than
1246 take it as it is written. But if there is no quote preceding a list,
1247 the first item of the list is special: it is a command for the computer
1248 to obey. (In Lisp, these commands are called @emph{functions}.) The list
1249 @code{(+ 2 2)} shown above did not have a quote in front of it, so Lisp
1250 understands that the @code{+} is an instruction to do something with the
1251 rest of the list: add the numbers that follow.
1254 If you are reading this inside of GNU Emacs in Info, here is how you can
1255 evaluate such a list: place your cursor immediately after the right
1256 hand parenthesis of the following list and then type @kbd{C-x C-e}:
1262 @c use code for the number four, not samp.
1264 You will see the number @code{4} appear in the echo area. (In the
1265 jargon, what you have just done is ``evaluate the list.'' The echo area
1266 is the line at the bottom of the screen that displays or ``echoes''
1267 text.) Now try the same thing with a quoted list: place the cursor
1268 right after the following list and type @kbd{C-x C-e}:
1271 '(this is a quoted list)
1275 You will see @code{(this is a quoted list)} appear in the echo area.
1277 @cindex Lisp interpreter, explained
1278 @cindex Interpreter, Lisp, explained
1279 In both cases, what you are doing is giving a command to the program
1280 inside of GNU Emacs called the @dfn{Lisp interpreter}---giving the
1281 interpreter a command to evaluate the expression. The name of the Lisp
1282 interpreter comes from the word for the task done by a human who comes
1283 up with the meaning of an expression---who ``interprets'' it.
1285 You can also evaluate an atom that is not part of a list---one that is
1286 not surrounded by parentheses; again, the Lisp interpreter translates
1287 from the humanly readable expression to the language of the computer.
1288 But before discussing this (@pxref{Variables}), we will discuss what the
1289 Lisp interpreter does when you make an error.
1292 @section Generate an Error Message
1293 @cindex Generate an error message
1294 @cindex Error message generation
1296 Partly so you won't worry if you do it accidentally, we will now give
1297 a command to the Lisp interpreter that generates an error message.
1298 This is a harmless activity; and indeed, we will often try to generate
1299 error messages intentionally. Once you understand the jargon, error
1300 messages can be informative. Instead of being called ``error''
1301 messages, they should be called ``help'' messages. They are like
1302 signposts to a traveler in a strange country; deciphering them can be
1303 hard, but once understood, they can point the way.
1305 The error message is generated by a built-in GNU Emacs debugger. We
1306 will ``enter the debugger''. You get out of the debugger by typing @code{q}.
1308 What we will do is evaluate a list that is not quoted and does not
1309 have a meaningful command as its first element. Here is a list almost
1310 exactly the same as the one we just used, but without the single-quote
1311 in front of it. Position the cursor right after it and type @kbd{C-x
1315 (this is an unquoted list)
1320 What you see depends on which version of Emacs you are running. GNU
1321 Emacs version 22 provides more information than version 20 and before.
1322 First, the more recent result of generating an error; then the
1323 earlier, version 20 result.
1327 In GNU Emacs version 22, a @file{*Backtrace*} window will open up and
1328 you will see the following in it:
1331 A @file{*Backtrace*} window will open up and you should see the
1336 ---------- Buffer: *Backtrace* ----------
1337 Debugger entered--Lisp error: (void-function this)
1338 (this is an unquoted list)
1339 eval((this is an unquoted list))
1340 eval-last-sexp-1(nil)
1342 call-interactively(eval-last-sexp)
1343 ---------- Buffer: *Backtrace* ----------
1349 Your cursor will be in this window (you may have to wait a few seconds
1350 before it becomes visible). To quit the debugger and make the
1351 debugger window go away, type:
1358 Please type @kbd{q} right now, so you become confident that you can
1359 get out of the debugger. Then, type @kbd{C-x C-e} again to re-enter
1362 @cindex @samp{function} defined
1363 Based on what we already know, we can almost read this error message.
1365 You read the @file{*Backtrace*} buffer from the bottom up; it tells
1366 you what Emacs did. When you typed @kbd{C-x C-e}, you made an
1367 interactive call to the command @code{eval-last-sexp}. @code{eval} is
1368 an abbreviation for ``evaluate'' and @code{sexp} is an abbreviation for
1369 ``symbolic expression''. The command means ``evaluate last symbolic
1370 expression'', which is the expression just before your cursor.
1372 Each line above tells you what the Lisp interpreter evaluated next.
1373 The most recent action is at the top. The buffer is called the
1374 @file{*Backtrace*} buffer because it enables you to track Emacs
1378 At the top of the @file{*Backtrace*} buffer, you see the line:
1381 Debugger entered--Lisp error: (void-function this)
1385 The Lisp interpreter tried to evaluate the first atom of the list, the
1386 word @samp{this}. It is this action that generated the error message
1387 @samp{void-function this}.
1389 The message contains the words @samp{void-function} and @samp{this}.
1391 @cindex @samp{function} defined
1392 The word @samp{function} was mentioned once before. It is a very
1393 important word. For our purposes, we can define it by saying that a
1394 @dfn{function} is a set of instructions to the computer that tell the
1395 computer to do something.
1397 Now we can begin to understand the error message: @samp{void-function
1398 this}. The function (that is, the word @samp{this}) does not have a
1399 definition of any set of instructions for the computer to carry out.
1401 The slightly odd word, @samp{void-function}, is designed to cover the
1402 way Emacs Lisp is implemented, which is that when a symbol does not
1403 have a function definition attached to it, the place that should
1404 contain the instructions is ``void''.
1406 On the other hand, since we were able to add 2 plus 2 successfully, by
1407 evaluating @code{(+ 2 2)}, we can infer that the symbol @code{+} must
1408 have a set of instructions for the computer to obey and those
1409 instructions must be to add the numbers that follow the @code{+}.
1411 It is possible to prevent Emacs entering the debugger in cases like
1412 this. We do not explain how to do that here, but we will mention what
1413 the result looks like, because you may encounter a similar situation
1414 if there is a bug in some Emacs code that you are using. In such
1415 cases, you will see only one line of error message; it will appear in
1416 the echo area and look like this:
1419 Symbol's function definition is void:@: this
1424 (Also, your terminal may beep at you---some do, some don't; and others
1425 blink. This is just a device to get your attention.)
1427 The message goes away as soon as you type a key, even just to
1430 We know the meaning of the word @samp{Symbol}. It refers to the first
1431 atom of the list, the word @samp{this}. The word @samp{function}
1432 refers to the instructions that tell the computer what to do.
1433 (Technically, the symbol tells the computer where to find the
1434 instructions, but this is a complication we can ignore for the
1437 The error message can be understood: @samp{Symbol's function
1438 definition is void:@: this}. The symbol (that is, the word
1439 @samp{this}) lacks instructions for the computer to carry out.
1441 @node Names & Definitions
1442 @section Symbol Names and Function Definitions
1443 @cindex Symbol names
1445 We can articulate another characteristic of Lisp based on what we have
1446 discussed so far---an important characteristic: a symbol, like
1447 @code{+}, is not itself the set of instructions for the computer to
1448 carry out. Instead, the symbol is used, perhaps temporarily, as a way
1449 of locating the definition or set of instructions. What we see is the
1450 name through which the instructions can be found. Names of people
1451 work the same way. I can be referred to as @samp{Bob}; however, I am
1452 not the letters @samp{B}, @samp{o}, @samp{b} but am, or was, the
1453 consciousness consistently associated with a particular life-form.
1454 The name is not me, but it can be used to refer to me.
1456 In Lisp, one set of instructions can be attached to several names.
1457 For example, the computer instructions for adding numbers can be
1458 linked to the symbol @code{plus} as well as to the symbol @code{+}
1459 (and are in some dialects of Lisp). Among humans, I can be referred
1460 to as @samp{Robert} as well as @samp{Bob} and by other words as well.
1462 On the other hand, a symbol can have only one function definition
1463 attached to it at a time. Otherwise, the computer would be confused as
1464 to which definition to use. If this were the case among people, only
1465 one person in the world could be named @samp{Bob}. However, the function
1466 definition to which the name refers can be changed readily.
1467 (@xref{Install, , Install a Function Definition}.)
1469 Since Emacs Lisp is large, it is customary to name symbols in a way
1470 that identifies the part of Emacs to which the function belongs.
1471 Thus, all the names for functions that deal with Texinfo start with
1472 @samp{texinfo-} and those for functions that deal with reading mail
1473 start with @samp{rmail-}.
1475 @node Lisp Interpreter
1476 @section The Lisp Interpreter
1477 @cindex Lisp interpreter, what it does
1478 @cindex Interpreter, what it does
1480 Based on what we have seen, we can now start to figure out what the
1481 Lisp interpreter does when we command it to evaluate a list.
1482 First, it looks to see whether there is a quote before the list; if
1483 there is, the interpreter just gives us the list. On the other
1484 hand, if there is no quote, the interpreter looks at the first element
1485 in the list and sees whether it has a function definition. If it does,
1486 the interpreter carries out the instructions in the function definition.
1487 Otherwise, the interpreter prints an error message.
1489 This is how Lisp works. Simple. There are added complications which we
1490 will get to in a minute, but these are the fundamentals. Of course, to
1491 write Lisp programs, you need to know how to write function definitions
1492 and attach them to names, and how to do this without confusing either
1493 yourself or the computer.
1496 * Complications:: Variables, Special forms, Lists within.
1497 * Byte Compiling:: Specially processing code for speed.
1502 @unnumberedsubsec Complications
1505 Now, for the first complication. In addition to lists, the Lisp
1506 interpreter can evaluate a symbol that is not quoted and does not have
1507 parentheses around it. The Lisp interpreter will attempt to determine
1508 the symbol's value as a @dfn{variable}. This situation is described
1509 in the section on variables. (@xref{Variables}.)
1511 @cindex Special form
1512 The second complication occurs because some functions are unusual and
1513 do not work in the usual manner. Those that don't are called
1514 @dfn{special forms}. They are used for special jobs, like defining a
1515 function, and there are not many of them. In the next few chapters,
1516 you will be introduced to several of the more important special forms.
1518 As well as special forms, there are also @dfn{macros}. A macro
1519 is a construct defined in Lisp, which differs from a function in that it
1520 translates a Lisp expression into another expression that is to be
1521 evaluated in place of the original expression. (@xref{Lisp macro}.)
1523 For the purposes of this introduction, you do not need to worry too much
1524 about whether something is a special form, macro, or ordinary function.
1525 For example, @code{if} is a special form (@pxref{if}), but @code{when}
1526 is a macro (@pxref{Lisp macro}). In earlier versions of Emacs,
1527 @code{defun} was a special form, but now it is a macro (@pxref{defun}).
1528 It still behaves in the same way.
1530 The final complication is this: if the function that the
1531 Lisp interpreter is looking at is not a special form, and if it is part
1532 of a list, the Lisp interpreter looks to see whether the list has a list
1533 inside of it. If there is an inner list, the Lisp interpreter first
1534 figures out what it should do with the inside list, and then it works on
1535 the outside list. If there is yet another list embedded inside the
1536 inner list, it works on that one first, and so on. It always works on
1537 the innermost list first. The interpreter works on the innermost list
1538 first, to evaluate the result of that list. The result may be
1539 used by the enclosing expression.
1541 Otherwise, the interpreter works left to right, from one expression to
1544 @node Byte Compiling
1545 @subsection Byte Compiling
1546 @cindex Byte compiling
1548 One other aspect of interpreting: the Lisp interpreter is able to
1549 interpret two kinds of entity: humanly readable code, on which we will
1550 focus exclusively, and specially processed code, called @dfn{byte
1551 compiled} code, which is not humanly readable. Byte compiled code
1552 runs faster than humanly readable code.
1554 You can transform humanly readable code into byte compiled code by
1555 running one of the compile commands such as @code{byte-compile-file}.
1556 Byte compiled code is usually stored in a file that ends with a
1557 @file{.elc} extension rather than a @file{.el} extension. You will
1558 see both kinds of file in the @file{emacs/lisp} directory; the files
1559 to read are those with @file{.el} extensions.
1561 As a practical matter, for most things you might do to customize or
1562 extend Emacs, you do not need to byte compile; and I will not discuss
1563 the topic here. @xref{Byte Compilation, , Byte Compilation, elisp,
1564 The GNU Emacs Lisp Reference Manual}, for a full description of byte
1571 When the Lisp interpreter works on an expression, the term for the
1572 activity is called @dfn{evaluation}. We say that the interpreter
1573 ``evaluates the expression''. I've used this term several times before.
1574 The word comes from its use in everyday language, ``to ascertain the
1575 value or amount of; to appraise'', according to @cite{Webster's New
1576 Collegiate Dictionary}.
1579 * How the Interpreter Acts:: Returns and Side Effects...
1580 * Evaluating Inner Lists:: Lists within lists...
1584 @node How the Interpreter Acts
1585 @unnumberedsubsec How the Lisp Interpreter Acts
1588 @cindex @samp{returned value} explained
1589 After evaluating an expression, the Lisp interpreter will most likely
1590 @dfn{return} the value that the computer produces by carrying out the
1591 instructions it found in the function definition, or perhaps it will
1592 give up on that function and produce an error message. (The interpreter
1593 may also find itself tossed, so to speak, to a different function or it
1594 may attempt to repeat continually what it is doing for ever and ever in
1595 what is called an ``infinite loop''. These actions are less common; and
1596 we can ignore them.) Most frequently, the interpreter returns a value.
1598 @cindex @samp{side effect} defined
1599 At the same time the interpreter returns a value, it may do something
1600 else as well, such as move a cursor or copy a file; this other kind of
1601 action is called a @dfn{side effect}. Actions that we humans think are
1602 important, such as printing results, are often ``side effects'' to the
1603 Lisp interpreter. The jargon can sound peculiar, but it turns out that
1604 it is fairly easy to learn to use side effects.
1606 In summary, evaluating a symbolic expression most commonly causes the
1607 Lisp interpreter to return a value and perhaps carry out a side effect;
1608 or else produce an error.
1610 @node Evaluating Inner Lists
1611 @subsection Evaluating Inner Lists
1612 @cindex Inner list evaluation
1613 @cindex Evaluating inner lists
1615 If evaluation applies to a list that is inside another list, the outer
1616 list may use the value returned by the first evaluation as information
1617 when the outer list is evaluated. This explains why inner expressions
1618 are evaluated first: the values they return are used by the outer
1622 We can investigate this process by evaluating another addition example.
1623 Place your cursor after the following expression and type @kbd{C-x C-e}:
1630 The number 8 will appear in the echo area.
1632 What happens is that the Lisp interpreter first evaluates the inner
1633 expression, @code{(+ 3 3)}, for which the value 6 is returned; then it
1634 evaluates the outer expression as if it were written @code{(+ 2 6)}, which
1635 returns the value 8. Since there are no more enclosing expressions to
1636 evaluate, the interpreter prints that value in the echo area.
1638 Now it is easy to understand the name of the command invoked by the
1639 keystrokes @kbd{C-x C-e}: the name is @code{eval-last-sexp}. The
1640 letters @code{sexp} are an abbreviation for ``symbolic expression'', and
1641 @code{eval} is an abbreviation for ``evaluate''. The command means
1642 ``evaluate last symbolic expression''.
1644 As an experiment, you can try evaluating the expression by putting the
1645 cursor at the beginning of the next line immediately following the
1646 expression, or inside the expression.
1649 Here is another copy of the expression:
1656 If you place the cursor at the beginning of the blank line that
1657 immediately follows the expression and type @kbd{C-x C-e}, you will
1658 still get the value 8 printed in the echo area. Now try putting the
1659 cursor inside the expression. If you put it right after the next to
1660 last parenthesis (so it appears to sit on top of the last parenthesis),
1661 you will get a 6 printed in the echo area! This is because the command
1662 evaluates the expression @code{(+ 3 3)}.
1664 Now put the cursor immediately after a number. Type @kbd{C-x C-e} and
1665 you will get the number itself. In Lisp, if you evaluate a number, you
1666 get the number itself---this is how numbers differ from symbols. If you
1667 evaluate a list starting with a symbol like @code{+}, you will get a
1668 value returned that is the result of the computer carrying out the
1669 instructions in the function definition attached to that name. If a
1670 symbol by itself is evaluated, something different happens, as we will
1671 see in the next section.
1677 In Emacs Lisp, a symbol can have a value attached to it just as it can
1678 have a function definition attached to it. The two are different.
1679 The function definition is a set of instructions that a computer will
1680 obey. A value, on the other hand, is something, such as number or a
1681 name, that can vary (which is why such a symbol is called a variable).
1682 The value of a symbol can be any expression in Lisp, such as a symbol,
1683 number, list, or string. A symbol that has a value is often called a
1686 A symbol can have both a function definition and a value attached to
1687 it at the same time. Or it can have just one or the other.
1688 The two are separate. This is somewhat similar
1689 to the way the name Cambridge can refer to the city in Massachusetts
1690 and have some information attached to the name as well, such as
1691 ``great programming center''.
1694 (Incidentally, in Emacs Lisp, a symbol can have two
1695 other things attached to it, too: a property list and a documentation
1696 string; these are discussed later.)
1699 Another way to think about this is to imagine a symbol as being a chest
1700 of drawers. The function definition is put in one drawer, the value in
1701 another, and so on. What is put in the drawer holding the value can be
1702 changed without affecting the contents of the drawer holding the
1703 function definition, and vice versa.
1706 * fill-column Example::
1707 * Void Function:: The error message for a symbol
1709 * Void Variable:: The error message for a symbol without a value.
1713 @node fill-column Example
1714 @unnumberedsubsec @code{fill-column}, an Example Variable
1717 @findex fill-column, @r{an example variable}
1718 @cindex Example variable, @code{fill-column}
1719 @cindex Variable, example of, @code{fill-column}
1720 The variable @code{fill-column} illustrates a symbol with a value
1721 attached to it: in every GNU Emacs buffer, this symbol is set to some
1722 value, usually 72 or 70, but sometimes to some other value. To find the
1723 value of this symbol, evaluate it by itself. If you are reading this in
1724 Info inside of GNU Emacs, you can do this by putting the cursor after
1725 the symbol and typing @kbd{C-x C-e}:
1732 After I typed @kbd{C-x C-e}, Emacs printed the number 72 in my echo
1733 area. This is the value for which @code{fill-column} is set for me as I
1734 write this. It may be different for you in your Info buffer. Notice
1735 that the value returned as a variable is printed in exactly the same way
1736 as the value returned by a function carrying out its instructions. From
1737 the point of view of the Lisp interpreter, a value returned is a value
1738 returned. What kind of expression it came from ceases to matter once
1741 A symbol can have any value attached to it or, to use the jargon, we can
1742 @dfn{bind} the variable to a value: to a number, such as 72; to a
1743 string, @code{"such as this"}; to a list, such as @code{(spruce pine
1744 oak)}; we can even bind a variable to a function definition.
1746 A symbol can be bound to a value in several ways. @xref{set & setq, ,
1747 Setting the Value of a Variable}, for information about one way to do
1751 @subsection Error Message for a Symbol Without a Function
1752 @cindex Symbol without function error
1753 @cindex Error for symbol without function
1755 When we evaluated @code{fill-column} to find its value as a variable,
1756 we did not place parentheses around the word. This is because we did
1757 not intend to use it as a function name.
1759 If @code{fill-column} were the first or only element of a list, the
1760 Lisp interpreter would attempt to find the function definition
1761 attached to it. But @code{fill-column} has no function definition.
1762 Try evaluating this:
1770 You will create a @file{*Backtrace*} buffer that says:
1774 ---------- Buffer: *Backtrace* ----------
1775 Debugger entered--Lisp error: (void-function fill-column)
1778 eval-last-sexp-1(nil)
1780 call-interactively(eval-last-sexp)
1781 ---------- Buffer: *Backtrace* ----------
1786 (Remember, to quit the debugger and make the debugger window go away,
1787 type @kbd{q} in the @file{*Backtrace*} buffer.)
1791 In GNU Emacs 20 and before, you will produce an error message that says:
1794 Symbol's function definition is void:@: fill-column
1798 (The message will go away as soon as you move the cursor or type
1803 @subsection Error Message for a Symbol Without a Value
1804 @cindex Symbol without value error
1805 @cindex Error for symbol without value
1807 If you attempt to evaluate a symbol that does not have a value bound to
1808 it, you will receive an error message. You can see this by
1809 experimenting with our 2 plus 2 addition. In the following expression,
1810 put your cursor right after the @code{+}, before the first number 2,
1819 In GNU Emacs 22, you will create a @file{*Backtrace*} buffer that
1824 ---------- Buffer: *Backtrace* ----------
1825 Debugger entered--Lisp error: (void-variable +)
1827 eval-last-sexp-1(nil)
1829 call-interactively(eval-last-sexp)
1830 ---------- Buffer: *Backtrace* ----------
1835 (Again, you can quit the debugger by
1836 typing @kbd{q} in the @file{*Backtrace*} buffer.)
1838 This backtrace is different from the very first error message we saw,
1839 which said, @samp{Debugger entered--Lisp error: (void-function this)}.
1840 In this case, the function does not have a value as a variable; while
1841 in the other error message, the function (the word @samp{this}) did not
1844 In this experiment with the @code{+}, what we did was cause the Lisp
1845 interpreter to evaluate the @code{+} and look for the value of the
1846 variable instead of the function definition. We did this by placing the
1847 cursor right after the symbol rather than after the parenthesis of the
1848 enclosing list as we did before. As a consequence, the Lisp interpreter
1849 evaluated the preceding s-expression, which in this case was
1852 Since @code{+} does not have a value bound to it, just the function
1853 definition, the error message reported that the symbol's value as a
1858 In GNU Emacs version 20 and before, your error message will say:
1861 Symbol's value as variable is void:@: +
1865 The meaning is the same as in GNU Emacs 22.
1871 @cindex Passing information to functions
1873 To see how information is passed to functions, let's look again at
1874 our old standby, the addition of two plus two. In Lisp, this is written
1881 If you evaluate this expression, the number 4 will appear in your echo
1882 area. What the Lisp interpreter does is add the numbers that follow
1885 @cindex @samp{argument} defined
1886 The numbers added by @code{+} are called the @dfn{arguments} of the
1887 function @code{+}. These numbers are the information that is given to
1888 or @dfn{passed} to the function.
1890 The word ``argument'' comes from the way it is used in mathematics and
1891 does not refer to a disputation between two people; instead it refers to
1892 the information presented to the function, in this case, to the
1893 @code{+}. In Lisp, the arguments to a function are the atoms or lists
1894 that follow the function. The values returned by the evaluation of
1895 these atoms or lists are passed to the function. Different functions
1896 require different numbers of arguments; some functions require none at
1897 all.@footnote{It is curious to track the path by which the word ``argument''
1898 came to have two different meanings, one in mathematics and the other in
1899 everyday English. According to the @cite{Oxford English Dictionary},
1900 the word derives from the Latin for @samp{to make clear, prove}; thus it
1901 came to mean, by one thread of derivation, ``the evidence offered as
1902 proof'', which is to say, ``the information offered'', which led to its
1903 meaning in Lisp. But in the other thread of derivation, it came to mean
1904 ``to assert in a manner against which others may make counter
1905 assertions'', which led to the meaning of the word as a disputation.
1906 (Note here that the English word has two different definitions attached
1907 to it at the same time. By contrast, in Emacs Lisp, a symbol cannot
1908 have two different function definitions at the same time.)}
1911 * Data types:: Types of data passed to a function.
1912 * Args as Variable or List:: An argument can be the value
1913 of a variable or list.
1914 * Variable Number of Arguments:: Some functions may take a
1915 variable number of arguments.
1916 * Wrong Type of Argument:: Passing an argument of the wrong type
1918 * message:: A useful function for sending messages.
1922 @subsection Arguments' Data Types
1924 @cindex Types of data
1925 @cindex Arguments' data types
1927 The type of data that should be passed to a function depends on what
1928 kind of information it uses. The arguments to a function such as
1929 @code{+} must have values that are numbers, since @code{+} adds numbers.
1930 Other functions use different kinds of data for their arguments.
1934 For example, the @code{concat} function links together or unites two or
1935 more strings of text to produce a string. The arguments are strings.
1936 Concatenating the two character strings @code{abc}, @code{def} produces
1937 the single string @code{abcdef}. This can be seen by evaluating the
1941 (concat "abc" "def")
1945 The value produced by evaluating this expression is @code{"abcdef"}.
1947 A function such as @code{substring} uses both a string and numbers as
1948 arguments. The function returns a part of the string, a substring of
1949 the first argument. This function takes three arguments. Its first
1950 argument is the string of characters, the second and third arguments
1951 are numbers that indicate the beginning (inclusive) and end
1952 (exclusive) of the substring. The numbers are a count of the number
1953 of characters (including spaces and punctuation) from the beginning of
1954 the string. Note that the characters in a string are numbered from
1958 For example, if you evaluate the following:
1961 (substring "The quick brown fox jumped." 16 19)
1965 you will see @code{"fox"} appear in the echo area. The arguments are the
1966 string and the two numbers.
1968 Note that the string passed to @code{substring} is a single atom even
1969 though it is made up of several words separated by spaces. Lisp counts
1970 everything between the two quotation marks as part of the string,
1971 including the spaces. You can think of the @code{substring} function as
1972 a kind of ``atom smasher'' since it takes an otherwise indivisible atom
1973 and extracts a part. However, @code{substring} is only able to extract
1974 a substring from an argument that is a string, not from another type of
1975 atom such as a number or symbol.
1977 @node Args as Variable or List
1978 @subsection An Argument as the Value of a Variable or List
1980 An argument can be a symbol that returns a value when it is evaluated.
1981 For example, when the symbol @code{fill-column} by itself is evaluated,
1982 it returns a number. This number can be used in an addition.
1985 Position the cursor after the following expression and type @kbd{C-x
1993 The value will be a number two more than what you get by evaluating
1994 @code{fill-column} alone. For me, this is 74, because my value of
1995 @code{fill-column} is 72.
1997 As we have just seen, an argument can be a symbol that returns a value
1998 when evaluated. In addition, an argument can be a list that returns a
1999 value when it is evaluated. For example, in the following expression,
2000 the arguments to the function @code{concat} are the strings
2001 @w{@code{"The "}} and @w{@code{" red foxes."}} and the list
2002 @code{(number-to-string (+ 2 fill-column))}.
2004 @c For GNU Emacs 22, need number-to-string
2006 (concat "The " (number-to-string (+ 2 fill-column)) " red foxes.")
2010 If you evaluate this expression---and if, as with my Emacs,
2011 @code{fill-column} evaluates to 72---@code{"The 74 red foxes."} will
2012 appear in the echo area. (Note that you must put spaces after the
2013 word @samp{The} and before the word @samp{red} so they will appear in
2014 the final string. The function @code{number-to-string} converts the
2015 integer that the addition function returns to a string.
2016 @code{number-to-string} is also known as @code{int-to-string}.)
2018 @node Variable Number of Arguments
2019 @subsection Variable Number of Arguments
2020 @cindex Variable number of arguments
2021 @cindex Arguments, variable number of
2023 Some functions, such as @code{concat}, @code{+} or @code{*}, take any
2024 number of arguments. (The @code{*} is the symbol for multiplication.)
2025 This can be seen by evaluating each of the following expressions in
2026 the usual way. What you will see in the echo area is printed in this
2027 text after @samp{@result{}}, which you may read as ``evaluates to''.
2030 In the first set, the functions have no arguments:
2041 In this set, the functions have one argument each:
2052 In this set, the functions have three arguments each:
2056 (+ 3 4 5) @result{} 12
2058 (* 3 4 5) @result{} 60
2062 @node Wrong Type of Argument
2063 @subsection Using the Wrong Type Object as an Argument
2064 @cindex Wrong type of argument
2065 @cindex Argument, wrong type of
2067 When a function is passed an argument of the wrong type, the Lisp
2068 interpreter produces an error message. For example, the @code{+}
2069 function expects the values of its arguments to be numbers. As an
2070 experiment we can pass it the quoted symbol @code{hello} instead of a
2071 number. Position the cursor after the following expression and type
2079 When you do this you will generate an error message. What has happened
2080 is that @code{+} has tried to add the 2 to the value returned by
2081 @code{'hello}, but the value returned by @code{'hello} is the symbol
2082 @code{hello}, not a number. Only numbers can be added. So @code{+}
2083 could not carry out its addition.
2086 You will create and enter a @file{*Backtrace*} buffer that says:
2091 ---------- Buffer: *Backtrace* ----------
2092 Debugger entered--Lisp error:
2093 (wrong-type-argument number-or-marker-p hello)
2095 eval((+ 2 (quote hello)))
2096 eval-last-sexp-1(nil)
2098 call-interactively(eval-last-sexp)
2099 ---------- Buffer: *Backtrace* ----------
2104 As usual, the error message tries to be helpful and makes sense after you
2105 learn how to read it.@footnote{@code{(quote hello)} is an expansion of
2106 the abbreviation @code{'hello}.}
2108 The first part of the error message is straightforward; it says
2109 @samp{wrong type argument}. Next comes the mysterious jargon word
2110 @w{@samp{number-or-marker-p}}. This word is trying to tell you what
2111 kind of argument the @code{+} expected.
2113 The symbol @code{number-or-marker-p} says that the Lisp interpreter is
2114 trying to determine whether the information presented it (the value of
2115 the argument) is a number or a marker (a special object representing a
2116 buffer position). What it does is test to see whether the @code{+} is
2117 being given numbers to add. It also tests to see whether the
2118 argument is something called a marker, which is a specific feature of
2119 Emacs Lisp. (In Emacs, locations in a buffer are recorded as markers.
2120 When the mark is set with the @kbd{C-@@} or @kbd{C-@key{SPC}} command,
2121 its position is kept as a marker. The mark can be considered a
2122 number---the number of characters the location is from the beginning
2123 of the buffer.) In Emacs Lisp, @code{+} can be used to add the
2124 numeric value of marker positions as numbers.
2126 The @samp{p} of @code{number-or-marker-p} is the embodiment of a
2127 practice started in the early days of Lisp programming. The @samp{p}
2128 stands for ``predicate''. In the jargon used by the early Lisp
2129 researchers, a predicate refers to a function to determine whether some
2130 property is true or false. So the @samp{p} tells us that
2131 @code{number-or-marker-p} is the name of a function that determines
2132 whether it is true or false that the argument supplied is a number or
2133 a marker. Other Lisp symbols that end in @samp{p} include @code{zerop},
2134 a function that tests whether its argument has the value of zero, and
2135 @code{listp}, a function that tests whether its argument is a list.
2137 Finally, the last part of the error message is the symbol @code{hello}.
2138 This is the value of the argument that was passed to @code{+}. If the
2139 addition had been passed the correct type of object, the value passed
2140 would have been a number, such as 37, rather than a symbol like
2141 @code{hello}. But then you would not have got the error message.
2145 In GNU Emacs version 20 and before, the echo area displays an error
2149 Wrong type argument:@: number-or-marker-p, hello
2152 This says, in different words, the same as the top line of the
2153 @file{*Backtrace*} buffer.
2157 @subsection The @code{message} Function
2160 Like @code{+}, the @code{message} function takes a variable number of
2161 arguments. It is used to send messages to the user and is so useful
2162 that we will describe it here.
2165 A message is printed in the echo area. For example, you can print a
2166 message in your echo area by evaluating the following list:
2169 (message "This message appears in the echo area!")
2172 The whole string between double quotation marks is a single argument
2173 and is printed @i{in toto}. (Note that in this example, the message
2174 itself will appear in the echo area within double quotes; that is
2175 because you see the value returned by the @code{message} function. In
2176 most uses of @code{message} in programs that you write, the text will
2177 be printed in the echo area as a side-effect, without the quotes.
2178 @xref{multiply-by-seven in detail, , @code{multiply-by-seven} in
2179 detail}, for an example of this.)
2181 However, if there is a @samp{%s} in the quoted string of characters, the
2182 @code{message} function does not print the @samp{%s} as such, but looks
2183 to the argument that follows the string. It evaluates the second
2184 argument and prints the value at the location in the string where the
2188 You can see this by positioning the cursor after the following
2189 expression and typing @kbd{C-x C-e}:
2192 (message "The name of this buffer is: %s." (buffer-name))
2196 In Info, @code{"The name of this buffer is: *info*."} will appear in the
2197 echo area. The function @code{buffer-name} returns the name of the
2198 buffer as a string, which the @code{message} function inserts in place
2201 To print a value as an integer, use @samp{%d} in the same way as
2202 @samp{%s}. For example, to print a message in the echo area that
2203 states the value of the @code{fill-column}, evaluate the following:
2206 (message "The value of fill-column is %d." fill-column)
2210 On my system, when I evaluate this list, @code{"The value of
2211 fill-column is 72."} appears in my echo area@footnote{Actually, you
2212 can use @code{%s} to print a number. It is non-specific. @code{%d}
2213 prints only the part of a number left of a decimal point, and not
2214 anything that is not a number.}.
2216 If there is more than one @samp{%s} in the quoted string, the value of
2217 the first argument following the quoted string is printed at the
2218 location of the first @samp{%s} and the value of the second argument is
2219 printed at the location of the second @samp{%s}, and so on.
2222 For example, if you evaluate the following,
2226 (message "There are %d %s in the office!"
2227 (- fill-column 14) "pink elephants")
2232 a rather whimsical message will appear in your echo area. On my system
2233 it says, @code{"There are 58 pink elephants in the office!"}.
2235 The expression @code{(- fill-column 14)} is evaluated and the resulting
2236 number is inserted in place of the @samp{%d}; and the string in double
2237 quotes, @code{"pink elephants"}, is treated as a single argument and
2238 inserted in place of the @samp{%s}. (That is to say, a string between
2239 double quotes evaluates to itself, like a number.)
2241 Finally, here is a somewhat complex example that not only illustrates
2242 the computation of a number, but also shows how you can use an
2243 expression within an expression to generate the text that is substituted
2248 (message "He saw %d %s"
2252 "The quick brown foxes jumped." 16 21)
2257 In this example, @code{message} has three arguments: the string,
2258 @code{"He saw %d %s"}, the expression, @code{(- fill-column 32)}, and
2259 the expression beginning with the function @code{concat}. The value
2260 resulting from the evaluation of @code{(- fill-column 32)} is inserted
2261 in place of the @samp{%d}; and the value returned by the expression
2262 beginning with @code{concat} is inserted in place of the @samp{%s}.
2264 When your fill column is 70 and you evaluate the expression, the
2265 message @code{"He saw 38 red foxes leaping."} appears in your echo
2269 @section Setting the Value of a Variable
2270 @cindex Variable, setting value
2271 @cindex Setting value of variable
2273 @cindex @samp{bind} defined
2274 There are several ways by which a variable can be given a value. One of
2275 the ways is to use either the function @code{set} or the function
2276 @code{setq}. Another way is to use @code{let} (@pxref{let}). (The
2277 jargon for this process is to @dfn{bind} a variable to a value.)
2279 The following sections not only describe how @code{set} and @code{setq}
2280 work but also illustrate how arguments are passed.
2283 * Using set:: Setting values.
2284 * Using setq:: Setting a quoted value.
2285 * Counting:: Using @code{setq} to count.
2289 @subsection Using @code{set}
2292 To set the value of the symbol @code{flowers} to the list @code{'(rose
2293 violet daisy buttercup)}, evaluate the following expression by
2294 positioning the cursor after the expression and typing @kbd{C-x C-e}.
2297 (set 'flowers '(rose violet daisy buttercup))
2301 The list @code{(rose violet daisy buttercup)} will appear in the echo
2302 area. This is what is @emph{returned} by the @code{set} function. As a
2303 side effect, the symbol @code{flowers} is bound to the list; that is,
2304 the symbol @code{flowers}, which can be viewed as a variable, is given
2305 the list as its value. (This process, by the way, illustrates how a
2306 side effect to the Lisp interpreter, setting the value, can be the
2307 primary effect that we humans are interested in. This is because every
2308 Lisp function must return a value if it does not get an error, but it
2309 will only have a side effect if it is designed to have one.)
2311 After evaluating the @code{set} expression, you can evaluate the symbol
2312 @code{flowers} and it will return the value you just set. Here is the
2313 symbol. Place your cursor after it and type @kbd{C-x C-e}.
2320 When you evaluate @code{flowers}, the list
2321 @code{(rose violet daisy buttercup)} appears in the echo area.
2323 Incidentally, if you evaluate @code{'flowers}, the variable with a quote
2324 in front of it, what you will see in the echo area is the symbol itself,
2325 @code{flowers}. Here is the quoted symbol, so you can try this:
2331 Note also, that when you use @code{set}, you need to quote both
2332 arguments to @code{set}, unless you want them evaluated. Since we do
2333 not want either argument evaluated, neither the variable
2334 @code{flowers} nor the list @code{(rose violet daisy buttercup)}, both
2335 are quoted. (When you use @code{set} without quoting its first
2336 argument, the first argument is evaluated before anything else is
2337 done. If you did this and @code{flowers} did not have a value
2338 already, you would get an error message that the @samp{Symbol's value
2339 as variable is void}; on the other hand, if @code{flowers} did return
2340 a value after it was evaluated, the @code{set} would attempt to set
2341 the value that was returned. There are situations where this is the
2342 right thing for the function to do; but such situations are rare.)
2345 @subsection Using @code{setq}
2348 As a practical matter, you almost always quote the first argument to
2349 @code{set}. The combination of @code{set} and a quoted first argument
2350 is so common that it has its own name: the special form @code{setq}.
2351 This special form is just like @code{set} except that the first argument
2352 is quoted automatically, so you don't need to type the quote mark
2353 yourself. Also, as an added convenience, @code{setq} permits you to set
2354 several different variables to different values, all in one expression.
2356 To set the value of the variable @code{carnivores} to the list
2357 @code{'(lion tiger leopard)} using @code{setq}, the following expression
2361 (setq carnivores '(lion tiger leopard))
2365 This is exactly the same as using @code{set} except the first argument
2366 is automatically quoted by @code{setq}. (The @samp{q} in @code{setq}
2367 means @code{quote}.)
2370 With @code{set}, the expression would look like this:
2373 (set 'carnivores '(lion tiger leopard))
2376 Also, @code{setq} can be used to assign different values to
2377 different variables. The first argument is bound to the value
2378 of the second argument, the third argument is bound to the value of the
2379 fourth argument, and so on. For example, you could use the following to
2380 assign a list of trees to the symbol @code{trees} and a list of herbivores
2381 to the symbol @code{herbivores}:
2385 (setq trees '(pine fir oak maple)
2386 herbivores '(gazelle antelope zebra))
2391 (The expression could just as well have been on one line, but it might
2392 not have fit on a page; and humans find it easier to read nicely
2395 Although I have been using the term ``assign'', there is another way of
2396 thinking about the workings of @code{set} and @code{setq}; and that is to
2397 say that @code{set} and @code{setq} make the symbol @emph{point} to the
2398 list. This latter way of thinking is very common and in forthcoming
2399 chapters we shall come upon at least one symbol that has ``pointer'' as
2400 part of its name. The name is chosen because the symbol has a value,
2401 specifically a list, attached to it; or, expressed another way,
2402 the symbol is set to ``point'' to the list.
2405 @subsection Counting
2408 Here is an example that shows how to use @code{setq} in a counter. You
2409 might use this to count how many times a part of your program repeats
2410 itself. First set a variable to zero; then add one to the number each
2411 time the program repeats itself. To do this, you need a variable that
2412 serves as a counter, and two expressions: an initial @code{setq}
2413 expression that sets the counter variable to zero; and a second
2414 @code{setq} expression that increments the counter each time it is
2419 (setq counter 0) ; @r{Let's call this the initializer.}
2421 (setq counter (+ counter 1)) ; @r{This is the incrementer.}
2423 counter ; @r{This is the counter.}
2428 (The text following the @samp{;} are comments. @xref{Change a
2429 defun, , Change a Function Definition}.)
2431 If you evaluate the first of these expressions, the initializer,
2432 @code{(setq counter 0)}, and then evaluate the third expression,
2433 @code{counter}, the number @code{0} will appear in the echo area. If
2434 you then evaluate the second expression, the incrementer, @code{(setq
2435 counter (+ counter 1))}, the counter will get the value 1. So if you
2436 again evaluate @code{counter}, the number @code{1} will appear in the
2437 echo area. Each time you evaluate the second expression, the value of
2438 the counter will be incremented.
2440 When you evaluate the incrementer, @code{(setq counter (+ counter 1))},
2441 the Lisp interpreter first evaluates the innermost list; this is the
2442 addition. In order to evaluate this list, it must evaluate the variable
2443 @code{counter} and the number @code{1}. When it evaluates the variable
2444 @code{counter}, it receives its current value. It passes this value and
2445 the number @code{1} to the @code{+} which adds them together. The sum
2446 is then returned as the value of the inner list and passed to the
2447 @code{setq} which sets the variable @code{counter} to this new value.
2448 Thus, the value of the variable, @code{counter}, is changed.
2453 Learning Lisp is like climbing a hill in which the first part is the
2454 steepest. You have now climbed the most difficult part; what remains
2455 becomes easier as you progress onwards.
2463 Lisp programs are made up of expressions, which are lists or single atoms.
2466 Lists are made up of zero or more atoms or inner lists, separated by whitespace and
2467 surrounded by parentheses. A list can be empty.
2470 Atoms are multi-character symbols, like @code{forward-paragraph}, single
2471 character symbols like @code{+}, strings of characters between double
2472 quotation marks, or numbers.
2475 A number evaluates to itself.
2478 A string between double quotes also evaluates to itself.
2481 When you evaluate a symbol by itself, its value is returned.
2484 When you evaluate a list, the Lisp interpreter looks at the first symbol
2485 in the list and then at the function definition bound to that symbol.
2486 Then the instructions in the function definition are carried out.
2489 A single quotation mark,
2496 , tells the Lisp interpreter that it should
2497 return the following expression as written, and not evaluate it as it
2498 would if the quote were not there.
2501 Arguments are the information passed to a function. The arguments to a
2502 function are computed by evaluating the rest of the elements of the list
2503 of which the function is the first element.
2506 A function always returns a value when it is evaluated (unless it gets
2507 an error); in addition, it may also carry out some action called a
2508 ``side effect''. In many cases, a function's primary purpose is to
2509 create a side effect.
2512 @node Error Message Exercises
2515 A few simple exercises:
2519 Generate an error message by evaluating an appropriate symbol that is
2520 not within parentheses.
2523 Generate an error message by evaluating an appropriate symbol that is
2524 between parentheses.
2527 Create a counter that increments by two rather than one.
2530 Write an expression that prints a message in the echo area when
2534 @node Practicing Evaluation
2535 @chapter Practicing Evaluation
2536 @cindex Practicing evaluation
2537 @cindex Evaluation practice
2539 Before learning how to write a function definition in Emacs Lisp, it is
2540 useful to spend a little time evaluating various expressions that have
2541 already been written. These expressions will be lists with the
2542 functions as their first (and often only) element. Since some of the
2543 functions associated with buffers are both simple and interesting, we
2544 will start with those. In this section, we will evaluate a few of
2545 these. In another section, we will study the code of several other
2546 buffer-related functions, to see how they were written.
2549 * How to Evaluate:: Typing editing commands or @kbd{C-x C-e}
2551 * Buffer Names:: Buffers and files are different.
2552 * Getting Buffers:: Getting a buffer itself, not merely its name.
2553 * Switching Buffers:: How to change to another buffer.
2554 * Buffer Size & Locations:: Where point is located and the size of
2556 * Evaluation Exercise::
2560 @node How to Evaluate
2561 @unnumberedsec How to Evaluate
2564 @i{Whenever you give an editing command} to Emacs Lisp, such as the
2565 command to move the cursor or to scroll the screen, @i{you are evaluating
2566 an expression,} the first element of which is a function. @i{This is
2569 @cindex @samp{interactive function} defined
2570 @cindex @samp{command} defined
2571 When you type keys, you cause the Lisp interpreter to evaluate an
2572 expression and that is how you get your results. Even typing plain text
2573 involves evaluating an Emacs Lisp function, in this case, one that uses
2574 @code{self-insert-command}, which simply inserts the character you
2575 typed. The functions you evaluate by typing keystrokes are called
2576 @dfn{interactive} functions, or @dfn{commands}; how you make a function
2577 interactive will be illustrated in the chapter on how to write function
2578 definitions. @xref{Interactive, , Making a Function Interactive}.
2580 In addition to typing keyboard commands, we have seen a second way to
2581 evaluate an expression: by positioning the cursor after a list and
2582 typing @kbd{C-x C-e}. This is what we will do in the rest of this
2583 section. There are other ways to evaluate an expression as well; these
2584 will be described as we come to them.
2586 Besides being used for practicing evaluation, the functions shown in the
2587 next few sections are important in their own right. A study of these
2588 functions makes clear the distinction between buffers and files, how to
2589 switch to a buffer, and how to determine a location within it.
2592 @section Buffer Names
2594 @findex buffer-file-name
2596 The two functions, @code{buffer-name} and @code{buffer-file-name}, show
2597 the difference between a file and a buffer. When you evaluate the
2598 following expression, @code{(buffer-name)}, the name of the buffer
2599 appears in the echo area. When you evaluate @code{(buffer-file-name)},
2600 the name of the file to which the buffer refers appears in the echo
2601 area. Usually, the name returned by @code{(buffer-name)} is the same as
2602 the name of the file to which it refers, and the name returned by
2603 @code{(buffer-file-name)} is the full path-name of the file.
2605 A file and a buffer are two different entities. A file is information
2606 recorded permanently in the computer (unless you delete it). A buffer,
2607 on the other hand, is information inside of Emacs that will vanish at
2608 the end of the editing session (or when you kill the buffer). Usually,
2609 a buffer contains information that you have copied from a file; we say
2610 the buffer is @dfn{visiting} that file. This copy is what you work on
2611 and modify. Changes to the buffer do not change the file, until you
2612 save the buffer. When you save the buffer, the buffer is copied to the file
2613 and is thus saved permanently.
2616 If you are reading this in Info inside of GNU Emacs, you can evaluate
2617 each of the following expressions by positioning the cursor after it and
2618 typing @kbd{C-x C-e}.
2629 When I do this in Info, the value returned by evaluating
2630 @code{(buffer-name)} is @file{"*info*"}, and the value returned by
2631 evaluating @code{(buffer-file-name)} is @file{nil}.
2633 On the other hand, while I am writing this document, the value
2634 returned by evaluating @code{(buffer-name)} is
2635 @file{"introduction.texinfo"}, and the value returned by evaluating
2636 @code{(buffer-file-name)} is
2637 @file{"/gnu/work/intro/introduction.texinfo"}.
2639 @cindex @code{nil}, history of word
2640 The former is the name of the buffer and the latter is the name of the
2641 file. In Info, the buffer name is @file{"*info*"}. Info does not
2642 point to any file, so the result of evaluating
2643 @code{(buffer-file-name)} is @file{nil}. The symbol @code{nil} is
2644 from the Latin word for ``nothing''; in this case, it means that the
2645 buffer is not associated with any file. (In Lisp, @code{nil} is also
2646 used to mean ``false'' and is a synonym for the empty list, @code{()}.)
2648 When I am writing, the name of my buffer is
2649 @file{"introduction.texinfo"}. The name of the file to which it
2650 points is @file{"/gnu/work/intro/introduction.texinfo"}.
2652 (In the expressions, the parentheses tell the Lisp interpreter to
2653 treat @w{@code{buffer-name}} and @w{@code{buffer-file-name}} as
2654 functions; without the parentheses, the interpreter would attempt to
2655 evaluate the symbols as variables. @xref{Variables}.)
2657 In spite of the distinction between files and buffers, you will often
2658 find that people refer to a file when they mean a buffer and vice versa.
2659 Indeed, most people say, ``I am editing a file,'' rather than saying,
2660 ``I am editing a buffer which I will soon save to a file.'' It is
2661 almost always clear from context what people mean. When dealing with
2662 computer programs, however, it is important to keep the distinction in mind,
2663 since the computer is not as smart as a person.
2665 @cindex Buffer, history of word
2666 The word ``buffer'', by the way, comes from the meaning of the word as a
2667 cushion that deadens the force of a collision. In early computers, a
2668 buffer cushioned the interaction between files and the computer's
2669 central processing unit. The drums or tapes that held a file and the
2670 central processing unit were pieces of equipment that were very
2671 different from each other, working at their own speeds, in spurts. The
2672 buffer made it possible for them to work together effectively.
2673 Eventually, the buffer grew from being an intermediary, a temporary
2674 holding place, to being the place where work is done. This
2675 transformation is rather like that of a small seaport that grew into a
2676 great city: once it was merely the place where cargo was warehoused
2677 temporarily before being loaded onto ships; then it became a business
2678 and cultural center in its own right.
2680 Not all buffers are associated with files. For example, a
2681 @file{*scratch*} buffer does not visit any file. Similarly, a
2682 @file{*Help*} buffer is not associated with any file.
2684 In the old days, when you lacked a @file{~/.emacs} file and started an
2685 Emacs session by typing the command @code{emacs} alone, without naming
2686 any files, Emacs started with the @file{*scratch*} buffer visible.
2687 Nowadays, you will see a splash screen. You can follow one of the
2688 commands suggested on the splash screen, visit a file, or press the
2689 spacebar to reach the @file{*scratch*} buffer.
2691 If you switch to the @file{*scratch*} buffer, type
2692 @code{(buffer-name)}, position the cursor after it, and then type
2693 @kbd{C-x C-e} to evaluate the expression. The name @code{"*scratch*"}
2694 will be returned and will appear in the echo area. @code{"*scratch*"}
2695 is the name of the buffer. When you type @code{(buffer-file-name)} in
2696 the @file{*scratch*} buffer and evaluate that, @code{nil} will appear
2697 in the echo area, just as it does when you evaluate
2698 @code{(buffer-file-name)} in Info.
2700 Incidentally, if you are in the @file{*scratch*} buffer and want the
2701 value returned by an expression to appear in the @file{*scratch*}
2702 buffer itself rather than in the echo area, type @kbd{C-u C-x C-e}
2703 instead of @kbd{C-x C-e}. This causes the value returned to appear
2704 after the expression. The buffer will look like this:
2707 (buffer-name)"*scratch*"
2711 You cannot do this in Info since Info is read-only and it will not allow
2712 you to change the contents of the buffer. But you can do this in any
2713 buffer you can edit; and when you write code or documentation (such as
2714 this book), this feature is very useful.
2716 @node Getting Buffers
2717 @section Getting Buffers
2718 @findex current-buffer
2719 @findex other-buffer
2720 @cindex Getting a buffer
2722 The @code{buffer-name} function returns the @emph{name} of the buffer;
2723 to get the buffer @emph{itself}, a different function is needed: the
2724 @code{current-buffer} function. If you use this function in code, what
2725 you get is the buffer itself.
2727 A name and the object or entity to which the name refers are different
2728 from each other. You are not your name. You are a person to whom
2729 others refer by name. If you ask to speak to George and someone hands you
2730 a card with the letters @samp{G}, @samp{e}, @samp{o}, @samp{r},
2731 @samp{g}, and @samp{e} written on it, you might be amused, but you would
2732 not be satisfied. You do not want to speak to the name, but to the
2733 person to whom the name refers. A buffer is similar: the name of the
2734 scratch buffer is @file{*scratch*}, but the name is not the buffer. To
2735 get a buffer itself, you need to use a function such as
2736 @code{current-buffer}.
2738 However, there is a slight complication: if you evaluate
2739 @code{current-buffer} in an expression on its own, as we will do here,
2740 what you see is a printed representation of the name of the buffer
2741 without the contents of the buffer. Emacs works this way for two
2742 reasons: the buffer may be thousands of lines long---too long to be
2743 conveniently displayed; and, another buffer may have the same contents
2744 but a different name, and it is important to distinguish between them.
2747 Here is an expression containing the function:
2754 If you evaluate this expression in Info in Emacs in the usual way,
2755 @file{#<buffer *info*>} will appear in the echo area. The special
2756 format indicates that the buffer itself is being returned, rather than
2759 Incidentally, while you can type a number or symbol into a program, you
2760 cannot do that with the printed representation of a buffer: the only way
2761 to get a buffer itself is with a function such as @code{current-buffer}.
2763 A related function is @code{other-buffer}. This returns the most
2764 recently selected buffer other than the one you are in currently, not
2765 a printed representation of its name. If you have recently switched
2766 back and forth from the @file{*scratch*} buffer, @code{other-buffer}
2767 will return that buffer.
2770 You can see this by evaluating the expression:
2777 You should see @file{#<buffer *scratch*>} appear in the echo area, or
2778 the name of whatever other buffer you switched back from most
2779 recently@footnote{Actually, by default, if the buffer from which you
2780 just switched is visible to you in another window, @code{other-buffer}
2781 will choose the most recent buffer that you cannot see; this is a
2782 subtlety that I often forget.}.
2784 @node Switching Buffers
2785 @section Switching Buffers
2786 @findex switch-to-buffer
2788 @cindex Switching to a buffer
2790 The @code{other-buffer} function actually provides a buffer when it is
2791 used as an argument to a function that requires one. We can see this
2792 by using @code{other-buffer} and @code{switch-to-buffer} to switch to a
2795 But first, a brief introduction to the @code{switch-to-buffer}
2796 function. When you switched back and forth from Info to the
2797 @file{*scratch*} buffer to evaluate @code{(buffer-name)}, you most
2798 likely typed @kbd{C-x b} and then typed @file{*scratch*}@footnote{Or
2799 rather, to save typing, you probably only typed @kbd{RET} if the
2800 default buffer was @file{*scratch*}, or if it was different, then you
2801 typed just part of the name, such as @code{*sc}, pressed your
2802 @kbd{TAB} key to cause it to expand to the full name, and then typed
2803 @kbd{RET}.} when prompted in the minibuffer for the name of
2804 the buffer to which you wanted to switch. The keystrokes, @kbd{C-x
2805 b}, cause the Lisp interpreter to evaluate the interactive function
2806 @code{switch-to-buffer}. As we said before, this is how Emacs works:
2807 different keystrokes call or run different functions. For example,
2808 @kbd{C-f} calls @code{forward-char}, @kbd{M-e} calls
2809 @code{forward-sentence}, and so on.
2811 By writing @code{switch-to-buffer} in an expression, and giving it a
2812 buffer to switch to, we can switch buffers just the way @kbd{C-x b}
2816 (switch-to-buffer (other-buffer))
2820 The symbol @code{switch-to-buffer} is the first element of the list,
2821 so the Lisp interpreter will treat it as a function and carry out the
2822 instructions that are attached to it. But before doing that, the
2823 interpreter will note that @code{other-buffer} is inside parentheses
2824 and work on that symbol first. @code{other-buffer} is the first (and
2825 in this case, the only) element of this list, so the Lisp interpreter
2826 calls or runs the function. It returns another buffer. Next, the
2827 interpreter runs @code{switch-to-buffer}, passing to it, as an
2828 argument, the other buffer, which is what Emacs will switch to. If
2829 you are reading this in Info, try this now. Evaluate the expression.
2830 (To get back, type @kbd{C-x b @key{RET}}.)@footnote{Remember, this
2831 expression will move you to your most recent other buffer that you
2832 cannot see. If you really want to go to your most recently selected
2833 buffer, even if you can still see it, you need to evaluate the
2834 following more complex expression:
2837 (switch-to-buffer (other-buffer (current-buffer) t))
2841 In this case, the first argument to @code{other-buffer} tells it which
2842 buffer to skip---the current one---and the second argument tells
2843 @code{other-buffer} it is OK to switch to a visible buffer. In
2844 regular use, @code{switch-to-buffer} takes you to a buffer not visible
2845 in windows since you would most likely use @kbd{C-x o}
2846 (@code{other-window}) to go to another visible buffer.}
2848 In the programming examples in later sections of this document, you will
2849 see the function @code{set-buffer} more often than
2850 @code{switch-to-buffer}. This is because of a difference between
2851 computer programs and humans: humans have eyes and expect to see the
2852 buffer on which they are working on their computer terminals. This is
2853 so obvious, it almost goes without saying. However, programs do not
2854 have eyes. When a computer program works on a buffer, that buffer does
2855 not need to be visible on the screen.
2857 @code{switch-to-buffer} is designed for humans and does two different
2858 things: it switches the buffer to which Emacs's attention is directed; and
2859 it switches the buffer displayed in the window to the new buffer.
2860 @code{set-buffer}, on the other hand, does only one thing: it switches
2861 the attention of the computer program to a different buffer. The buffer
2862 on the screen remains unchanged (of course, normally nothing happens
2863 there until the command finishes running).
2865 @cindex @samp{call} defined
2866 Also, we have just introduced another jargon term, the word @dfn{call}.
2867 When you evaluate a list in which the first symbol is a function, you
2868 are calling that function. The use of the term comes from the notion of
2869 the function as an entity that can do something for you if you ``call''
2870 it---just as a plumber is an entity who can fix a leak if you call him
2873 @node Buffer Size & Locations
2874 @section Buffer Size and the Location of Point
2875 @cindex Size of buffer
2877 @cindex Point location
2878 @cindex Location of point
2880 Finally, let's look at several rather simple functions,
2881 @code{buffer-size}, @code{point}, @code{point-min}, and
2882 @code{point-max}. These give information about the size of a buffer and
2883 the location of point within it.
2885 The function @code{buffer-size} tells you the size of the current
2886 buffer; that is, the function returns a count of the number of
2887 characters in the buffer.
2894 You can evaluate this in the usual way, by positioning the
2895 cursor after the expression and typing @kbd{C-x C-e}.
2897 @cindex @samp{point} defined
2898 In Emacs, the current position of the cursor is called @dfn{point}.
2899 The expression @code{(point)} returns a number that tells you where the
2900 cursor is located as a count of the number of characters from the
2901 beginning of the buffer up to point.
2904 You can see the character count for point in this buffer by evaluating
2905 the following expression in the usual way:
2912 As I write this, the value of point is 65724. The @code{point}
2913 function is frequently used in some of the examples later in this
2917 The value of point depends, of course, on its location within the
2918 buffer. If you evaluate point in this spot, the number will be larger:
2925 For me, the value of point in this location is 66043, which means that
2926 there are 319 characters (including spaces) between the two
2927 expressions. (Doubtless, you will see different numbers, since I will
2928 have edited this since I first evaluated point.)
2930 @cindex @samp{narrowing} defined
2931 The function @code{point-min} is somewhat similar to @code{point}, but
2932 it returns the value of the minimum permissible value of point in the
2933 current buffer. This is the number 1 unless @dfn{narrowing} is in
2934 effect. (Narrowing is a mechanism whereby you can restrict yourself,
2935 or a program, to operations on just a part of a buffer.
2936 @xref{Narrowing & Widening, , Narrowing and Widening}.) Likewise, the
2937 function @code{point-max} returns the value of the maximum permissible
2938 value of point in the current buffer.
2940 @node Evaluation Exercise
2943 Find a file with which you are working and move towards its middle.
2944 Find its buffer name, file name, length, and your position in the file.
2946 @node Writing Defuns
2947 @chapter How To Write Function Definitions
2948 @cindex Definition writing
2949 @cindex Function definition writing
2950 @cindex Writing a function definition
2952 When the Lisp interpreter evaluates a list, it looks to see whether the
2953 first symbol on the list has a function definition attached to it; or,
2954 put another way, whether the symbol points to a function definition. If
2955 it does, the computer carries out the instructions in the definition. A
2956 symbol that has a function definition is called, simply, a function
2957 (although, properly speaking, the definition is the function and the
2958 symbol refers to it.)
2961 * Primitive Functions::
2962 * defun:: The @code{defun} macro.
2963 * Install:: Install a function definition.
2964 * Interactive:: Making a function interactive.
2965 * Interactive Options:: Different options for @code{interactive}.
2966 * Permanent Installation:: Installing code permanently.
2967 * let:: Creating and initializing local variables.
2969 * else:: If--then--else expressions.
2970 * Truth & Falsehood:: What Lisp considers false and true.
2971 * save-excursion:: Keeping track of point and buffer.
2977 @node Primitive Functions
2978 @unnumberedsec An Aside about Primitive Functions
2980 @cindex Primitive functions
2981 @cindex Functions, primitive
2983 @cindex C language primitives
2984 @cindex Primitives written in C
2985 All functions are defined in terms of other functions, except for a few
2986 @dfn{primitive} functions that are written in the C programming
2987 language. When you write functions' definitions, you will write them in
2988 Emacs Lisp and use other functions as your building blocks. Some of the
2989 functions you will use will themselves be written in Emacs Lisp (perhaps
2990 by you) and some will be primitives written in C@. The primitive
2991 functions are used exactly like those written in Emacs Lisp and behave
2992 like them. They are written in C so we can easily run GNU Emacs on any
2993 computer that has sufficient power and can run C.
2995 Let me re-emphasize this: when you write code in Emacs Lisp, you do not
2996 distinguish between the use of functions written in C and the use of
2997 functions written in Emacs Lisp. The difference is irrelevant. I
2998 mention the distinction only because it is interesting to know. Indeed,
2999 unless you investigate, you won't know whether an already-written
3000 function is written in Emacs Lisp or C.
3003 @section The @code{defun} Macro
3006 @cindex @samp{function definition} defined
3007 In Lisp, a symbol such as @code{mark-whole-buffer} has code attached to
3008 it that tells the computer what to do when the function is called.
3009 This code is called the @dfn{function definition} and is created by
3010 evaluating a Lisp expression that starts with the symbol @code{defun}
3011 (which is an abbreviation for @emph{define function}).
3013 In subsequent sections, we will look at function definitions from the
3014 Emacs source code, such as @code{mark-whole-buffer}. In this section,
3015 we will describe a simple function definition so you can see how it
3016 looks. This function definition uses arithmetic because it makes for a
3017 simple example. Some people dislike examples using arithmetic; however,
3018 if you are such a person, do not despair. Hardly any of the code we
3019 will study in the remainder of this introduction involves arithmetic or
3020 mathematics. The examples mostly involve text in one way or another.
3022 A function definition has up to five parts following the word
3027 The name of the symbol to which the function definition should be
3031 A list of the arguments that will be passed to the function. If no
3032 arguments will be passed to the function, this is an empty list,
3036 Documentation describing the function. (Technically optional, but
3037 strongly recommended.)
3040 Optionally, an expression to make the function interactive so you can
3041 use it by typing @kbd{M-x} and then the name of the function; or by
3042 typing an appropriate key or keychord.
3044 @cindex @samp{body} defined
3046 The code that instructs the computer what to do: the @dfn{body} of the
3047 function definition.
3050 It is helpful to think of the five parts of a function definition as
3051 being organized in a template, with slots for each part:
3055 (defun @var{function-name} (@var{arguments}@dots{})
3056 "@var{optional-documentation}@dots{}"
3057 (interactive @var{argument-passing-info}) ; @r{optional}
3062 As an example, here is the code for a function that multiplies its
3063 argument by 7. (This example is not interactive. @xref{Interactive,
3064 , Making a Function Interactive}, for that information.)
3068 (defun multiply-by-seven (number)
3069 "Multiply NUMBER by seven."
3074 This definition begins with a parenthesis and the symbol @code{defun},
3075 followed by the name of the function.
3077 @cindex @samp{argument list} defined
3078 The name of the function is followed by a list that contains the
3079 arguments that will be passed to the function. This list is called
3080 the @dfn{argument list}. In this example, the list has only one
3081 element, the symbol, @code{number}. When the function is used, the
3082 symbol will be bound to the value that is used as the argument to the
3085 Instead of choosing the word @code{number} for the name of the argument,
3086 I could have picked any other name. For example, I could have chosen
3087 the word @code{multiplicand}. I picked the word ``number'' because it
3088 tells what kind of value is intended for this slot; but I could just as
3089 well have chosen the word ``multiplicand'' to indicate the role that the
3090 value placed in this slot will play in the workings of the function. I
3091 could have called it @code{foogle}, but that would have been a bad
3092 choice because it would not tell humans what it means. The choice of
3093 name is up to the programmer and should be chosen to make the meaning of
3096 Indeed, you can choose any name you wish for a symbol in an argument
3097 list, even the name of a symbol used in some other function: the name
3098 you use in an argument list is private to that particular definition.
3099 In that definition, the name refers to a different entity than any use
3100 of the same name outside the function definition. Suppose you have a
3101 nick-name ``Shorty'' in your family; when your family members refer to
3102 ``Shorty'', they mean you. But outside your family, in a movie, for
3103 example, the name ``Shorty'' refers to someone else. Because a name in an
3104 argument list is private to the function definition, you can change the
3105 value of such a symbol inside the body of a function without changing
3106 its value outside the function. The effect is similar to that produced
3107 by a @code{let} expression. (@xref{let, , @code{let}}.)
3110 Note also that we discuss the word ``number'' in two different ways: as a
3111 symbol that appears in the code, and as the name of something that will
3112 be replaced by a something else during the evaluation of the function.
3113 In the first case, @code{number} is a symbol, not a number; it happens
3114 that within the function, it is a variable who value is the number in
3115 question, but our primary interest in it is as a symbol. On the other
3116 hand, when we are talking about the function, our interest is that we
3117 will substitute a number for the word @var{number}. To keep this
3118 distinction clear, we use different typography for the two
3119 circumstances. When we talk about this function, or about how it works,
3120 we refer to this number by writing @var{number}. In the function
3121 itself, we refer to it by writing @code{number}.
3124 The argument list is followed by the documentation string that
3125 describes the function. This is what you see when you type
3126 @w{@kbd{C-h f}} and the name of a function. Incidentally, when you
3127 write a documentation string like this, you should make the first line
3128 a complete sentence since some commands, such as @code{apropos}, print
3129 only the first line of a multi-line documentation string. Also, you
3130 should not indent the second line of a documentation string, if you
3131 have one, because that looks odd when you use @kbd{C-h f}
3132 (@code{describe-function}). The documentation string is optional, but
3133 it is so useful, it should be included in almost every function you
3136 @findex * @r{(multiplication)}
3137 The third line of the example consists of the body of the function
3138 definition. (Most functions' definitions, of course, are longer than
3139 this.) In this function, the body is the list, @code{(* 7 number)}, which
3140 says to multiply the value of @var{number} by 7. (In Emacs Lisp,
3141 @code{*} is the function for multiplication, just as @code{+} is the
3142 function for addition.)
3144 When you use the @code{multiply-by-seven} function, the argument
3145 @code{number} evaluates to the actual number you want used. Here is an
3146 example that shows how @code{multiply-by-seven} is used; but don't try
3147 to evaluate this yet!
3150 (multiply-by-seven 3)
3154 The symbol @code{number}, specified in the function definition in the
3155 next section, is given or ``bound to'' the value 3 in the actual use of
3156 the function. Note that although @code{number} was inside parentheses
3157 in the function definition, the argument passed to the
3158 @code{multiply-by-seven} function is not in parentheses. The
3159 parentheses are written in the function definition so the computer can
3160 figure out where the argument list ends and the rest of the function
3163 If you evaluate this example, you are likely to get an error message.
3164 (Go ahead, try it!) This is because we have written the function
3165 definition, but not yet told the computer about the definition---we have
3166 not yet installed (or ``loaded'') the function definition in Emacs.
3167 Installing a function is the process that tells the Lisp interpreter the
3168 definition of the function. Installation is described in the next
3172 @section Install a Function Definition
3173 @cindex Install a Function Definition
3174 @cindex Definition installation
3175 @cindex Function definition installation
3177 If you are reading this inside of Info in Emacs, you can try out the
3178 @code{multiply-by-seven} function by first evaluating the function
3179 definition and then evaluating @code{(multiply-by-seven 3)}. A copy of
3180 the function definition follows. Place the cursor after the last
3181 parenthesis of the function definition and type @kbd{C-x C-e}. When you
3182 do this, @code{multiply-by-seven} will appear in the echo area. (What
3183 this means is that when a function definition is evaluated, the value it
3184 returns is the name of the defined function.) At the same time, this
3185 action installs the function definition.
3189 (defun multiply-by-seven (number)
3190 "Multiply NUMBER by seven."
3196 By evaluating this @code{defun}, you have just installed
3197 @code{multiply-by-seven} in Emacs. The function is now just as much a
3198 part of Emacs as @code{forward-word} or any other editing function you
3199 use. (@code{multiply-by-seven} will stay installed until you quit
3200 Emacs. To reload code automatically whenever you start Emacs, see
3201 @ref{Permanent Installation, , Installing Code Permanently}.)
3204 * Effect of installation::
3205 * Change a defun:: How to change a function definition.
3209 @node Effect of installation
3210 @unnumberedsubsec The effect of installation
3213 You can see the effect of installing @code{multiply-by-seven} by
3214 evaluating the following sample. Place the cursor after the following
3215 expression and type @kbd{C-x C-e}. The number 21 will appear in the
3219 (multiply-by-seven 3)
3222 If you wish, you can read the documentation for the function by typing
3223 @kbd{C-h f} (@code{describe-function}) and then the name of the
3224 function, @code{multiply-by-seven}. When you do this, a
3225 @file{*Help*} window will appear on your screen that says:
3229 multiply-by-seven is a Lisp function.
3231 (multiply-by-seven NUMBER)
3233 Multiply NUMBER by seven.
3238 (To return to a single window on your screen, type @kbd{C-x 1}.)
3240 @node Change a defun
3241 @subsection Change a Function Definition
3242 @cindex Changing a function definition
3243 @cindex Function definition, how to change
3244 @cindex Definition, how to change
3246 If you want to change the code in @code{multiply-by-seven}, just rewrite
3247 it. To install the new version in place of the old one, evaluate the
3248 function definition again. This is how you modify code in Emacs. It is
3251 As an example, you can change the @code{multiply-by-seven} function to
3252 add the number to itself seven times instead of multiplying the number
3253 by seven. It produces the same answer, but by a different path. At
3254 the same time, we will add a comment to the code; a comment is text
3255 that the Lisp interpreter ignores, but that a human reader may find
3256 useful or enlightening. The comment is that this is the ``second
3261 (defun multiply-by-seven (number) ; @r{Second version.}
3262 "Multiply NUMBER by seven."
3263 (+ number number number number number number number))
3267 @cindex Comments in Lisp code
3268 The comment follows a semicolon, @samp{;}. In Lisp, everything on a
3269 line that follows a semicolon is a comment. The end of the line is the
3270 end of the comment. To stretch a comment over two or more lines, begin
3271 each line with a semicolon.
3273 @xref{Beginning init File, , Beginning a @file{.emacs}
3274 File}, and @ref{Comments, , Comments, elisp, The GNU Emacs Lisp
3275 Reference Manual}, for more about comments.
3277 You can install this version of the @code{multiply-by-seven} function by
3278 evaluating it in the same way you evaluated the first function: place
3279 the cursor after the last parenthesis and type @kbd{C-x C-e}.
3281 In summary, this is how you write code in Emacs Lisp: you write a
3282 function; install it; test it; and then make fixes or enhancements and
3286 @section Make a Function Interactive
3287 @cindex Interactive functions
3290 You make a function interactive by placing a list that begins with
3291 the special form @code{interactive} immediately after the
3292 documentation. A user can invoke an interactive function by typing
3293 @kbd{M-x} and then the name of the function; or by typing the keys to
3294 which it is bound, for example, by typing @kbd{C-n} for
3295 @code{next-line} or @kbd{C-x h} for @code{mark-whole-buffer}.
3297 Interestingly, when you call an interactive function interactively,
3298 the value returned is not automatically displayed in the echo area.
3299 This is because you often call an interactive function for its side
3300 effects, such as moving forward by a word or line, and not for the
3301 value returned. If the returned value were displayed in the echo area
3302 each time you typed a key, it would be very distracting.
3305 * Interactive multiply-by-seven:: An overview.
3306 * multiply-by-seven in detail:: The interactive version.
3310 @node Interactive multiply-by-seven
3311 @unnumberedsubsec An Interactive @code{multiply-by-seven}, An Overview
3314 Both the use of the special form @code{interactive} and one way to
3315 display a value in the echo area can be illustrated by creating an
3316 interactive version of @code{multiply-by-seven}.
3323 (defun multiply-by-seven (number) ; @r{Interactive version.}
3324 "Multiply NUMBER by seven."
3326 (message "The result is %d" (* 7 number)))
3331 You can install this code by placing your cursor after it and typing
3332 @kbd{C-x C-e}. The name of the function will appear in your echo area.
3333 Then, you can use this code by typing @kbd{C-u} and a number and then
3334 typing @kbd{M-x multiply-by-seven} and pressing @key{RET}. The phrase
3335 @samp{The result is @dots{}} followed by the product will appear in the
3338 Speaking more generally, you invoke a function like this in either of two
3343 By typing a prefix argument that contains the number to be passed, and
3344 then typing @kbd{M-x} and the name of the function, as with
3345 @kbd{C-u 3 M-x forward-sentence}; or,
3348 By typing whatever key or keychord the function is bound to, as with
3353 Both the examples just mentioned work identically to move point forward
3354 three sentences. (Since @code{multiply-by-seven} is not bound to a key,
3355 it could not be used as an example of key binding.)
3357 (@xref{Keybindings, , Some Keybindings}, to learn how to bind a command
3360 A prefix argument is passed to an interactive function by typing the
3361 @key{META} key followed by a number, for example, @kbd{M-3 M-e}, or by
3362 typing @kbd{C-u} and then a number, for example, @kbd{C-u 3 M-e} (if you
3363 type @kbd{C-u} without a number, it defaults to 4).
3365 @node multiply-by-seven in detail
3366 @subsection An Interactive @code{multiply-by-seven}
3368 Let's look at the use of the special form @code{interactive} and then at
3369 the function @code{message} in the interactive version of
3370 @code{multiply-by-seven}. You will recall that the function definition
3375 (defun multiply-by-seven (number) ; @r{Interactive version.}
3376 "Multiply NUMBER by seven."
3378 (message "The result is %d" (* 7 number)))
3382 In this function, the expression, @code{(interactive "p")}, is a list of
3383 two elements. The @code{"p"} tells Emacs to pass the prefix argument to
3384 the function and use its value for the argument of the function.
3387 The argument will be a number. This means that the symbol
3388 @code{number} will be bound to a number in the line:
3391 (message "The result is %d" (* 7 number))
3396 For example, if your prefix argument is 5, the Lisp interpreter will
3397 evaluate the line as if it were:
3400 (message "The result is %d" (* 7 5))
3404 (If you are reading this in GNU Emacs, you can evaluate this expression
3405 yourself.) First, the interpreter will evaluate the inner list, which
3406 is @code{(* 7 5)}. This returns a value of 35. Next, it
3407 will evaluate the outer list, passing the values of the second and
3408 subsequent elements of the list to the function @code{message}.
3410 As we have seen, @code{message} is an Emacs Lisp function especially
3411 designed for sending a one line message to a user. (@xref{message, ,
3412 The @code{message} function}.) In summary, the @code{message}
3413 function prints its first argument in the echo area as is, except for
3414 occurrences of @samp{%d} or @samp{%s} (and various other %-sequences
3415 which we have not mentioned). When it sees a control sequence, the
3416 function looks to the second or subsequent arguments and prints the
3417 value of the argument in the location in the string where the control
3418 sequence is located.
3420 In the interactive @code{multiply-by-seven} function, the control string
3421 is @samp{%d}, which requires a number, and the value returned by
3422 evaluating @code{(* 7 5)} is the number 35. Consequently, the number 35
3423 is printed in place of the @samp{%d} and the message is @samp{The result
3426 (Note that when you call the function @code{multiply-by-seven}, the
3427 message is printed without quotes, but when you call @code{message}, the
3428 text is printed in double quotes. This is because the value returned by
3429 @code{message} is what appears in the echo area when you evaluate an
3430 expression whose first element is @code{message}; but when embedded in a
3431 function, @code{message} prints the text as a side effect without
3434 @node Interactive Options
3435 @section Different Options for @code{interactive}
3436 @cindex Options for @code{interactive}
3437 @cindex Interactive options
3439 In the example, @code{multiply-by-seven} used @code{"p"} as the
3440 argument to @code{interactive}. This argument told Emacs to interpret
3441 your typing either @kbd{C-u} followed by a number or @key{META}
3442 followed by a number as a command to pass that number to the function
3443 as its argument. Emacs has more than twenty characters predefined for
3444 use with @code{interactive}. In almost every case, one of these
3445 options will enable you to pass the right information interactively to
3446 a function. (@xref{Interactive Codes, , Code Characters for
3447 @code{interactive}, elisp, The GNU Emacs Lisp Reference Manual}.)
3450 Consider the function @code{zap-to-char}. Its interactive expression
3454 (interactive "p\ncZap to char: ")
3457 The first part of the argument to @code{interactive} is @samp{p}, with
3458 which you are already familiar. This argument tells Emacs to
3459 interpret a ``prefix'', as a number to be passed to the function. You
3460 can specify a prefix either by typing @kbd{C-u} followed by a number
3461 or by typing @key{META} followed by a number. The prefix is the
3462 number of specified characters. Thus, if your prefix is three and the
3463 specified character is @samp{x}, then you will delete all the text up
3464 to and including the third next @samp{x}. If you do not set a prefix,
3465 then you delete all the text up to and including the specified
3466 character, but no more.
3468 The @samp{c} tells the function the name of the character to which to delete.
3470 More formally, a function with two or more arguments can have
3471 information passed to each argument by adding parts to the string that
3472 follows @code{interactive}. When you do this, the information is
3473 passed to each argument in the same order it is specified in the
3474 @code{interactive} list. In the string, each part is separated from
3475 the next part by a @samp{\n}, which is a newline. For example, you
3476 can follow @samp{p} with a @samp{\n} and an @samp{cZap to char:@: }.
3477 This causes Emacs to pass the value of the prefix argument (if there
3478 is one) and the character.
3480 In this case, the function definition looks like the following, where
3481 @code{arg} and @code{char} are the symbols to which @code{interactive}
3482 binds the prefix argument and the specified character:
3486 (defun @var{name-of-function} (arg char)
3487 "@var{documentation}@dots{}"
3488 (interactive "p\ncZap to char: ")
3489 @var{body-of-function}@dots{})
3494 (The space after the colon in the prompt makes it look better when you
3495 are prompted. @xref{copy-to-buffer, , The Definition of
3496 @code{copy-to-buffer}}, for an example.)
3498 When a function does not take arguments, @code{interactive} does not
3499 require any. Such a function contains the simple expression
3500 @code{(interactive)}. The @code{mark-whole-buffer} function is like
3503 Alternatively, if the special letter-codes are not right for your
3504 application, you can pass your own arguments to @code{interactive} as
3507 @xref{append-to-buffer, , The Definition of @code{append-to-buffer}},
3508 for an example. @xref{Using Interactive, , Using @code{Interactive},
3509 elisp, The GNU Emacs Lisp Reference Manual}, for a more complete
3510 explanation about this technique.
3512 @node Permanent Installation
3513 @section Install Code Permanently
3514 @cindex Install code permanently
3515 @cindex Permanent code installation
3516 @cindex Code installation
3518 When you install a function definition by evaluating it, it will stay
3519 installed until you quit Emacs. The next time you start a new session
3520 of Emacs, the function will not be installed unless you evaluate the
3521 function definition again.
3523 At some point, you may want to have code installed automatically
3524 whenever you start a new session of Emacs. There are several ways of
3529 If you have code that is just for yourself, you can put the code for the
3530 function definition in your @file{.emacs} initialization file. When you
3531 start Emacs, your @file{.emacs} file is automatically evaluated and all
3532 the function definitions within it are installed.
3533 @xref{Emacs Initialization, , Your @file{.emacs} File}.
3536 Alternatively, you can put the function definitions that you want
3537 installed in one or more files of their own and use the @code{load}
3538 function to cause Emacs to evaluate and thereby install each of the
3539 functions in the files.
3540 @xref{Loading Files, , Loading Files}.
3543 Thirdly, if you have code that your whole site will use, it is usual
3544 to put it in a file called @file{site-init.el} that is loaded when
3545 Emacs is built. This makes the code available to everyone who uses
3546 your machine. (See the @file{INSTALL} file that is part of the Emacs
3550 Finally, if you have code that everyone who uses Emacs may want, you
3551 can post it on a computer network or send a copy to the Free Software
3552 Foundation. (When you do this, please license the code and its
3553 documentation under a license that permits other people to run, copy,
3554 study, modify, and redistribute the code and which protects you from
3555 having your work taken from you.) If you send a copy of your code to
3556 the Free Software Foundation, and properly protect yourself and
3557 others, it may be included in the next release of Emacs. In large
3558 part, this is how Emacs has grown over the past years, by donations.
3564 The @code{let} expression is a special form in Lisp that you will need
3565 to use in most function definitions.
3567 @code{let} is used to attach or bind a symbol to a value in such a way
3568 that the Lisp interpreter will not confuse the variable with a
3569 variable of the same name that is not part of the function.
3571 To understand why the @code{let} special form is necessary, consider
3572 the situation in which you own a home that you generally refer to as
3573 ``the house'', as in the sentence, ``The house needs painting.'' If you
3574 are visiting a friend and your host refers to ``the house'', he is
3575 likely to be referring to @emph{his} house, not yours, that is, to a
3578 If your friend is referring to his house and you think he is referring
3579 to your house, you may be in for some confusion. The same thing could
3580 happen in Lisp if a variable that is used inside of one function has
3581 the same name as a variable that is used inside of another function,
3582 and the two are not intended to refer to the same value. The
3583 @code{let} special form prevents this kind of confusion.
3586 * Prevent confusion::
3587 * Parts of let Expression::
3588 * Sample let Expression::
3589 * Uninitialized let Variables::
3593 @node Prevent confusion
3594 @unnumberedsubsec @code{let} Prevents Confusion
3597 @cindex @samp{local variable} defined
3598 @cindex @samp{variable, local}, defined
3599 The @code{let} special form prevents confusion. @code{let} creates a
3600 name for a @dfn{local variable} that overshadows any use of the same
3601 name outside the @code{let} expression. This is like understanding
3602 that whenever your host refers to ``the house'', he means his house, not
3603 yours. (Symbols used in argument lists work the same way.
3604 @xref{defun, , The @code{defun} Macro}.)
3606 Local variables created by a @code{let} expression retain their value
3607 @emph{only} within the @code{let} expression itself (and within
3608 expressions called within the @code{let} expression); the local
3609 variables have no effect outside the @code{let} expression.
3611 Another way to think about @code{let} is that it is like a @code{setq}
3612 that is temporary and local. The values set by @code{let} are
3613 automatically undone when the @code{let} is finished. The setting
3614 only affects expressions that are inside the bounds of the @code{let}
3615 expression. In computer science jargon, we would say ``the binding of
3616 a symbol is visible only in functions called in the @code{let} form;
3617 in Emacs Lisp, scoping is dynamic, not lexical.''
3619 @code{let} can create more than one variable at once. Also,
3620 @code{let} gives each variable it creates an initial value, either a
3621 value specified by you, or @code{nil}. (In the jargon, this is called
3622 ``binding the variable to the value''.) After @code{let} has created
3623 and bound the variables, it executes the code in the body of the
3624 @code{let}, and returns the value of the last expression in the body,
3625 as the value of the whole @code{let} expression. (``Execute'' is a jargon
3626 term that means to evaluate a list; it comes from the use of the word
3627 meaning ``to give practical effect to'' (@cite{Oxford English
3628 Dictionary}). Since you evaluate an expression to perform an action,
3629 ``execute'' has evolved as a synonym to ``evaluate''.)
3631 @node Parts of let Expression
3632 @subsection The Parts of a @code{let} Expression
3633 @cindex @code{let} expression, parts of
3634 @cindex Parts of @code{let} expression
3636 @cindex @samp{varlist} defined
3637 A @code{let} expression is a list of three parts. The first part is
3638 the symbol @code{let}. The second part is a list, called a
3639 @dfn{varlist}, each element of which is either a symbol by itself or a
3640 two-element list, the first element of which is a symbol. The third
3641 part of the @code{let} expression is the body of the @code{let}. The
3642 body usually consists of one or more lists.
3645 A template for a @code{let} expression looks like this:
3648 (let @var{varlist} @var{body}@dots{})
3652 The symbols in the varlist are the variables that are given initial
3653 values by the @code{let} special form. Symbols by themselves are given
3654 the initial value of @code{nil}; and each symbol that is the first
3655 element of a two-element list is bound to the value that is returned
3656 when the Lisp interpreter evaluates the second element.
3658 Thus, a varlist might look like this: @code{(thread (needles 3))}. In
3659 this case, in a @code{let} expression, Emacs binds the symbol
3660 @code{thread} to an initial value of @code{nil}, and binds the symbol
3661 @code{needles} to an initial value of 3.
3663 When you write a @code{let} expression, what you do is put the
3664 appropriate expressions in the slots of the @code{let} expression
3667 If the varlist is composed of two-element lists, as is often the case,
3668 the template for the @code{let} expression looks like this:
3672 (let ((@var{variable} @var{value})
3673 (@var{variable} @var{value})
3679 @node Sample let Expression
3680 @subsection Sample @code{let} Expression
3681 @cindex Sample @code{let} expression
3682 @cindex @code{let} expression sample
3684 The following expression creates and gives initial values
3685 to the two variables @code{zebra} and @code{tiger}. The body of the
3686 @code{let} expression is a list which calls the @code{message} function.
3690 (let ((zebra 'stripes)
3692 (message "One kind of animal has %s and another is %s."
3697 Here, the varlist is @code{((zebra 'stripes) (tiger 'fierce))}.
3699 The two variables are @code{zebra} and @code{tiger}. Each variable is
3700 the first element of a two-element list and each value is the second
3701 element of its two-element list. In the varlist, Emacs binds the
3702 variable @code{zebra} to the value @code{stripes}@footnote{According
3703 to Jared Diamond in @cite{Guns, Germs, and Steel}, ``@dots{} zebras
3704 become impossibly dangerous as they grow older'' but the claim here is
3705 that they do not become fierce like a tiger. (1997, W. W. Norton and
3706 Co., ISBN 0-393-03894-2, page 171)}, and binds the
3707 variable @code{tiger} to the value @code{fierce}. In this example,
3708 both values are symbols preceded by a quote. The values could just as
3709 well have been another list or a string. The body of the @code{let}
3710 follows after the list holding the variables. In this example, the
3711 body is a list that uses the @code{message} function to print a string
3715 You may evaluate the example in the usual fashion, by placing the
3716 cursor after the last parenthesis and typing @kbd{C-x C-e}. When you do
3717 this, the following will appear in the echo area:
3720 "One kind of animal has stripes and another is fierce."
3723 As we have seen before, the @code{message} function prints its first
3724 argument, except for @samp{%s}. In this example, the value of the variable
3725 @code{zebra} is printed at the location of the first @samp{%s} and the
3726 value of the variable @code{tiger} is printed at the location of the
3729 @node Uninitialized let Variables
3730 @subsection Uninitialized Variables in a @code{let} Statement
3731 @cindex Uninitialized @code{let} variables
3732 @cindex @code{let} variables uninitialized
3734 If you do not bind the variables in a @code{let} statement to specific
3735 initial values, they will automatically be bound to an initial value of
3736 @code{nil}, as in the following expression:
3745 "Here are %d variables with %s, %s, and %s value."
3746 birch pine fir oak))
3751 Here, the varlist is @code{((birch 3) pine fir (oak 'some))}.
3754 If you evaluate this expression in the usual way, the following will
3755 appear in your echo area:
3758 "Here are 3 variables with nil, nil, and some value."
3762 In this example, Emacs binds the symbol @code{birch} to the number 3,
3763 binds the symbols @code{pine} and @code{fir} to @code{nil}, and binds
3764 the symbol @code{oak} to the value @code{some}.
3766 Note that in the first part of the @code{let}, the variables @code{pine}
3767 and @code{fir} stand alone as atoms that are not surrounded by
3768 parentheses; this is because they are being bound to @code{nil}, the
3769 empty list. But @code{oak} is bound to @code{some} and so is a part of
3770 the list @code{(oak 'some)}. Similarly, @code{birch} is bound to the
3771 number 3 and so is in a list with that number. (Since a number
3772 evaluates to itself, the number does not need to be quoted. Also, the
3773 number is printed in the message using a @samp{%d} rather than a
3774 @samp{%s}.) The four variables as a group are put into a list to
3775 delimit them from the body of the @code{let}.
3778 @section The @code{if} Special Form
3780 @cindex Conditional with @code{if}
3782 A third special form, in addition to @code{defun} and @code{let}, is the
3783 conditional @code{if}. This form is used to instruct the computer to
3784 make decisions. You can write function definitions without using
3785 @code{if}, but it is used often enough, and is important enough, to be
3786 included here. It is used, for example, in the code for the
3787 function @code{beginning-of-buffer}.
3789 The basic idea behind an @code{if}, is that ``@emph{if} a test is true,
3790 @emph{then} an expression is evaluated.'' If the test is not true, the
3791 expression is not evaluated. For example, you might make a decision
3792 such as, ``if it is warm and sunny, then go to the beach!''
3795 * if in more detail::
3796 * type-of-animal in detail:: An example of an @code{if} expression.
3800 @node if in more detail
3801 @unnumberedsubsec @code{if} in more detail
3804 @cindex @samp{if-part} defined
3805 @cindex @samp{then-part} defined
3806 An @code{if} expression written in Lisp does not use the word ``then'';
3807 the test and the action are the second and third elements of the list
3808 whose first element is @code{if}. Nonetheless, the test part of an
3809 @code{if} expression is often called the @dfn{if-part} and the second
3810 argument is often called the @dfn{then-part}.
3812 Also, when an @code{if} expression is written, the true-or-false-test
3813 is usually written on the same line as the symbol @code{if}, but the
3814 action to carry out if the test is true, the ``then-part'', is written
3815 on the second and subsequent lines. This makes the @code{if}
3816 expression easier to read.
3820 (if @var{true-or-false-test}
3821 @var{action-to-carry-out-if-test-is-true})
3826 The true-or-false-test will be an expression that
3827 is evaluated by the Lisp interpreter.
3829 Here is an example that you can evaluate in the usual manner. The test
3830 is whether the number 5 is greater than the number 4. Since it is, the
3831 message @samp{5 is greater than 4!} will be printed.
3835 (if (> 5 4) ; @r{if-part}
3836 (message "5 is greater than 4!")) ; @r{then-part}
3841 (The function @code{>} tests whether its first argument is greater than
3842 its second argument and returns true if it is.)
3843 @findex > (greater than)
3845 Of course, in actual use, the test in an @code{if} expression will not
3846 be fixed for all time as it is by the expression @code{(> 5 4)}.
3847 Instead, at least one of the variables used in the test will be bound to
3848 a value that is not known ahead of time. (If the value were known ahead
3849 of time, we would not need to run the test!)
3851 For example, the value may be bound to an argument of a function
3852 definition. In the following function definition, the character of the
3853 animal is a value that is passed to the function. If the value bound to
3854 @code{characteristic} is @code{fierce}, then the message, @samp{It's a
3855 tiger!} will be printed; otherwise, @code{nil} will be returned.
3859 (defun type-of-animal (characteristic)
3860 "Print message in echo area depending on CHARACTERISTIC.
3861 If the CHARACTERISTIC is the symbol `fierce',
3862 then warn of a tiger."
3863 (if (equal characteristic 'fierce)
3864 (message "It's a tiger!")))
3870 If you are reading this inside of GNU Emacs, you can evaluate the
3871 function definition in the usual way to install it in Emacs, and then you
3872 can evaluate the following two expressions to see the results:
3876 (type-of-animal 'fierce)
3878 (type-of-animal 'zebra)
3883 @c Following sentences rewritten to prevent overfull hbox.
3885 When you evaluate @code{(type-of-animal 'fierce)}, you will see the
3886 following message printed in the echo area: @code{"It's a tiger!"}; and
3887 when you evaluate @code{(type-of-animal 'zebra)} you will see @code{nil}
3888 printed in the echo area.
3890 @node type-of-animal in detail
3891 @subsection The @code{type-of-animal} Function in Detail
3893 Let's look at the @code{type-of-animal} function in detail.
3895 The function definition for @code{type-of-animal} was written by filling
3896 the slots of two templates, one for a function definition as a whole, and
3897 a second for an @code{if} expression.
3900 The template for every function that is not interactive is:
3904 (defun @var{name-of-function} (@var{argument-list})
3905 "@var{documentation}@dots{}"
3911 The parts of the function that match this template look like this:
3915 (defun type-of-animal (characteristic)
3916 "Print message in echo area depending on CHARACTERISTIC.
3917 If the CHARACTERISTIC is the symbol `fierce',
3918 then warn of a tiger."
3919 @var{body: the} @code{if} @var{expression})
3923 The name of function is @code{type-of-animal}; it is passed the value
3924 of one argument. The argument list is followed by a multi-line
3925 documentation string. The documentation string is included in the
3926 example because it is a good habit to write documentation string for
3927 every function definition. The body of the function definition
3928 consists of the @code{if} expression.
3931 The template for an @code{if} expression looks like this:
3935 (if @var{true-or-false-test}
3936 @var{action-to-carry-out-if-the-test-returns-true})
3941 In the @code{type-of-animal} function, the code for the @code{if}
3946 (if (equal characteristic 'fierce)
3947 (message "It's a tiger!")))
3952 Here, the true-or-false-test is the expression:
3955 (equal characteristic 'fierce)
3959 In Lisp, @code{equal} is a function that determines whether its first
3960 argument is equal to its second argument. The second argument is the
3961 quoted symbol @code{'fierce} and the first argument is the value of the
3962 symbol @code{characteristic}---in other words, the argument passed to
3965 In the first exercise of @code{type-of-animal}, the argument
3966 @code{fierce} is passed to @code{type-of-animal}. Since @code{fierce}
3967 is equal to @code{fierce}, the expression, @code{(equal characteristic
3968 'fierce)}, returns a value of true. When this happens, the @code{if}
3969 evaluates the second argument or then-part of the @code{if}:
3970 @code{(message "It's tiger!")}.
3972 On the other hand, in the second exercise of @code{type-of-animal}, the
3973 argument @code{zebra} is passed to @code{type-of-animal}. @code{zebra}
3974 is not equal to @code{fierce}, so the then-part is not evaluated and
3975 @code{nil} is returned by the @code{if} expression.
3978 @section If--then--else Expressions
3981 An @code{if} expression may have an optional third argument, called
3982 the @dfn{else-part}, for the case when the true-or-false-test returns
3983 false. When this happens, the second argument or then-part of the
3984 overall @code{if} expression is @emph{not} evaluated, but the third or
3985 else-part @emph{is} evaluated. You might think of this as the cloudy
3986 day alternative for the decision ``if it is warm and sunny, then go to
3987 the beach, else read a book!''.
3989 The word ``else'' is not written in the Lisp code; the else-part of an
3990 @code{if} expression comes after the then-part. In the written Lisp, the
3991 else-part is usually written to start on a line of its own and is
3992 indented less than the then-part:
3996 (if @var{true-or-false-test}
3997 @var{action-to-carry-out-if-the-test-returns-true}
3998 @var{action-to-carry-out-if-the-test-returns-false})
4002 For example, the following @code{if} expression prints the message @samp{4
4003 is not greater than 5!} when you evaluate it in the usual way:
4007 (if (> 4 5) ; @r{if-part}
4008 (message "4 falsely greater than 5!") ; @r{then-part}
4009 (message "4 is not greater than 5!")) ; @r{else-part}
4014 Note that the different levels of indentation make it easy to
4015 distinguish the then-part from the else-part. (GNU Emacs has several
4016 commands that automatically indent @code{if} expressions correctly.
4017 @xref{Typing Lists, , GNU Emacs Helps You Type Lists}.)
4019 We can extend the @code{type-of-animal} function to include an
4020 else-part by simply incorporating an additional part to the @code{if}
4024 You can see the consequences of doing this if you evaluate the following
4025 version of the @code{type-of-animal} function definition to install it
4026 and then evaluate the two subsequent expressions to pass different
4027 arguments to the function.
4031 (defun type-of-animal (characteristic) ; @r{Second version.}
4032 "Print message in echo area depending on CHARACTERISTIC.
4033 If the CHARACTERISTIC is the symbol `fierce',
4034 then warn of a tiger;
4035 else say it's not fierce."
4036 (if (equal characteristic 'fierce)
4037 (message "It's a tiger!")
4038 (message "It's not fierce!")))
4045 (type-of-animal 'fierce)
4047 (type-of-animal 'zebra)
4052 @c Following sentence rewritten to prevent overfull hbox.
4054 When you evaluate @code{(type-of-animal 'fierce)}, you will see the
4055 following message printed in the echo area: @code{"It's a tiger!"}; but
4056 when you evaluate @code{(type-of-animal 'zebra)}, you will see
4057 @code{"It's not fierce!"}.
4059 (Of course, if the @var{characteristic} were @code{ferocious}, the
4060 message @code{"It's not fierce!"} would be printed; and it would be
4061 misleading! When you write code, you need to take into account the
4062 possibility that some such argument will be tested by the @code{if}
4063 and write your program accordingly.)
4065 @node Truth & Falsehood
4066 @section Truth and Falsehood in Emacs Lisp
4067 @cindex Truth and falsehood in Emacs Lisp
4068 @cindex Falsehood and truth in Emacs Lisp
4071 There is an important aspect to the truth test in an @code{if}
4072 expression. So far, we have spoken of ``true'' and ``false'' as values of
4073 predicates as if they were new kinds of Emacs Lisp objects. In fact,
4074 ``false'' is just our old friend @code{nil}. Anything else---anything
4075 at all---is ``true''.
4077 The expression that tests for truth is interpreted as @dfn{true}
4078 if the result of evaluating it is a value that is not @code{nil}. In
4079 other words, the result of the test is considered true if the value
4080 returned is a number such as 47, a string such as @code{"hello"}, or a
4081 symbol (other than @code{nil}) such as @code{flowers}, or a list (so
4082 long as it is not empty), or even a buffer!
4085 * nil explained:: @code{nil} has two meanings.
4090 @unnumberedsubsec An explanation of @code{nil}
4093 Before illustrating a test for truth, we need an explanation of @code{nil}.
4095 In Emacs Lisp, the symbol @code{nil} has two meanings. First, it means the
4096 empty list. Second, it means false and is the value returned when a
4097 true-or-false-test tests false. @code{nil} can be written as an empty
4098 list, @code{()}, or as @code{nil}. As far as the Lisp interpreter is
4099 concerned, @code{()} and @code{nil} are the same. Humans, however, tend
4100 to use @code{nil} for false and @code{()} for the empty list.
4102 In Emacs Lisp, any value that is not @code{nil}---is not the empty
4103 list---is considered true. This means that if an evaluation returns
4104 something that is not an empty list, an @code{if} expression will test
4105 true. For example, if a number is put in the slot for the test, it
4106 will be evaluated and will return itself, since that is what numbers
4107 do when evaluated. In this conditional, the @code{if} expression will
4108 test true. The expression tests false only when @code{nil}, an empty
4109 list, is returned by evaluating the expression.
4111 You can see this by evaluating the two expressions in the following examples.
4113 In the first example, the number 4 is evaluated as the test in the
4114 @code{if} expression and returns itself; consequently, the then-part
4115 of the expression is evaluated and returned: @samp{true} appears in
4116 the echo area. In the second example, the @code{nil} indicates false;
4117 consequently, the else-part of the expression is evaluated and
4118 returned: @samp{false} appears in the echo area.
4135 Incidentally, if some other useful value is not available for a test that
4136 returns true, then the Lisp interpreter will return the symbol @code{t}
4137 for true. For example, the expression @code{(> 5 4)} returns @code{t}
4138 when evaluated, as you can see by evaluating it in the usual way:
4146 On the other hand, this function returns @code{nil} if the test is false.
4152 @node save-excursion
4153 @section @code{save-excursion}
4154 @findex save-excursion
4155 @cindex Region, what it is
4156 @cindex Preserving point and buffer
4157 @cindex Point and buffer preservation
4161 The @code{save-excursion} function is the third and final special form
4162 that we will discuss in this chapter.
4164 In Emacs Lisp programs used for editing, the @code{save-excursion}
4165 function is very common. It saves the location of point,
4166 executes the body of the function, and then restores point to
4167 its previous position if its location was changed. Its primary
4168 purpose is to keep the user from being surprised and disturbed by
4169 unexpected movement of point.
4172 * Point and mark:: A review of various locations.
4173 * Template for save-excursion::
4177 @node Point and mark
4178 @unnumberedsubsec Point and Mark
4181 Before discussing @code{save-excursion}, however, it may be useful
4182 first to review what point and mark are in GNU Emacs. @dfn{Point} is
4183 the current location of the cursor. Wherever the cursor
4184 is, that is point. More precisely, on terminals where the cursor
4185 appears to be on top of a character, point is immediately before the
4186 character. In Emacs Lisp, point is an integer. The first character in
4187 a buffer is number one, the second is number two, and so on. The
4188 function @code{point} returns the current position of the cursor as a
4189 number. Each buffer has its own value for point.
4191 The @dfn{mark} is another position in the buffer; its value can be set
4192 with a command such as @kbd{C-@key{SPC}} (@code{set-mark-command}). If
4193 a mark has been set, you can use the command @kbd{C-x C-x}
4194 (@code{exchange-point-and-mark}) to cause the cursor to jump to the mark
4195 and set the mark to be the previous position of point. In addition, if
4196 you set another mark, the position of the previous mark is saved in the
4197 mark ring. Many mark positions can be saved this way. You can jump the
4198 cursor to a saved mark by typing @kbd{C-u C-@key{SPC}} one or more
4201 The part of the buffer between point and mark is called @dfn{the
4202 region}. Numerous commands work on the region, including
4203 @code{center-region}, @code{count-lines-region}, @code{kill-region}, and
4204 @code{print-region}.
4206 The @code{save-excursion} special form saves the location of point and
4207 restores this position after the code within the body of the
4208 special form is evaluated by the Lisp interpreter. Thus, if point were
4209 in the beginning of a piece of text and some code moved point to the end
4210 of the buffer, the @code{save-excursion} would put point back to where
4211 it was before, after the expressions in the body of the function were
4214 In Emacs, a function frequently moves point as part of its internal
4215 workings even though a user would not expect this. For example,
4216 @code{count-lines-region} moves point. To prevent the user from being
4217 bothered by jumps that are both unexpected and (from the user's point of
4218 view) unnecessary, @code{save-excursion} is often used to keep point in
4219 the location expected by the user. The use of
4220 @code{save-excursion} is good housekeeping.
4222 To make sure the house stays clean, @code{save-excursion} restores the
4223 value of point even if something goes wrong in the code inside
4224 of it (or, to be more precise and to use the proper jargon, ``in case of
4225 abnormal exit''). This feature is very helpful.
4227 In addition to recording the value of point,
4228 @code{save-excursion} keeps track of the current buffer, and restores
4229 it, too. This means you can write code that will change the buffer and
4230 have @code{save-excursion} switch you back to the original buffer.
4231 This is how @code{save-excursion} is used in @code{append-to-buffer}.
4232 (@xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.)
4234 @node Template for save-excursion
4235 @subsection Template for a @code{save-excursion} Expression
4238 The template for code using @code{save-excursion} is simple:
4248 The body of the function is one or more expressions that will be
4249 evaluated in sequence by the Lisp interpreter. If there is more than
4250 one expression in the body, the value of the last one will be returned
4251 as the value of the @code{save-excursion} function. The other
4252 expressions in the body are evaluated only for their side effects; and
4253 @code{save-excursion} itself is used only for its side effect (which
4254 is restoring the position of point).
4257 In more detail, the template for a @code{save-excursion} expression
4263 @var{first-expression-in-body}
4264 @var{second-expression-in-body}
4265 @var{third-expression-in-body}
4267 @var{last-expression-in-body})
4272 An expression, of course, may be a symbol on its own or a list.
4274 In Emacs Lisp code, a @code{save-excursion} expression often occurs
4275 within the body of a @code{let} expression. It looks like this:
4288 In the last few chapters we have introduced a macro and a fair number
4289 of functions and special forms. Here they are described in brief,
4290 along with a few similar functions that have not been mentioned yet.
4293 @item eval-last-sexp
4294 Evaluate the last symbolic expression before the current location of
4295 point. The value is printed in the echo area unless the function is
4296 invoked with an argument; in that case, the output is printed in the
4297 current buffer. This command is normally bound to @kbd{C-x C-e}.
4300 Define function. This macro has up to five parts: the name, a
4301 template for the arguments that will be passed to the function,
4302 documentation, an optional interactive declaration, and the body of
4306 For example, in an early version of Emacs, the function definition was
4307 as follows. (It is slightly more complex now that it seeks the first
4308 non-whitespace character rather than the first visible character.)
4312 (defun back-to-indentation ()
4313 "Move point to first visible character on line."
4315 (beginning-of-line 1)
4316 (skip-chars-forward " \t"))
4323 (defun backward-to-indentation (&optional arg)
4324 "Move backward ARG lines and position at first nonblank character."
4326 (forward-line (- (or arg 1)))
4327 (skip-chars-forward " \t"))
4329 (defun back-to-indentation ()
4330 "Move point to the first non-whitespace character on this line."
4332 (beginning-of-line 1)
4333 (skip-syntax-forward " " (line-end-position))
4334 ;; Move back over chars that have whitespace syntax but have the p flag.
4335 (backward-prefix-chars))
4339 Declare to the interpreter that the function can be used
4340 interactively. This special form may be followed by a string with one
4341 or more parts that pass the information to the arguments of the
4342 function, in sequence. These parts may also tell the interpreter to
4343 prompt for information. Parts of the string are separated by
4344 newlines, @samp{\n}.
4347 Common code characters are:
4351 The name of an existing buffer.
4354 The name of an existing file.
4357 The numeric prefix argument. (Note that this @code{p} is lower case.)
4360 Point and the mark, as two numeric arguments, smallest first. This
4361 is the only code letter that specifies two successive arguments
4365 @xref{Interactive Codes, , Code Characters for @samp{interactive},
4366 elisp, The GNU Emacs Lisp Reference Manual}, for a complete list of
4370 Declare that a list of variables is for use within the body of the
4371 @code{let} and give them an initial value, either @code{nil} or a
4372 specified value; then evaluate the rest of the expressions in the body
4373 of the @code{let} and return the value of the last one. Inside the
4374 body of the @code{let}, the Lisp interpreter does not see the values of
4375 the variables of the same names that are bound outside of the
4383 (let ((foo (buffer-name))
4384 (bar (buffer-size)))
4386 "This buffer is %s and has %d characters."
4391 @item save-excursion
4392 Record the values of point and the current buffer before
4393 evaluating the body of this special form. Restore the value of point and
4401 (message "We are %d characters into this buffer."
4404 (goto-char (point-min)) (point))))
4409 Evaluate the first argument to the function; if it is true, evaluate
4410 the second argument; else evaluate the third argument, if there is one.
4412 The @code{if} special form is called a @dfn{conditional}. There are
4413 other conditionals in Emacs Lisp, but @code{if} is perhaps the most
4421 (if (= 22 emacs-major-version)
4422 (message "This is version 22 Emacs")
4423 (message "This is not version 22 Emacs"))
4432 The @code{<} function tests whether its first argument is smaller than
4433 its second argument. A corresponding function, @code{>}, tests whether
4434 the first argument is greater than the second. Likewise, @code{<=}
4435 tests whether the first argument is less than or equal to the second and
4436 @code{>=} tests whether the first argument is greater than or equal to
4437 the second. In all cases, both arguments must be numbers or markers
4438 (markers indicate positions in buffers).
4442 The @code{=} function tests whether two arguments, both numbers or
4448 Test whether two objects are the same. @code{equal} uses one meaning
4449 of the word ``same'' and @code{eq} uses another: @code{equal} returns
4450 true if the two objects have a similar structure and contents, such as
4451 two copies of the same book. On the other hand, @code{eq}, returns
4452 true if both arguments are actually the same object.
4461 The @code{string-lessp} function tests whether its first argument is
4462 smaller than the second argument. A shorter, alternative name for the
4463 same function (a @code{defalias}) is @code{string<}.
4465 The arguments to @code{string-lessp} must be strings or symbols; the
4466 ordering is lexicographic, so case is significant. The print names of
4467 symbols are used instead of the symbols themselves.
4469 @cindex @samp{empty string} defined
4470 An empty string, @samp{""}, a string with no characters in it, is
4471 smaller than any string of characters.
4473 @code{string-equal} provides the corresponding test for equality. Its
4474 shorter, alternative name is @code{string=}. There are no string test
4475 functions that correspond to @var{>}, @code{>=}, or @code{<=}.
4478 Print a message in the echo area. The first argument is a string that
4479 can contain @samp{%s}, @samp{%d}, or @samp{%c} to print the value of
4480 arguments that follow the string. The argument used by @samp{%s} must
4481 be a string or a symbol; the argument used by @samp{%d} must be a
4482 number. The argument used by @samp{%c} must be an @sc{ascii} code
4483 number; it will be printed as the character with that @sc{ascii} code.
4484 (Various other %-sequences have not been mentioned.)
4488 The @code{setq} function sets the value of its first argument to the
4489 value of the second argument. The first argument is automatically
4490 quoted by @code{setq}. It does the same for succeeding pairs of
4491 arguments. Another function, @code{set}, takes only two arguments and
4492 evaluates both of them before setting the value returned by its first
4493 argument to the value returned by its second argument.
4496 Without an argument, return the name of the buffer, as a string.
4498 @item buffer-file-name
4499 Without an argument, return the name of the file the buffer is
4502 @item current-buffer
4503 Return the buffer in which Emacs is active; it may not be
4504 the buffer that is visible on the screen.
4507 Return the most recently selected buffer (other than the buffer passed
4508 to @code{other-buffer} as an argument and other than the current
4511 @item switch-to-buffer
4512 Select a buffer for Emacs to be active in and display it in the current
4513 window so users can look at it. Usually bound to @kbd{C-x b}.
4516 Switch Emacs's attention to a buffer on which programs will run. Don't
4517 alter what the window is showing.
4520 Return the number of characters in the current buffer.
4523 Return the value of the current position of the cursor, as an
4524 integer counting the number of characters from the beginning of the
4528 Return the minimum permissible value of point in
4529 the current buffer. This is 1, unless narrowing is in effect.
4532 Return the value of the maximum permissible value of point in the
4533 current buffer. This is the end of the buffer, unless narrowing is in
4538 @node defun Exercises
4543 Write a non-interactive function that doubles the value of its
4544 argument, a number. Make that function interactive.
4547 Write a function that tests whether the current value of
4548 @code{fill-column} is greater than the argument passed to the function,
4549 and if so, prints an appropriate message.
4552 @node Buffer Walk Through
4553 @chapter A Few Buffer-Related Functions
4555 In this chapter we study in detail several of the functions used in GNU
4556 Emacs. This is called a ``walk-through''. These functions are used as
4557 examples of Lisp code, but are not imaginary examples; with the
4558 exception of the first, simplified function definition, these functions
4559 show the actual code used in GNU Emacs. You can learn a great deal from
4560 these definitions. The functions described here are all related to
4561 buffers. Later, we will study other functions.
4564 * Finding More:: How to find more information.
4565 * simplified-beginning-of-buffer:: Shows @code{goto-char},
4566 @code{point-min}, and @code{push-mark}.
4567 * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}.
4568 * append-to-buffer:: Uses @code{save-excursion} and
4569 @code{insert-buffer-substring}.
4570 * Buffer Related Review:: Review.
4571 * Buffer Exercises::
4575 @section Finding More Information
4577 @findex describe-function, @r{introduced}
4578 @cindex Find function documentation
4579 In this walk-through, I will describe each new function as we come to
4580 it, sometimes in detail and sometimes briefly. If you are interested,
4581 you can get the full documentation of any Emacs Lisp function at any
4582 time by typing @kbd{C-h f} and then the name of the function (and then
4583 @key{RET}). Similarly, you can get the full documentation for a
4584 variable by typing @kbd{C-h v} and then the name of the variable (and
4587 @cindex Find source of function
4588 @c In version 22, tells location both of C and of Emacs Lisp
4589 Also, @code{describe-function} will tell you the location of the
4590 function definition.
4592 Put point into the name of the file that contains the function and
4593 press the @key{RET} key. In this case, @key{RET} means
4594 @code{push-button} rather than ``return'' or ``enter''. Emacs will take
4595 you directly to the function definition.
4600 If you move point over the file name and press
4601 the @key{RET} key, which in this case means @code{help-follow} rather
4602 than ``return'' or ``enter'', Emacs will take you directly to the function
4606 More generally, if you want to see a function in its original source
4607 file, you can use the @code{find-tag} function to jump to it.
4608 @code{find-tag} works with a wide variety of languages, not just
4609 Lisp, and C, and it works with non-programming text as well. For
4610 example, @code{find-tag} will jump to the various nodes in the
4611 Texinfo source file of this document.
4612 The @code{find-tag} function depends on ``tags tables'' that record
4613 the locations of the functions, variables, and other items to which
4614 @code{find-tag} jumps.
4616 To use the @code{find-tag} command, type @kbd{M-.} (i.e., press the
4617 period key while holding down the @key{META} key, or else type the
4618 @key{ESC} key and then type the period key), and then, at the prompt,
4619 type in the name of the function whose source code you want to see,
4620 such as @code{mark-whole-buffer}, and then type @key{RET}. Emacs will
4621 switch buffers and display the source code for the function on your
4622 screen. To switch back to your current buffer, type @kbd{C-x b
4623 @key{RET}}. (On some keyboards, the @key{META} key is labeled
4626 @c !!! 22.1.1 tags table location in this paragraph
4627 @cindex TAGS table, specifying
4629 Depending on how the initial default values of your copy of Emacs are
4630 set, you may also need to specify the location of your ``tags table'',
4631 which is a file called @file{TAGS}. For example, if you are
4632 interested in Emacs sources, the tags table you will most likely want,
4633 if it has already been created for you, will be in a subdirectory of
4634 the @file{/usr/local/share/emacs/} directory; thus you would use the
4635 @code{M-x visit-tags-table} command and specify a pathname such as
4636 @file{/usr/local/share/emacs/22.1.1/lisp/TAGS}. If the tags table
4637 has not already been created, you will have to create it yourself. It
4638 will be in a file such as @file{/usr/local/src/emacs/src/TAGS}.
4641 To create a @file{TAGS} file in a specific directory, switch to that
4642 directory in Emacs using @kbd{M-x cd} command, or list the directory
4643 with @kbd{C-x d} (@code{dired}). Then run the compile command, with
4644 @w{@code{etags *.el}} as the command to execute:
4647 M-x compile RET etags *.el RET
4650 For more information, see @ref{etags, , Create Your Own @file{TAGS} File}.
4652 After you become more familiar with Emacs Lisp, you will find that you will
4653 frequently use @code{find-tag} to navigate your way around source code;
4654 and you will create your own @file{TAGS} tables.
4656 @cindex Library, as term for ``file''
4657 Incidentally, the files that contain Lisp code are conventionally
4658 called @dfn{libraries}. The metaphor is derived from that of a
4659 specialized library, such as a law library or an engineering library,
4660 rather than a general library. Each library, or file, contains
4661 functions that relate to a particular topic or activity, such as
4662 @file{abbrev.el} for handling abbreviations and other typing
4663 shortcuts, and @file{help.el} for help. (Sometimes several
4664 libraries provide code for a single activity, as the various
4665 @file{rmail@dots{}} files provide code for reading electronic mail.)
4666 In @cite{The GNU Emacs Manual}, you will see sentences such as ``The
4667 @kbd{C-h p} command lets you search the standard Emacs Lisp libraries
4668 by topic keywords.''
4670 @node simplified-beginning-of-buffer
4671 @section A Simplified @code{beginning-of-buffer} Definition
4672 @findex simplified-beginning-of-buffer
4674 The @code{beginning-of-buffer} command is a good function to start with
4675 since you are likely to be familiar with it and it is easy to
4676 understand. Used as an interactive command, @code{beginning-of-buffer}
4677 moves the cursor to the beginning of the buffer, leaving the mark at the
4678 previous position. It is generally bound to @kbd{M-<}.
4680 In this section, we will discuss a shortened version of the function
4681 that shows how it is most frequently used. This shortened function
4682 works as written, but it does not contain the code for a complex option.
4683 In another section, we will describe the entire function.
4684 (@xref{beginning-of-buffer, , Complete Definition of
4685 @code{beginning-of-buffer}}.)
4687 Before looking at the code, let's consider what the function
4688 definition has to contain: it must include an expression that makes
4689 the function interactive so it can be called by typing @kbd{M-x
4690 beginning-of-buffer} or by typing a keychord such as @kbd{M-<}; it
4691 must include code to leave a mark at the original position in the
4692 buffer; and it must include code to move the cursor to the beginning
4696 Here is the complete text of the shortened version of the function:
4700 (defun simplified-beginning-of-buffer ()
4701 "Move point to the beginning of the buffer;
4702 leave mark at previous position."
4705 (goto-char (point-min)))
4709 Like all function definitions, this definition has five parts following
4710 the macro @code{defun}:
4714 The name: in this example, @code{simplified-beginning-of-buffer}.
4717 A list of the arguments: in this example, an empty list, @code{()},
4720 The documentation string.
4723 The interactive expression.
4730 In this function definition, the argument list is empty; this means that
4731 this function does not require any arguments. (When we look at the
4732 definition for the complete function, we will see that it may be passed
4733 an optional argument.)
4735 The interactive expression tells Emacs that the function is intended to
4736 be used interactively. In this example, @code{interactive} does not have
4737 an argument because @code{simplified-beginning-of-buffer} does not
4741 The body of the function consists of the two lines:
4746 (goto-char (point-min))
4750 The first of these lines is the expression, @code{(push-mark)}. When
4751 this expression is evaluated by the Lisp interpreter, it sets a mark at
4752 the current position of the cursor, wherever that may be. The position
4753 of this mark is saved in the mark ring.
4755 The next line is @code{(goto-char (point-min))}. This expression
4756 jumps the cursor to the minimum point in the buffer, that is, to the
4757 beginning of the buffer (or to the beginning of the accessible portion
4758 of the buffer if it is narrowed. @xref{Narrowing & Widening, ,
4759 Narrowing and Widening}.)
4761 The @code{push-mark} command sets a mark at the place where the cursor
4762 was located before it was moved to the beginning of the buffer by the
4763 @code{(goto-char (point-min))} expression. Consequently, you can, if
4764 you wish, go back to where you were originally by typing @kbd{C-x C-x}.
4766 That is all there is to the function definition!
4768 @findex describe-function
4769 When you are reading code such as this and come upon an unfamiliar
4770 function, such as @code{goto-char}, you can find out what it does by
4771 using the @code{describe-function} command. To use this command, type
4772 @kbd{C-h f} and then type in the name of the function and press
4773 @key{RET}. The @code{describe-function} command will print the
4774 function's documentation string in a @file{*Help*} window. For
4775 example, the documentation for @code{goto-char} is:
4779 Set point to POSITION, a number or marker.
4780 Beginning of buffer is position (point-min), end is (point-max).
4785 The function's one argument is the desired position.
4788 (The prompt for @code{describe-function} will offer you the symbol
4789 under or preceding the cursor, so you can save typing by positioning
4790 the cursor right over or after the function and then typing @kbd{C-h f
4793 The @code{end-of-buffer} function definition is written in the same way as
4794 the @code{beginning-of-buffer} definition except that the body of the
4795 function contains the expression @code{(goto-char (point-max))} in place
4796 of @code{(goto-char (point-min))}.
4798 @node mark-whole-buffer
4799 @section The Definition of @code{mark-whole-buffer}
4800 @findex mark-whole-buffer
4802 The @code{mark-whole-buffer} function is no harder to understand than the
4803 @code{simplified-beginning-of-buffer} function. In this case, however,
4804 we will look at the complete function, not a shortened version.
4806 The @code{mark-whole-buffer} function is not as commonly used as the
4807 @code{beginning-of-buffer} function, but is useful nonetheless: it
4808 marks a whole buffer as a region by putting point at the beginning and
4809 a mark at the end of the buffer. It is generally bound to @kbd{C-x
4813 * mark-whole-buffer overview::
4814 * Body of mark-whole-buffer:: Only three lines of code.
4818 @node mark-whole-buffer overview
4819 @unnumberedsubsec An overview of @code{mark-whole-buffer}
4823 In GNU Emacs 22, the code for the complete function looks like this:
4827 (defun mark-whole-buffer ()
4828 "Put point at beginning and mark at end of buffer.
4829 You probably should not use this function in Lisp programs;
4830 it is usually a mistake for a Lisp function to use any subroutine
4831 that uses or sets the mark."
4834 (push-mark (point-max) nil t)
4835 (goto-char (point-min)))
4840 Like all other functions, the @code{mark-whole-buffer} function fits
4841 into the template for a function definition. The template looks like
4846 (defun @var{name-of-function} (@var{argument-list})
4847 "@var{documentation}@dots{}"
4848 (@var{interactive-expression}@dots{})
4853 Here is how the function works: the name of the function is
4854 @code{mark-whole-buffer}; it is followed by an empty argument list,
4855 @samp{()}, which means that the function does not require arguments.
4856 The documentation comes next.
4858 The next line is an @code{(interactive)} expression that tells Emacs
4859 that the function will be used interactively. These details are similar
4860 to the @code{simplified-beginning-of-buffer} function described in the
4864 @node Body of mark-whole-buffer
4865 @subsection Body of @code{mark-whole-buffer}
4867 The body of the @code{mark-whole-buffer} function consists of three
4874 (push-mark (point-max) nil t)
4875 (goto-char (point-min))
4879 The first of these lines is the expression, @code{(push-mark (point))}.
4881 This line does exactly the same job as the first line of the body of
4882 the @code{simplified-beginning-of-buffer} function, which is written
4883 @code{(push-mark)}. In both cases, the Lisp interpreter sets a mark
4884 at the current position of the cursor.
4886 I don't know why the expression in @code{mark-whole-buffer} is written
4887 @code{(push-mark (point))} and the expression in
4888 @code{beginning-of-buffer} is written @code{(push-mark)}. Perhaps
4889 whoever wrote the code did not know that the arguments for
4890 @code{push-mark} are optional and that if @code{push-mark} is not
4891 passed an argument, the function automatically sets mark at the
4892 location of point by default. Or perhaps the expression was written
4893 so as to parallel the structure of the next line. In any case, the
4894 line causes Emacs to determine the position of point and set a mark
4897 In earlier versions of GNU Emacs, the next line of
4898 @code{mark-whole-buffer} was @code{(push-mark (point-max))}. This
4899 expression sets a mark at the point in the buffer that has the highest
4900 number. This will be the end of the buffer (or, if the buffer is
4901 narrowed, the end of the accessible portion of the buffer.
4902 @xref{Narrowing & Widening, , Narrowing and Widening}, for more about
4903 narrowing.) After this mark has been set, the previous mark, the one
4904 set at point, is no longer set, but Emacs remembers its position, just
4905 as all other recent marks are always remembered. This means that you
4906 can, if you wish, go back to that position by typing @kbd{C-u
4910 In GNU Emacs 22, the @code{(point-max)} is slightly more complicated.
4914 (push-mark (point-max) nil t)
4918 The expression works nearly the same as before. It sets a mark at the
4919 highest numbered place in the buffer that it can. However, in this
4920 version, @code{push-mark} has two additional arguments. The second
4921 argument to @code{push-mark} is @code{nil}. This tells the function
4922 it @emph{should} display a message that says ``Mark set'' when it pushes
4923 the mark. The third argument is @code{t}. This tells
4924 @code{push-mark} to activate the mark when Transient Mark mode is
4925 turned on. Transient Mark mode highlights the currently active
4926 region. It is often turned off.
4928 Finally, the last line of the function is @code{(goto-char
4929 (point-min)))}. This is written exactly the same way as it is written
4930 in @code{beginning-of-buffer}. The expression moves the cursor to
4931 the minimum point in the buffer, that is, to the beginning of the buffer
4932 (or to the beginning of the accessible portion of the buffer). As a
4933 result of this, point is placed at the beginning of the buffer and mark
4934 is set at the end of the buffer. The whole buffer is, therefore, the
4937 @node append-to-buffer
4938 @section The Definition of @code{append-to-buffer}
4939 @findex append-to-buffer
4941 The @code{append-to-buffer} command is more complex than the
4942 @code{mark-whole-buffer} command. What it does is copy the region
4943 (that is, the part of the buffer between point and mark) from the
4944 current buffer to a specified buffer.
4947 * append-to-buffer overview::
4948 * append interactive:: A two part interactive expression.
4949 * append-to-buffer body:: Incorporates a @code{let} expression.
4950 * append save-excursion:: How the @code{save-excursion} works.
4954 @node append-to-buffer overview
4955 @unnumberedsubsec An Overview of @code{append-to-buffer}
4958 @findex insert-buffer-substring
4959 The @code{append-to-buffer} command uses the
4960 @code{insert-buffer-substring} function to copy the region.
4961 @code{insert-buffer-substring} is described by its name: it takes a
4962 string of characters from part of a buffer, a ``substring'', and
4963 inserts them into another buffer.
4965 Most of @code{append-to-buffer} is
4966 concerned with setting up the conditions for
4967 @code{insert-buffer-substring} to work: the code must specify both the
4968 buffer to which the text will go, the window it comes from and goes
4969 to, and the region that will be copied.
4972 Here is the complete text of the function:
4976 (defun append-to-buffer (buffer start end)
4977 "Append to specified buffer the text of the region.
4978 It is inserted into that buffer before its point.
4982 When calling from a program, give three arguments:
4983 BUFFER (or buffer name), START and END.
4984 START and END specify the portion of the current buffer to be copied."
4986 (list (read-buffer "Append to buffer: " (other-buffer
4987 (current-buffer) t))
4988 (region-beginning) (region-end)))
4991 (let ((oldbuf (current-buffer)))
4993 (let* ((append-to (get-buffer-create buffer))
4994 (windows (get-buffer-window-list append-to t t))
4996 (set-buffer append-to)
4997 (setq point (point))
4998 (barf-if-buffer-read-only)
4999 (insert-buffer-substring oldbuf start end)
5000 (dolist (window windows)
5001 (when (= (window-point window) point)
5002 (set-window-point window (point))))))))
5006 The function can be understood by looking at it as a series of
5007 filled-in templates.
5009 The outermost template is for the function definition. In this
5010 function, it looks like this (with several slots filled in):
5014 (defun append-to-buffer (buffer start end)
5015 "@var{documentation}@dots{}"
5016 (interactive @dots{})
5021 The first line of the function includes its name and three arguments.
5022 The arguments are the @code{buffer} to which the text will be copied, and
5023 the @code{start} and @code{end} of the region in the current buffer that
5026 The next part of the function is the documentation, which is clear and
5027 complete. As is conventional, the three arguments are written in
5028 upper case so you will notice them easily. Even better, they are
5029 described in the same order as in the argument list.
5031 Note that the documentation distinguishes between a buffer and its
5032 name. (The function can handle either.)
5034 @node append interactive
5035 @subsection The @code{append-to-buffer} Interactive Expression
5037 Since the @code{append-to-buffer} function will be used interactively,
5038 the function must have an @code{interactive} expression. (For a
5039 review of @code{interactive}, see @ref{Interactive, , Making a
5040 Function Interactive}.) The expression reads as follows:
5046 "Append to buffer: "
5047 (other-buffer (current-buffer) t))
5054 This expression is not one with letters standing for parts, as
5055 described earlier. Instead, it starts a list with these parts:
5057 The first part of the list is an expression to read the name of a
5058 buffer and return it as a string. That is @code{read-buffer}. The
5059 function requires a prompt as its first argument, @samp{"Append to
5060 buffer: "}. Its second argument tells the command what value to
5061 provide if you don't specify anything.
5063 In this case that second argument is an expression containing the
5064 function @code{other-buffer}, an exception, and a @samp{t}, standing
5067 The first argument to @code{other-buffer}, the exception, is yet
5068 another function, @code{current-buffer}. That is not going to be
5069 returned. The second argument is the symbol for true, @code{t}. that
5070 tells @code{other-buffer} that it may show visible buffers (except in
5071 this case, it will not show the current buffer, which makes sense).
5074 The expression looks like this:
5077 (other-buffer (current-buffer) t)
5080 The second and third arguments to the @code{list} expression are
5081 @code{(region-beginning)} and @code{(region-end)}. These two
5082 functions specify the beginning and end of the text to be appended.
5085 Originally, the command used the letters @samp{B} and @samp{r}.
5086 The whole @code{interactive} expression looked like this:
5089 (interactive "BAppend to buffer:@: \nr")
5093 But when that was done, the default value of the buffer switched to
5094 was invisible. That was not wanted.
5096 (The prompt was separated from the second argument with a newline,
5097 @samp{\n}. It was followed by an @samp{r} that told Emacs to bind the
5098 two arguments that follow the symbol @code{buffer} in the function's
5099 argument list (that is, @code{start} and @code{end}) to the values of
5100 point and mark. That argument worked fine.)
5102 @node append-to-buffer body
5103 @subsection The Body of @code{append-to-buffer}
5106 in GNU Emacs 22 in /usr/local/src/emacs/lisp/simple.el
5108 (defun append-to-buffer (buffer start end)
5109 "Append to specified buffer the text of the region.
5110 It is inserted into that buffer before its point.
5112 When calling from a program, give three arguments:
5113 BUFFER (or buffer name), START and END.
5114 START and END specify the portion of the current buffer to be copied."
5116 (list (read-buffer "Append to buffer: " (other-buffer (current-buffer) t))
5117 (region-beginning) (region-end)))
5118 (let ((oldbuf (current-buffer)))
5120 (let* ((append-to (get-buffer-create buffer))
5121 (windows (get-buffer-window-list append-to t t))
5123 (set-buffer append-to)
5124 (setq point (point))
5125 (barf-if-buffer-read-only)
5126 (insert-buffer-substring oldbuf start end)
5127 (dolist (window windows)
5128 (when (= (window-point window) point)
5129 (set-window-point window (point))))))))
5132 The body of the @code{append-to-buffer} function begins with @code{let}.
5134 As we have seen before (@pxref{let, , @code{let}}), the purpose of a
5135 @code{let} expression is to create and give initial values to one or
5136 more variables that will only be used within the body of the
5137 @code{let}. This means that such a variable will not be confused with
5138 any variable of the same name outside the @code{let} expression.
5140 We can see how the @code{let} expression fits into the function as a
5141 whole by showing a template for @code{append-to-buffer} with the
5142 @code{let} expression in outline:
5146 (defun append-to-buffer (buffer start end)
5147 "@var{documentation}@dots{}"
5148 (interactive @dots{})
5149 (let ((@var{variable} @var{value}))
5154 The @code{let} expression has three elements:
5158 The symbol @code{let};
5161 A varlist containing, in this case, a single two-element list,
5162 @code{(@var{variable} @var{value})};
5165 The body of the @code{let} expression.
5169 In the @code{append-to-buffer} function, the varlist looks like this:
5172 (oldbuf (current-buffer))
5176 In this part of the @code{let} expression, the one variable,
5177 @code{oldbuf}, is bound to the value returned by the
5178 @code{(current-buffer)} expression. The variable, @code{oldbuf}, is
5179 used to keep track of the buffer in which you are working and from
5180 which you will copy.
5182 The element or elements of a varlist are surrounded by a set of
5183 parentheses so the Lisp interpreter can distinguish the varlist from
5184 the body of the @code{let}. As a consequence, the two-element list
5185 within the varlist is surrounded by a circumscribing set of parentheses.
5186 The line looks like this:
5190 (let ((oldbuf (current-buffer)))
5196 The two parentheses before @code{oldbuf} might surprise you if you did
5197 not realize that the first parenthesis before @code{oldbuf} marks the
5198 boundary of the varlist and the second parenthesis marks the beginning
5199 of the two-element list, @code{(oldbuf (current-buffer))}.
5201 @node append save-excursion
5202 @subsection @code{save-excursion} in @code{append-to-buffer}
5204 The body of the @code{let} expression in @code{append-to-buffer}
5205 consists of a @code{save-excursion} expression.
5207 The @code{save-excursion} function saves the location of point, and restores it
5208 to that position after the expressions in the
5209 body of the @code{save-excursion} complete execution. In addition,
5210 @code{save-excursion} keeps track of the original buffer, and
5211 restores it. This is how @code{save-excursion} is used in
5212 @code{append-to-buffer}.
5215 @cindex Indentation for formatting
5216 @cindex Formatting convention
5217 Incidentally, it is worth noting here that a Lisp function is normally
5218 formatted so that everything that is enclosed in a multi-line spread is
5219 indented more to the right than the first symbol. In this function
5220 definition, the @code{let} is indented more than the @code{defun}, and
5221 the @code{save-excursion} is indented more than the @code{let}, like
5237 This formatting convention makes it easy to see that the lines in
5238 the body of the @code{save-excursion} are enclosed by the parentheses
5239 associated with @code{save-excursion}, just as the
5240 @code{save-excursion} itself is enclosed by the parentheses associated
5241 with the @code{let}:
5245 (let ((oldbuf (current-buffer)))
5248 (set-buffer @dots{})
5249 (insert-buffer-substring oldbuf start end)
5255 The use of the @code{save-excursion} function can be viewed as a process
5256 of filling in the slots of a template:
5261 @var{first-expression-in-body}
5262 @var{second-expression-in-body}
5264 @var{last-expression-in-body})
5270 In this function, the body of the @code{save-excursion} contains only
5271 one expression, the @code{let*} expression. You know about a
5272 @code{let} function. The @code{let*} function is different. It has a
5273 @samp{*} in its name. It enables Emacs to set each variable in its
5274 varlist in sequence, one after another.
5276 Its critical feature is that variables later in the varlist can make
5277 use of the values to which Emacs set variables earlier in the varlist.
5278 @xref{fwd-para let, , The @code{let*} expression}.
5280 We will skip functions like @code{let*} and focus on two: the
5281 @code{set-buffer} function and the @code{insert-buffer-substring}
5285 In the old days, the @code{set-buffer} expression was simply
5288 (set-buffer (get-buffer-create buffer))
5296 (set-buffer append-to)
5300 @code{append-to} is bound to @code{(get-buffer-create buffer)} earlier
5301 on in the @code{let*} expression. That extra binding would not be
5302 necessary except for that @code{append-to} is used later in the
5303 varlist as an argument to @code{get-buffer-window-list}.
5308 (let ((oldbuf (current-buffer)))
5310 (let* ((append-to (get-buffer-create buffer))
5311 (windows (get-buffer-window-list append-to t t))
5313 (set-buffer append-to)
5314 (setq point (point))
5315 (barf-if-buffer-read-only)
5316 (insert-buffer-substring oldbuf start end)
5317 (dolist (window windows)
5318 (when (= (window-point window) point)
5319 (set-window-point window (point))))))))
5322 The @code{append-to-buffer} function definition inserts text from the
5323 buffer in which you are currently to a named buffer. It happens that
5324 @code{insert-buffer-substring} copies text from another buffer to the
5325 current buffer, just the reverse---that is why the
5326 @code{append-to-buffer} definition starts out with a @code{let} that
5327 binds the local symbol @code{oldbuf} to the value returned by
5328 @code{current-buffer}.
5331 The @code{insert-buffer-substring} expression looks like this:
5334 (insert-buffer-substring oldbuf start end)
5338 The @code{insert-buffer-substring} function copies a string
5339 @emph{from} the buffer specified as its first argument and inserts the
5340 string into the present buffer. In this case, the argument to
5341 @code{insert-buffer-substring} is the value of the variable created
5342 and bound by the @code{let}, namely the value of @code{oldbuf}, which
5343 was the current buffer when you gave the @code{append-to-buffer}
5346 After @code{insert-buffer-substring} has done its work,
5347 @code{save-excursion} will restore the action to the original buffer
5348 and @code{append-to-buffer} will have done its job.
5351 Written in skeletal form, the workings of the body look like this:
5355 (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer})
5356 (save-excursion ; @r{Keep track of buffer.}
5358 @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer})
5360 @var{change-back-to-original-buffer-when-finished}
5361 @var{let-the-local-meaning-of-}@code{oldbuf}@var{-disappear-when-finished}
5365 In summary, @code{append-to-buffer} works as follows: it saves the
5366 value of the current buffer in the variable called @code{oldbuf}. It
5367 gets the new buffer (creating one if need be) and switches Emacs's
5368 attention to it. Using the value of @code{oldbuf}, it inserts the
5369 region of text from the old buffer into the new buffer; and then using
5370 @code{save-excursion}, it brings you back to your original buffer.
5372 In looking at @code{append-to-buffer}, you have explored a fairly
5373 complex function. It shows how to use @code{let} and
5374 @code{save-excursion}, and how to change to and come back from another
5375 buffer. Many function definitions use @code{let},
5376 @code{save-excursion}, and @code{set-buffer} this way.
5378 @node Buffer Related Review
5381 Here is a brief summary of the various functions discussed in this chapter.
5384 @item describe-function
5385 @itemx describe-variable
5386 Print the documentation for a function or variable.
5387 Conventionally bound to @kbd{C-h f} and @kbd{C-h v}.
5390 Find the file containing the source for a function or variable and
5391 switch buffers to it, positioning point at the beginning of the item.
5392 Conventionally bound to @kbd{M-.} (that's a period following the
5395 @item save-excursion
5396 Save the location of point and restore its value after the
5397 arguments to @code{save-excursion} have been evaluated. Also, remember
5398 the current buffer and return to it.
5401 Set mark at a location and record the value of the previous mark on the
5402 mark ring. The mark is a location in the buffer that will keep its
5403 relative position even if text is added to or removed from the buffer.
5406 Set point to the location specified by the value of the argument, which
5407 can be a number, a marker, or an expression that returns the number of
5408 a position, such as @code{(point-min)}.
5410 @item insert-buffer-substring
5411 Copy a region of text from a buffer that is passed to the function as
5412 an argument and insert the region into the current buffer.
5414 @item mark-whole-buffer
5415 Mark the whole buffer as a region. Normally bound to @kbd{C-x h}.
5418 Switch the attention of Emacs to another buffer, but do not change the
5419 window being displayed. Used when the program rather than a human is
5420 to work on a different buffer.
5422 @item get-buffer-create
5424 Find a named buffer or create one if a buffer of that name does not
5425 exist. The @code{get-buffer} function returns @code{nil} if the named
5426 buffer does not exist.
5430 @node Buffer Exercises
5435 Write your own @code{simplified-end-of-buffer} function definition;
5436 then test it to see whether it works.
5439 Use @code{if} and @code{get-buffer} to write a function that prints a
5440 message telling you whether a buffer exists.
5443 Using @code{find-tag}, find the source for the @code{copy-to-buffer}
5448 @chapter A Few More Complex Functions
5450 In this chapter, we build on what we have learned in previous chapters
5451 by looking at more complex functions. The @code{copy-to-buffer}
5452 function illustrates use of two @code{save-excursion} expressions in
5453 one definition, while the @code{insert-buffer} function illustrates
5454 use of an asterisk in an @code{interactive} expression, use of
5455 @code{or}, and the important distinction between a name and the object
5456 to which the name refers.
5459 * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}.
5460 * insert-buffer:: Read-only, and with @code{or}.
5461 * beginning-of-buffer:: Shows @code{goto-char},
5462 @code{point-min}, and @code{push-mark}.
5463 * Second Buffer Related Review::
5464 * optional Exercise::
5467 @node copy-to-buffer
5468 @section The Definition of @code{copy-to-buffer}
5469 @findex copy-to-buffer
5471 After understanding how @code{append-to-buffer} works, it is easy to
5472 understand @code{copy-to-buffer}. This function copies text into a
5473 buffer, but instead of adding to the second buffer, it replaces all the
5474 previous text in the second buffer.
5477 The body of @code{copy-to-buffer} looks like this,
5482 (interactive "BCopy to buffer: \nr")
5483 (let ((oldbuf (current-buffer)))
5484 (with-current-buffer (get-buffer-create buffer)
5485 (barf-if-buffer-read-only)
5488 (insert-buffer-substring oldbuf start end)))))
5492 The @code{copy-to-buffer} function has a simpler @code{interactive}
5493 expression than @code{append-to-buffer}.
5496 The definition then says
5499 (with-current-buffer (get-buffer-create buffer) @dots{}
5502 First, look at the earliest inner expression; that is evaluated first.
5503 That expression starts with @code{get-buffer-create buffer}. The
5504 function tells the computer to use the buffer with the name specified
5505 as the one to which you are copying, or if such a buffer does not
5506 exist, to create it. Then, the @code{with-current-buffer} function
5507 evaluates its body with that buffer temporarily current.
5509 (This demonstrates another way to shift the computer's attention but
5510 not the user's. The @code{append-to-buffer} function showed how to do
5511 the same with @code{save-excursion} and @code{set-buffer}.
5512 @code{with-current-buffer} is a newer, and arguably easier,
5515 The @code{barf-if-buffer-read-only} function sends you an error
5516 message saying the buffer is read-only if you cannot modify it.
5518 The next line has the @code{erase-buffer} function as its sole
5519 contents. That function erases the buffer.
5521 Finally, the last two lines contain the @code{save-excursion}
5522 expression with @code{insert-buffer-substring} as its body.
5523 The @code{insert-buffer-substring} expression copies the text from
5524 the buffer you are in (and you have not seen the computer shift its
5525 attention, so you don't know that that buffer is now called
5528 Incidentally, this is what is meant by ``replacement''. To replace text,
5529 Emacs erases the previous text and then inserts new text.
5532 In outline, the body of @code{copy-to-buffer} looks like this:
5536 (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer})
5537 (@var{with-the-buffer-you-are-copying-to}
5538 (@var{but-do-not-erase-or-copy-to-a-read-only-buffer})
5541 @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer})))
5546 @section The Definition of @code{insert-buffer}
5547 @findex insert-buffer
5549 @code{insert-buffer} is yet another buffer-related function. This
5550 command copies another buffer @emph{into} the current buffer. It is the
5551 reverse of @code{append-to-buffer} or @code{copy-to-buffer}, since they
5552 copy a region of text @emph{from} the current buffer to another buffer.
5554 Here is a discussion based on the original code. The code was
5555 simplified in 2003 and is harder to understand.
5557 (@xref{New insert-buffer, , New Body for @code{insert-buffer}}, to see
5558 a discussion of the new body.)
5560 In addition, this code illustrates the use of @code{interactive} with a
5561 buffer that might be @dfn{read-only} and the important distinction
5562 between the name of an object and the object actually referred to.
5565 * insert-buffer code::
5566 * insert-buffer interactive:: When you can read, but not write.
5567 * insert-buffer body:: The body has an @code{or} and a @code{let}.
5568 * if & or:: Using an @code{if} instead of an @code{or}.
5569 * Insert or:: How the @code{or} expression works.
5570 * Insert let:: Two @code{save-excursion} expressions.
5571 * New insert-buffer::
5575 @node insert-buffer code
5576 @unnumberedsubsec The Code for @code{insert-buffer}
5580 Here is the earlier code:
5584 (defun insert-buffer (buffer)
5585 "Insert after point the contents of BUFFER.
5586 Puts mark after the inserted text.
5587 BUFFER may be a buffer or a buffer name."
5588 (interactive "*bInsert buffer:@: ")
5591 (or (bufferp buffer)
5592 (setq buffer (get-buffer buffer)))
5593 (let (start end newmark)
5597 (setq start (point-min) end (point-max)))
5600 (insert-buffer-substring buffer start end)
5601 (setq newmark (point)))
5602 (push-mark newmark)))
5607 As with other function definitions, you can use a template to see an
5608 outline of the function:
5612 (defun insert-buffer (buffer)
5613 "@var{documentation}@dots{}"
5614 (interactive "*bInsert buffer:@: ")
5619 @node insert-buffer interactive
5620 @subsection The Interactive Expression in @code{insert-buffer}
5621 @findex interactive, @r{example use of}
5623 In @code{insert-buffer}, the argument to the @code{interactive}
5624 declaration has two parts, an asterisk, @samp{*}, and @samp{bInsert
5628 * Read-only buffer:: When a buffer cannot be modified.
5629 * b for interactive:: An existing buffer or else its name.
5632 @node Read-only buffer
5633 @unnumberedsubsubsec A Read-only Buffer
5634 @cindex Read-only buffer
5635 @cindex Asterisk for read-only buffer
5636 @findex * @r{for read-only buffer}
5638 The asterisk is for the situation when the current buffer is a
5639 read-only buffer---a buffer that cannot be modified. If
5640 @code{insert-buffer} is called when the current buffer is read-only, a
5641 message to this effect is printed in the echo area and the terminal
5642 may beep or blink at you; you will not be permitted to insert anything
5643 into current buffer. The asterisk does not need to be followed by a
5644 newline to separate it from the next argument.
5646 @node b for interactive
5647 @unnumberedsubsubsec @samp{b} in an Interactive Expression
5649 The next argument in the interactive expression starts with a lower
5650 case @samp{b}. (This is different from the code for
5651 @code{append-to-buffer}, which uses an upper-case @samp{B}.
5652 @xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.)
5653 The lower-case @samp{b} tells the Lisp interpreter that the argument
5654 for @code{insert-buffer} should be an existing buffer or else its
5655 name. (The upper-case @samp{B} option provides for the possibility
5656 that the buffer does not exist.) Emacs will prompt you for the name
5657 of the buffer, offering you a default buffer, with name completion
5658 enabled. If the buffer does not exist, you receive a message that
5659 says ``No match''; your terminal may beep at you as well.
5661 The new and simplified code generates a list for @code{interactive}.
5662 It uses the @code{barf-if-buffer-read-only} and @code{read-buffer}
5663 functions with which we are already familiar and the @code{progn}
5664 special form with which we are not. (It will be described later.)
5666 @node insert-buffer body
5667 @subsection The Body of the @code{insert-buffer} Function
5669 The body of the @code{insert-buffer} function has two major parts: an
5670 @code{or} expression and a @code{let} expression. The purpose of the
5671 @code{or} expression is to ensure that the argument @code{buffer} is
5672 bound to a buffer and not just the name of a buffer. The body of the
5673 @code{let} expression contains the code which copies the other buffer
5674 into the current buffer.
5677 In outline, the two expressions fit into the @code{insert-buffer}
5682 (defun insert-buffer (buffer)
5683 "@var{documentation}@dots{}"
5684 (interactive "*bInsert buffer:@: ")
5689 (let (@var{varlist})
5690 @var{body-of-}@code{let}@dots{} )
5694 To understand how the @code{or} expression ensures that the argument
5695 @code{buffer} is bound to a buffer and not to the name of a buffer, it
5696 is first necessary to understand the @code{or} function.
5698 Before doing this, let me rewrite this part of the function using
5699 @code{if} so that you can see what is done in a manner that will be familiar.
5702 @subsection @code{insert-buffer} With an @code{if} Instead of an @code{or}
5704 The job to be done is to make sure the value of @code{buffer} is a
5705 buffer itself and not the name of a buffer. If the value is the name,
5706 then the buffer itself must be got.
5708 You can imagine yourself at a conference where an usher is wandering
5709 around holding a list with your name on it and looking for you: the
5710 usher is ``bound'' to your name, not to you; but when the usher finds
5711 you and takes your arm, the usher becomes ``bound'' to you.
5714 In Lisp, you might describe this situation like this:
5718 (if (not (holding-on-to-guest))
5719 (find-and-take-arm-of-guest))
5723 We want to do the same thing with a buffer---if we do not have the
5724 buffer itself, we want to get it.
5727 Using a predicate called @code{bufferp} that tells us whether we have a
5728 buffer (rather than its name), we can write the code like this:
5732 (if (not (bufferp buffer)) ; @r{if-part}
5733 (setq buffer (get-buffer buffer))) ; @r{then-part}
5738 Here, the true-or-false-test of the @code{if} expression is
5739 @w{@code{(not (bufferp buffer))}}; and the then-part is the expression
5740 @w{@code{(setq buffer (get-buffer buffer))}}.
5742 In the test, the function @code{bufferp} returns true if its argument is
5743 a buffer---but false if its argument is the name of the buffer. (The
5744 last character of the function name @code{bufferp} is the character
5745 @samp{p}; as we saw earlier, such use of @samp{p} is a convention that
5746 indicates that the function is a predicate, which is a term that means
5747 that the function will determine whether some property is true or false.
5748 @xref{Wrong Type of Argument, , Using the Wrong Type Object as an
5752 The function @code{not} precedes the expression @code{(bufferp buffer)},
5753 so the true-or-false-test looks like this:
5756 (not (bufferp buffer))
5760 @code{not} is a function that returns true if its argument is false
5761 and false if its argument is true. So if @code{(bufferp buffer)}
5762 returns true, the @code{not} expression returns false and vice versa:
5763 what is ``not true'' is false and what is ``not false'' is true.
5765 Using this test, the @code{if} expression works as follows: when the
5766 value of the variable @code{buffer} is actually a buffer rather than
5767 its name, the true-or-false-test returns false and the @code{if}
5768 expression does not evaluate the then-part. This is fine, since we do
5769 not need to do anything to the variable @code{buffer} if it really is
5772 On the other hand, when the value of @code{buffer} is not a buffer
5773 itself, but the name of a buffer, the true-or-false-test returns true
5774 and the then-part of the expression is evaluated. In this case, the
5775 then-part is @code{(setq buffer (get-buffer buffer))}. This
5776 expression uses the @code{get-buffer} function to return an actual
5777 buffer itself, given its name. The @code{setq} then sets the variable
5778 @code{buffer} to the value of the buffer itself, replacing its previous
5779 value (which was the name of the buffer).
5782 @subsection The @code{or} in the Body
5784 The purpose of the @code{or} expression in the @code{insert-buffer}
5785 function is to ensure that the argument @code{buffer} is bound to a
5786 buffer and not just to the name of a buffer. The previous section shows
5787 how the job could have been done using an @code{if} expression.
5788 However, the @code{insert-buffer} function actually uses @code{or}.
5789 To understand this, it is necessary to understand how @code{or} works.
5792 An @code{or} function can have any number of arguments. It evaluates
5793 each argument in turn and returns the value of the first of its
5794 arguments that is not @code{nil}. Also, and this is a crucial feature
5795 of @code{or}, it does not evaluate any subsequent arguments after
5796 returning the first non-@code{nil} value.
5799 The @code{or} expression looks like this:
5803 (or (bufferp buffer)
5804 (setq buffer (get-buffer buffer)))
5809 The first argument to @code{or} is the expression @code{(bufferp buffer)}.
5810 This expression returns true (a non-@code{nil} value) if the buffer is
5811 actually a buffer, and not just the name of a buffer. In the @code{or}
5812 expression, if this is the case, the @code{or} expression returns this
5813 true value and does not evaluate the next expression---and this is fine
5814 with us, since we do not want to do anything to the value of
5815 @code{buffer} if it really is a buffer.
5817 On the other hand, if the value of @code{(bufferp buffer)} is @code{nil},
5818 which it will be if the value of @code{buffer} is the name of a buffer,
5819 the Lisp interpreter evaluates the next element of the @code{or}
5820 expression. This is the expression @code{(setq buffer (get-buffer
5821 buffer))}. This expression returns a non-@code{nil} value, which
5822 is the value to which it sets the variable @code{buffer}---and this
5823 value is a buffer itself, not the name of a buffer.
5825 The result of all this is that the symbol @code{buffer} is always
5826 bound to a buffer itself rather than to the name of a buffer. All
5827 this is necessary because the @code{set-buffer} function in a
5828 following line only works with a buffer itself, not with the name to a
5832 Incidentally, using @code{or}, the situation with the usher would be
5836 (or (holding-on-to-guest) (find-and-take-arm-of-guest))
5840 @subsection The @code{let} Expression in @code{insert-buffer}
5842 After ensuring that the variable @code{buffer} refers to a buffer itself
5843 and not just to the name of a buffer, the @code{insert-buffer function}
5844 continues with a @code{let} expression. This specifies three local
5845 variables, @code{start}, @code{end}, and @code{newmark} and binds them
5846 to the initial value @code{nil}. These variables are used inside the
5847 remainder of the @code{let} and temporarily hide any other occurrence of
5848 variables of the same name in Emacs until the end of the @code{let}.
5851 The body of the @code{let} contains two @code{save-excursion}
5852 expressions. First, we will look at the inner @code{save-excursion}
5853 expression in detail. The expression looks like this:
5859 (setq start (point-min) end (point-max)))
5864 The expression @code{(set-buffer buffer)} changes Emacs's attention
5865 from the current buffer to the one from which the text will copied.
5866 In that buffer, the variables @code{start} and @code{end} are set to
5867 the beginning and end of the buffer, using the commands
5868 @code{point-min} and @code{point-max}. Note that we have here an
5869 illustration of how @code{setq} is able to set two variables in the
5870 same expression. The first argument of @code{setq} is set to the
5871 value of its second, and its third argument is set to the value of its
5874 After the body of the inner @code{save-excursion} is evaluated, the
5875 @code{save-excursion} restores the original buffer, but @code{start} and
5876 @code{end} remain set to the values of the beginning and end of the
5877 buffer from which the text will be copied.
5880 The outer @code{save-excursion} expression looks like this:
5885 (@var{inner-}@code{save-excursion}@var{-expression}
5886 (@var{go-to-new-buffer-and-set-}@code{start}@var{-and-}@code{end})
5887 (insert-buffer-substring buffer start end)
5888 (setq newmark (point)))
5893 The @code{insert-buffer-substring} function copies the text
5894 @emph{into} the current buffer @emph{from} the region indicated by
5895 @code{start} and @code{end} in @code{buffer}. Since the whole of the
5896 second buffer lies between @code{start} and @code{end}, the whole of
5897 the second buffer is copied into the buffer you are editing. Next,
5898 the value of point, which will be at the end of the inserted text, is
5899 recorded in the variable @code{newmark}.
5901 After the body of the outer @code{save-excursion} is evaluated, point
5902 is relocated to its original place.
5904 However, it is convenient to locate a mark at the end of the newly
5905 inserted text and locate point at its beginning. The @code{newmark}
5906 variable records the end of the inserted text. In the last line of
5907 the @code{let} expression, the @code{(push-mark newmark)} expression
5908 function sets a mark to this location. (The previous location of the
5909 mark is still accessible; it is recorded on the mark ring and you can
5910 go back to it with @kbd{C-u C-@key{SPC}}.) Meanwhile, point is
5911 located at the beginning of the inserted text, which is where it was
5912 before you called the insert function, the position of which was saved
5913 by the first @code{save-excursion}.
5916 The whole @code{let} expression looks like this:
5920 (let (start end newmark)
5924 (setq start (point-min) end (point-max)))
5925 (insert-buffer-substring buffer start end)
5926 (setq newmark (point)))
5927 (push-mark newmark))
5931 Like the @code{append-to-buffer} function, the @code{insert-buffer}
5932 function uses @code{let}, @code{save-excursion}, and
5933 @code{set-buffer}. In addition, the function illustrates one way to
5934 use @code{or}. All these functions are building blocks that we will
5935 find and use again and again.
5937 @node New insert-buffer
5938 @subsection New Body for @code{insert-buffer}
5939 @findex insert-buffer, new version body
5940 @findex new version body for insert-buffer
5942 The body in the GNU Emacs 22 version is more confusing than the original.
5945 It consists of two expressions,
5951 (insert-buffer-substring (get-buffer buffer))
5959 except, and this is what confuses novices, very important work is done
5960 inside the @code{push-mark} expression.
5962 The @code{get-buffer} function returns a buffer with the name
5963 provided. You will note that the function is @emph{not} called
5964 @code{get-buffer-create}; it does not create a buffer if one does not
5965 already exist. The buffer returned by @code{get-buffer}, an existing
5966 buffer, is passed to @code{insert-buffer-substring}, which inserts the
5967 whole of the buffer (since you did not specify anything else).
5969 The location into which the buffer is inserted is recorded by
5970 @code{push-mark}. Then the function returns @code{nil}, the value of
5971 its last command. Put another way, the @code{insert-buffer} function
5972 exists only to produce a side effect, inserting another buffer, not to
5975 @node beginning-of-buffer
5976 @section Complete Definition of @code{beginning-of-buffer}
5977 @findex beginning-of-buffer
5979 The basic structure of the @code{beginning-of-buffer} function has
5980 already been discussed. (@xref{simplified-beginning-of-buffer, , A
5981 Simplified @code{beginning-of-buffer} Definition}.)
5982 This section describes the complex part of the definition.
5984 As previously described, when invoked without an argument,
5985 @code{beginning-of-buffer} moves the cursor to the beginning of the
5986 buffer (in truth, the beginning of the accessible portion of the
5987 buffer), leaving the mark at the previous position. However, when the
5988 command is invoked with a number between one and ten, the function
5989 considers that number to be a fraction of the length of the buffer,
5990 measured in tenths, and Emacs moves the cursor that fraction of the
5991 way from the beginning of the buffer. Thus, you can either call this
5992 function with the key command @kbd{M-<}, which will move the cursor to
5993 the beginning of the buffer, or with a key command such as @kbd{C-u 7
5994 M-<} which will move the cursor to a point 70% of the way through the
5995 buffer. If a number bigger than ten is used for the argument, it
5996 moves to the end of the buffer.
5998 The @code{beginning-of-buffer} function can be called with or without an
5999 argument. The use of the argument is optional.
6002 * Optional Arguments::
6003 * beginning-of-buffer opt arg:: Example with optional argument.
6004 * beginning-of-buffer complete::
6007 @node Optional Arguments
6008 @subsection Optional Arguments
6010 Unless told otherwise, Lisp expects that a function with an argument in
6011 its function definition will be called with a value for that argument.
6012 If that does not happen, you get an error and a message that says
6013 @samp{Wrong number of arguments}.
6015 @cindex Optional arguments
6018 However, optional arguments are a feature of Lisp: a particular
6019 @dfn{keyword} is used to tell the Lisp interpreter that an argument is
6020 optional. The keyword is @code{&optional}. (The @samp{&} in front of
6021 @samp{optional} is part of the keyword.) In a function definition, if
6022 an argument follows the keyword @code{&optional}, no value need be
6023 passed to that argument when the function is called.
6026 The first line of the function definition of @code{beginning-of-buffer}
6027 therefore looks like this:
6030 (defun beginning-of-buffer (&optional arg)
6034 In outline, the whole function looks like this:
6038 (defun beginning-of-buffer (&optional arg)
6039 "@var{documentation}@dots{}"
6041 (or (@var{is-the-argument-a-cons-cell} arg)
6042 (and @var{are-both-transient-mark-mode-and-mark-active-true})
6044 (let (@var{determine-size-and-set-it})
6046 (@var{if-there-is-an-argument}
6047 @var{figure-out-where-to-go}
6054 The function is similar to the @code{simplified-beginning-of-buffer}
6055 function except that the @code{interactive} expression has @code{"P"}
6056 as an argument and the @code{goto-char} function is followed by an
6057 if-then-else expression that figures out where to put the cursor if
6058 there is an argument that is not a cons cell.
6060 (Since I do not explain a cons cell for many more chapters, please
6061 consider ignoring the function @code{consp}. @xref{List
6062 Implementation, , How Lists are Implemented}, and @ref{Cons Cell Type,
6063 , Cons Cell and List Types, elisp, The GNU Emacs Lisp Reference
6066 The @code{"P"} in the @code{interactive} expression tells Emacs to
6067 pass a prefix argument, if there is one, to the function in raw form.
6068 A prefix argument is made by typing the @key{META} key followed by a
6069 number, or by typing @kbd{C-u} and then a number. (If you don't type
6070 a number, @kbd{C-u} defaults to a cons cell with a 4. A lowercase
6071 @code{"p"} in the @code{interactive} expression causes the function to
6072 convert a prefix arg to a number.)
6074 The true-or-false-test of the @code{if} expression looks complex, but
6075 it is not: it checks whether @code{arg} has a value that is not
6076 @code{nil} and whether it is a cons cell. (That is what @code{consp}
6077 does; it checks whether its argument is a cons cell.) If @code{arg}
6078 has a value that is not @code{nil} (and is not a cons cell), which
6079 will be the case if @code{beginning-of-buffer} is called with a
6080 numeric argument, then this true-or-false-test will return true and
6081 the then-part of the @code{if} expression will be evaluated. On the
6082 other hand, if @code{beginning-of-buffer} is not called with an
6083 argument, the value of @code{arg} will be @code{nil} and the else-part
6084 of the @code{if} expression will be evaluated. The else-part is
6085 simply @code{point-min}, and when this is the outcome, the whole
6086 @code{goto-char} expression is @code{(goto-char (point-min))}, which
6087 is how we saw the @code{beginning-of-buffer} function in its
6090 @node beginning-of-buffer opt arg
6091 @subsection @code{beginning-of-buffer} with an Argument
6093 When @code{beginning-of-buffer} is called with an argument, an
6094 expression is evaluated which calculates what value to pass to
6095 @code{goto-char}. This expression is rather complicated at first sight.
6096 It includes an inner @code{if} expression and much arithmetic. It looks
6101 (if (> (buffer-size) 10000)
6102 ;; @r{Avoid overflow for large buffer sizes!}
6103 (* (prefix-numeric-value arg)
6108 size (prefix-numeric-value arg))) 10)))
6113 * Disentangle beginning-of-buffer::
6114 * Large buffer case::
6115 * Small buffer case::
6119 @node Disentangle beginning-of-buffer
6120 @unnumberedsubsubsec Disentangle @code{beginning-of-buffer}
6123 Like other complex-looking expressions, the conditional expression
6124 within @code{beginning-of-buffer} can be disentangled by looking at it
6125 as parts of a template, in this case, the template for an if-then-else
6126 expression. In skeletal form, the expression looks like this:
6130 (if (@var{buffer-is-large}
6131 @var{divide-buffer-size-by-10-and-multiply-by-arg}
6132 @var{else-use-alternate-calculation}
6136 The true-or-false-test of this inner @code{if} expression checks the
6137 size of the buffer. The reason for this is that the old version 18
6138 Emacs used numbers that are no bigger than eight million or so and in
6139 the computation that followed, the programmer feared that Emacs might
6140 try to use over-large numbers if the buffer were large. The term
6141 ``overflow'', mentioned in the comment, means numbers that are over
6142 large. More recent versions of Emacs use larger numbers, but this
6143 code has not been touched, if only because people now look at buffers
6144 that are far, far larger than ever before.
6146 There are two cases: if the buffer is large and if it is not.
6148 @node Large buffer case
6149 @unnumberedsubsubsec What happens in a large buffer
6151 In @code{beginning-of-buffer}, the inner @code{if} expression tests
6152 whether the size of the buffer is greater than 10,000 characters. To do
6153 this, it uses the @code{>} function and the computation of @code{size}
6154 that comes from the let expression.
6156 In the old days, the function @code{buffer-size} was used. Not only
6157 was that function called several times, it gave the size of the whole
6158 buffer, not the accessible part. The computation makes much more
6159 sense when it handles just the accessible part. (@xref{Narrowing &
6160 Widening, , Narrowing and Widening}, for more information on focusing
6161 attention to an ``accessible'' part.)
6164 The line looks like this:
6172 When the buffer is large, the then-part of the @code{if} expression is
6173 evaluated. It reads like this (after formatting for easy reading):
6178 (prefix-numeric-value arg)
6184 This expression is a multiplication, with two arguments to the function
6187 The first argument is @code{(prefix-numeric-value arg)}. When
6188 @code{"P"} is used as the argument for @code{interactive}, the value
6189 passed to the function as its argument is passed a ``raw prefix
6190 argument'', and not a number. (It is a number in a list.) To perform
6191 the arithmetic, a conversion is necessary, and
6192 @code{prefix-numeric-value} does the job.
6194 @findex / @r{(division)}
6196 The second argument is @code{(/ size 10)}. This expression divides
6197 the numeric value by ten---the numeric value of the size of the
6198 accessible portion of the buffer. This produces a number that tells
6199 how many characters make up one tenth of the buffer size. (In Lisp,
6200 @code{/} is used for division, just as @code{*} is used for
6204 In the multiplication expression as a whole, this amount is multiplied
6205 by the value of the prefix argument---the multiplication looks like this:
6209 (* @var{numeric-value-of-prefix-arg}
6210 @var{number-of-characters-in-one-tenth-of-the-accessible-buffer})
6215 If, for example, the prefix argument is @samp{7}, the one-tenth value
6216 will be multiplied by 7 to give a position 70% of the way through.
6219 The result of all this is that if the accessible portion of the buffer
6220 is large, the @code{goto-char} expression reads like this:
6224 (goto-char (* (prefix-numeric-value arg)
6229 This puts the cursor where we want it.
6231 @node Small buffer case
6232 @unnumberedsubsubsec What happens in a small buffer
6234 If the buffer contains fewer than 10,000 characters, a slightly
6235 different computation is performed. You might think this is not
6236 necessary, since the first computation could do the job. However, in
6237 a small buffer, the first method may not put the cursor on exactly the
6238 desired line; the second method does a better job.
6241 The code looks like this:
6243 @c Keep this on one line.
6245 (/ (+ 10 (* size (prefix-numeric-value arg))) 10))
6250 This is code in which you figure out what happens by discovering how the
6251 functions are embedded in parentheses. It is easier to read if you
6252 reformat it with each expression indented more deeply than its
6253 enclosing expression:
6261 (prefix-numeric-value arg)))
6268 Looking at parentheses, we see that the innermost operation is
6269 @code{(prefix-numeric-value arg)}, which converts the raw argument to
6270 a number. In the following expression, this number is multiplied by
6271 the size of the accessible portion of the buffer:
6274 (* size (prefix-numeric-value arg))
6278 This multiplication creates a number that may be larger than the size of
6279 the buffer---seven times larger if the argument is 7, for example. Ten
6280 is then added to this number and finally the large number is divided by
6281 ten to provide a value that is one character larger than the percentage
6282 position in the buffer.
6284 The number that results from all this is passed to @code{goto-char} and
6285 the cursor is moved to that point.
6288 @node beginning-of-buffer complete
6289 @subsection The Complete @code{beginning-of-buffer}
6292 Here is the complete text of the @code{beginning-of-buffer} function:
6298 (defun beginning-of-buffer (&optional arg)
6299 "Move point to the beginning of the buffer;
6300 leave mark at previous position.
6301 With \\[universal-argument] prefix,
6302 do not set mark at previous position.
6304 put point N/10 of the way from the beginning.
6306 If the buffer is narrowed,
6307 this command uses the beginning and size
6308 of the accessible part of the buffer.
6312 Don't use this command in Lisp programs!
6313 \(goto-char (point-min)) is faster
6314 and avoids clobbering the mark."
6317 (and transient-mark-mode mark-active)
6321 (let ((size (- (point-max) (point-min))))
6322 (goto-char (if (and arg (not (consp arg)))
6325 ;; Avoid overflow for large buffer sizes!
6326 (* (prefix-numeric-value arg)
6328 (/ (+ 10 (* size (prefix-numeric-value arg)))
6331 (if (and arg (not (consp arg))) (forward-line 1)))
6336 From before GNU Emacs 22
6339 (defun beginning-of-buffer (&optional arg)
6340 "Move point to the beginning of the buffer;
6341 leave mark at previous position.
6342 With arg N, put point N/10 of the way
6343 from the true beginning.
6346 Don't use this in Lisp programs!
6347 \(goto-char (point-min)) is faster
6348 and does not set the mark."
6355 (if (> (buffer-size) 10000)
6356 ;; @r{Avoid overflow for large buffer sizes!}
6357 (* (prefix-numeric-value arg)
6358 (/ (buffer-size) 10))
6361 (/ (+ 10 (* (buffer-size)
6362 (prefix-numeric-value arg)))
6365 (if arg (forward-line 1)))
6371 Except for two small points, the previous discussion shows how this
6372 function works. The first point deals with a detail in the
6373 documentation string, and the second point concerns the last line of
6377 In the documentation string, there is reference to an expression:
6380 \\[universal-argument]
6384 A @samp{\\} is used before the first square bracket of this
6385 expression. This @samp{\\} tells the Lisp interpreter to substitute
6386 whatever key is currently bound to the @samp{[@dots{}]}. In the case
6387 of @code{universal-argument}, that is usually @kbd{C-u}, but it might
6388 be different. (@xref{Documentation Tips, , Tips for Documentation
6389 Strings, elisp, The GNU Emacs Lisp Reference Manual}, for more
6393 Finally, the last line of the @code{beginning-of-buffer} command says
6394 to move point to the beginning of the next line if the command is
6395 invoked with an argument:
6398 (if (and arg (not (consp arg))) (forward-line 1))
6402 This puts the cursor at the beginning of the first line after the
6403 appropriate tenths position in the buffer. This is a flourish that
6404 means that the cursor is always located @emph{at least} the requested
6405 tenths of the way through the buffer, which is a nicety that is,
6406 perhaps, not necessary, but which, if it did not occur, would be sure
6407 to draw complaints. (The @code{(not (consp arg))} portion is so that
6408 if you specify the command with a @kbd{C-u}, but without a number,
6409 that is to say, if the ``raw prefix argument'' is simply a cons cell,
6410 the command does not put you at the beginning of the second line.)
6412 @node Second Buffer Related Review
6415 Here is a brief summary of some of the topics covered in this chapter.
6419 Evaluate each argument in sequence, and return the value of the first
6420 argument that is not @code{nil}; if none return a value that is not
6421 @code{nil}, return @code{nil}. In brief, return the first true value
6422 of the arguments; return a true value if one @emph{or} any of the
6426 Evaluate each argument in sequence, and if any are @code{nil}, return
6427 @code{nil}; if none are @code{nil}, return the value of the last
6428 argument. In brief, return a true value only if all the arguments are
6429 true; return a true value if one @emph{and} each of the others is
6433 A keyword used to indicate that an argument to a function definition
6434 is optional; this means that the function can be evaluated without the
6435 argument, if desired.
6437 @item prefix-numeric-value
6438 Convert the ``raw prefix argument'' produced by @code{(interactive
6439 "P")} to a numeric value.
6442 Move point forward to the beginning of the next line, or if the argument
6443 is greater than one, forward that many lines. If it can't move as far
6444 forward as it is supposed to, @code{forward-line} goes forward as far as
6445 it can and then returns a count of the number of additional lines it was
6446 supposed to move but couldn't.
6449 Delete the entire contents of the current buffer.
6452 Return @code{t} if its argument is a buffer; otherwise return @code{nil}.
6455 @node optional Exercise
6456 @section @code{optional} Argument Exercise
6458 Write an interactive function with an optional argument that tests
6459 whether its argument, a number, is greater than or equal to, or else,
6460 less than the value of @code{fill-column}, and tells you which, in a
6461 message. However, if you do not pass an argument to the function, use
6462 56 as a default value.
6464 @node Narrowing & Widening
6465 @chapter Narrowing and Widening
6466 @cindex Focusing attention (narrowing)
6470 Narrowing is a feature of Emacs that makes it possible for you to focus
6471 on a specific part of a buffer, and work without accidentally changing
6472 other parts. Narrowing is normally disabled since it can confuse
6476 * Narrowing advantages:: The advantages of narrowing
6477 * save-restriction:: The @code{save-restriction} special form.
6478 * what-line:: The number of the line that point is on.
6483 @node Narrowing advantages
6484 @unnumberedsec The Advantages of Narrowing
6487 With narrowing, the rest of a buffer is made invisible, as if it weren't
6488 there. This is an advantage if, for example, you want to replace a word
6489 in one part of a buffer but not in another: you narrow to the part you want
6490 and the replacement is carried out only in that section, not in the rest
6491 of the buffer. Searches will only work within a narrowed region, not
6492 outside of one, so if you are fixing a part of a document, you can keep
6493 yourself from accidentally finding parts you do not need to fix by
6494 narrowing just to the region you want.
6495 (The key binding for @code{narrow-to-region} is @kbd{C-x n n}.)
6497 However, narrowing does make the rest of the buffer invisible, which
6498 can scare people who inadvertently invoke narrowing and think they
6499 have deleted a part of their file. Moreover, the @code{undo} command
6500 (which is usually bound to @kbd{C-x u}) does not turn off narrowing
6501 (nor should it), so people can become quite desperate if they do not
6502 know that they can return the rest of a buffer to visibility with the
6503 @code{widen} command.
6504 (The key binding for @code{widen} is @kbd{C-x n w}.)
6506 Narrowing is just as useful to the Lisp interpreter as to a human.
6507 Often, an Emacs Lisp function is designed to work on just part of a
6508 buffer; or conversely, an Emacs Lisp function needs to work on all of a
6509 buffer that has been narrowed. The @code{what-line} function, for
6510 example, removes the narrowing from a buffer, if it has any narrowing
6511 and when it has finished its job, restores the narrowing to what it was.
6512 On the other hand, the @code{count-lines} function
6513 uses narrowing to restrict itself to just that portion
6514 of the buffer in which it is interested and then restores the previous
6517 @node save-restriction
6518 @section The @code{save-restriction} Special Form
6519 @findex save-restriction
6521 In Emacs Lisp, you can use the @code{save-restriction} special form to
6522 keep track of whatever narrowing is in effect, if any. When the Lisp
6523 interpreter meets with @code{save-restriction}, it executes the code
6524 in the body of the @code{save-restriction} expression, and then undoes
6525 any changes to narrowing that the code caused. If, for example, the
6526 buffer is narrowed and the code that follows @code{save-restriction}
6527 gets rid of the narrowing, @code{save-restriction} returns the buffer
6528 to its narrowed region afterwards. In the @code{what-line} command,
6529 any narrowing the buffer may have is undone by the @code{widen}
6530 command that immediately follows the @code{save-restriction} command.
6531 Any original narrowing is restored just before the completion of the
6535 The template for a @code{save-restriction} expression is simple:
6545 The body of the @code{save-restriction} is one or more expressions that
6546 will be evaluated in sequence by the Lisp interpreter.
6548 Finally, a point to note: when you use both @code{save-excursion} and
6549 @code{save-restriction}, one right after the other, you should use
6550 @code{save-excursion} outermost. If you write them in reverse order,
6551 you may fail to record narrowing in the buffer to which Emacs switches
6552 after calling @code{save-excursion}. Thus, when written together,
6553 @code{save-excursion} and @code{save-restriction} should be written
6564 In other circumstances, when not written together, the
6565 @code{save-excursion} and @code{save-restriction} special forms must
6566 be written in the order appropriate to the function.
6582 /usr/local/src/emacs/lisp/simple.el
6585 "Print the current buffer line number and narrowed line number of point."
6587 (let ((start (point-min))
6588 (n (line-number-at-pos)))
6590 (message "Line %d" n)
6594 (message "line %d (narrowed line %d)"
6595 (+ n (line-number-at-pos start) -1) n))))))
6597 (defun line-number-at-pos (&optional pos)
6598 "Return (narrowed) buffer line number at position POS.
6599 If POS is nil, use current buffer location.
6600 Counting starts at (point-min), so the value refers
6601 to the contents of the accessible portion of the buffer."
6602 (let ((opoint (or pos (point))) start)
6604 (goto-char (point-min))
6605 (setq start (point))
6608 (1+ (count-lines start (point))))))
6610 (defun count-lines (start end)
6611 "Return number of lines between START and END.
6612 This is usually the number of newlines between them,
6613 but can be one more if START is not equal to END
6614 and the greater of them is not at the start of a line."
6617 (narrow-to-region start end)
6618 (goto-char (point-min))
6619 (if (eq selective-display t)
6622 (while (re-search-forward "[\n\C-m]" nil t 40)
6623 (setq done (+ 40 done)))
6624 (while (re-search-forward "[\n\C-m]" nil t 1)
6625 (setq done (+ 1 done)))
6626 (goto-char (point-max))
6627 (if (and (/= start end)
6631 (- (buffer-size) (forward-line (buffer-size)))))))
6635 @section @code{what-line}
6637 @cindex Widening, example of
6639 The @code{what-line} command tells you the number of the line in which
6640 the cursor is located. The function illustrates the use of the
6641 @code{save-restriction} and @code{save-excursion} commands. Here is the
6642 original text of the function:
6647 "Print the current line number (in the buffer) of point."
6654 (1+ (count-lines 1 (point)))))))
6658 (In recent versions of GNU Emacs, the @code{what-line} function has
6659 been expanded to tell you your line number in a narrowed buffer as
6660 well as your line number in a widened buffer. The recent version is
6661 more complex than the version shown here. If you feel adventurous,
6662 you might want to look at it after figuring out how this version
6663 works. You will probably need to use @kbd{C-h f}
6664 (@code{describe-function}). The newer version uses a conditional to
6665 determine whether the buffer has been narrowed.
6667 (Also, it uses @code{line-number-at-pos}, which among other simple
6668 expressions, such as @code{(goto-char (point-min))}, moves point to
6669 the beginning of the current line with @code{(forward-line 0)} rather
6670 than @code{beginning-of-line}.)
6672 The @code{what-line} function as shown here has a documentation line
6673 and is interactive, as you would expect. The next two lines use the
6674 functions @code{save-restriction} and @code{widen}.
6676 The @code{save-restriction} special form notes whatever narrowing is in
6677 effect, if any, in the current buffer and restores that narrowing after
6678 the code in the body of the @code{save-restriction} has been evaluated.
6680 The @code{save-restriction} special form is followed by @code{widen}.
6681 This function undoes any narrowing the current buffer may have had
6682 when @code{what-line} was called. (The narrowing that was there is
6683 the narrowing that @code{save-restriction} remembers.) This widening
6684 makes it possible for the line counting commands to count from the
6685 beginning of the buffer. Otherwise, they would have been limited to
6686 counting within the accessible region. Any original narrowing is
6687 restored just before the completion of the function by the
6688 @code{save-restriction} special form.
6690 The call to @code{widen} is followed by @code{save-excursion}, which
6691 saves the location of the cursor (i.e., of point), and
6692 restores it after the code in the body of the @code{save-excursion}
6693 uses the @code{beginning-of-line} function to move point.
6695 (Note that the @code{(widen)} expression comes between the
6696 @code{save-restriction} and @code{save-excursion} special forms. When
6697 you write the two @code{save- @dots{}} expressions in sequence, write
6698 @code{save-excursion} outermost.)
6701 The last two lines of the @code{what-line} function are functions to
6702 count the number of lines in the buffer and then print the number in the
6708 (1+ (count-lines 1 (point)))))))
6712 The @code{message} function prints a one-line message at the bottom of
6713 the Emacs screen. The first argument is inside of quotation marks and
6714 is printed as a string of characters. However, it may contain a
6715 @samp{%d} expression to print a following argument. @samp{%d} prints
6716 the argument as a decimal, so the message will say something such as
6720 The number that is printed in place of the @samp{%d} is computed by the
6721 last line of the function:
6724 (1+ (count-lines 1 (point)))
6730 (defun count-lines (start end)
6731 "Return number of lines between START and END.
6732 This is usually the number of newlines between them,
6733 but can be one more if START is not equal to END
6734 and the greater of them is not at the start of a line."
6737 (narrow-to-region start end)
6738 (goto-char (point-min))
6739 (if (eq selective-display t)
6742 (while (re-search-forward "[\n\C-m]" nil t 40)
6743 (setq done (+ 40 done)))
6744 (while (re-search-forward "[\n\C-m]" nil t 1)
6745 (setq done (+ 1 done)))
6746 (goto-char (point-max))
6747 (if (and (/= start end)
6751 (- (buffer-size) (forward-line (buffer-size)))))))
6755 What this does is count the lines from the first position of the
6756 buffer, indicated by the @code{1}, up to @code{(point)}, and then add
6757 one to that number. (The @code{1+} function adds one to its
6758 argument.) We add one to it because line 2 has only one line before
6759 it, and @code{count-lines} counts only the lines @emph{before} the
6762 After @code{count-lines} has done its job, and the message has been
6763 printed in the echo area, the @code{save-excursion} restores point to
6764 its original position; and @code{save-restriction} restores
6765 the original narrowing, if any.
6767 @node narrow Exercise
6768 @section Exercise with Narrowing
6770 Write a function that will display the first 60 characters of the
6771 current buffer, even if you have narrowed the buffer to its latter
6772 half so that the first line is inaccessible. Restore point, mark, and
6773 narrowing. For this exercise, you need to use a whole potpourri of
6774 functions, including @code{save-restriction}, @code{widen},
6775 @code{goto-char}, @code{point-min}, @code{message}, and
6776 @code{buffer-substring}.
6778 @cindex Properties, mention of @code{buffer-substring-no-properties}
6779 (@code{buffer-substring} is a previously unmentioned function you will
6780 have to investigate yourself; or perhaps you will have to use
6781 @code{buffer-substring-no-properties} or
6782 @code{filter-buffer-substring} @dots{}, yet other functions. Text
6783 properties are a feature otherwise not discussed here. @xref{Text
6784 Properties, , Text Properties, elisp, The GNU Emacs Lisp Reference
6787 Additionally, do you really need @code{goto-char} or @code{point-min}?
6788 Or can you write the function without them?
6790 @node car cdr & cons
6791 @chapter @code{car}, @code{cdr}, @code{cons}: Fundamental Functions
6792 @findex car, @r{introduced}
6793 @findex cdr, @r{introduced}
6795 In Lisp, @code{car}, @code{cdr}, and @code{cons} are fundamental
6796 functions. The @code{cons} function is used to construct lists, and
6797 the @code{car} and @code{cdr} functions are used to take them apart.
6799 In the walk through of the @code{copy-region-as-kill} function, we
6800 will see @code{cons} as well as two variants on @code{cdr},
6801 namely, @code{setcdr} and @code{nthcdr}. (@xref{copy-region-as-kill}.)
6804 * Strange Names:: An historical aside: why the strange names?
6805 * car & cdr:: Functions for extracting part of a list.
6806 * cons:: Constructing a list.
6807 * nthcdr:: Calling @code{cdr} repeatedly.
6809 * setcar:: Changing the first element of a list.
6810 * setcdr:: Changing the rest of a list.
6816 @unnumberedsec Strange Names
6819 The name of the @code{cons} function is not unreasonable: it is an
6820 abbreviation of the word ``construct''. The origins of the names for
6821 @code{car} and @code{cdr}, on the other hand, are esoteric: @code{car}
6822 is an acronym from the phrase ``Contents of the Address part of the
6823 Register''; and @code{cdr} (pronounced ``could-er'') is an acronym from
6824 the phrase ``Contents of the Decrement part of the Register''. These
6825 phrases refer to specific pieces of hardware on the very early
6826 computer on which the original Lisp was developed. Besides being
6827 obsolete, the phrases have been completely irrelevant for more than 25
6828 years to anyone thinking about Lisp. Nonetheless, although a few
6829 brave scholars have begun to use more reasonable names for these
6830 functions, the old terms are still in use. In particular, since the
6831 terms are used in the Emacs Lisp source code, we will use them in this
6835 @section @code{car} and @code{cdr}
6837 The @sc{car} of a list is, quite simply, the first item in the list.
6838 Thus the @sc{car} of the list @code{(rose violet daisy buttercup)} is
6842 If you are reading this in Info in GNU Emacs, you can see this by
6843 evaluating the following:
6846 (car '(rose violet daisy buttercup))
6850 After evaluating the expression, @code{rose} will appear in the echo
6853 Clearly, a more reasonable name for the @code{car} function would be
6854 @code{first} and this is often suggested.
6856 @code{car} does not remove the first item from the list; it only reports
6857 what it is. After @code{car} has been applied to a list, the list is
6858 still the same as it was. In the jargon, @code{car} is
6859 ``non-destructive''. This feature turns out to be important.
6861 The @sc{cdr} of a list is the rest of the list, that is, the
6862 @code{cdr} function returns the part of the list that follows the
6863 first item. Thus, while the @sc{car} of the list @code{'(rose violet
6864 daisy buttercup)} is @code{rose}, the rest of the list, the value
6865 returned by the @code{cdr} function, is @code{(violet daisy
6869 You can see this by evaluating the following in the usual way:
6872 (cdr '(rose violet daisy buttercup))
6876 When you evaluate this, @code{(violet daisy buttercup)} will appear in
6879 Like @code{car}, @code{cdr} does not remove any elements from the
6880 list---it just returns a report of what the second and subsequent
6883 Incidentally, in the example, the list of flowers is quoted. If it were
6884 not, the Lisp interpreter would try to evaluate the list by calling
6885 @code{rose} as a function. In this example, we do not want to do that.
6887 Clearly, a more reasonable name for @code{cdr} would be @code{rest}.
6889 (There is a lesson here: when you name new functions, consider very
6890 carefully what you are doing, since you may be stuck with the names
6891 for far longer than you expect. The reason this document perpetuates
6892 these names is that the Emacs Lisp source code uses them, and if I did
6893 not use them, you would have a hard time reading the code; but do,
6894 please, try to avoid using these terms yourself. The people who come
6895 after you will be grateful to you.)
6897 When @code{car} and @code{cdr} are applied to a list made up of symbols,
6898 such as the list @code{(pine fir oak maple)}, the element of the list
6899 returned by the function @code{car} is the symbol @code{pine} without
6900 any parentheses around it. @code{pine} is the first element in the
6901 list. However, the @sc{cdr} of the list is a list itself, @code{(fir
6902 oak maple)}, as you can see by evaluating the following expressions in
6907 (car '(pine fir oak maple))
6909 (cdr '(pine fir oak maple))
6913 On the other hand, in a list of lists, the first element is itself a
6914 list. @code{car} returns this first element as a list. For example,
6915 the following list contains three sub-lists, a list of carnivores, a
6916 list of herbivores and a list of sea mammals:
6920 (car '((lion tiger cheetah)
6921 (gazelle antelope zebra)
6922 (whale dolphin seal)))
6927 In this example, the first element or @sc{car} of the list is the list of
6928 carnivores, @code{(lion tiger cheetah)}, and the rest of the list is
6929 @code{((gazelle antelope zebra) (whale dolphin seal))}.
6933 (cdr '((lion tiger cheetah)
6934 (gazelle antelope zebra)
6935 (whale dolphin seal)))
6939 It is worth saying again that @code{car} and @code{cdr} are
6940 non-destructive---that is, they do not modify or change lists to which
6941 they are applied. This is very important for how they are used.
6943 Also, in the first chapter, in the discussion about atoms, I said that
6944 in Lisp, ``certain kinds of atom, such as an array, can be separated
6945 into parts; but the mechanism for doing this is different from the
6946 mechanism for splitting a list. As far as Lisp is concerned, the
6947 atoms of a list are unsplittable.'' (@xref{Lisp Atoms}.) The
6948 @code{car} and @code{cdr} functions are used for splitting lists and
6949 are considered fundamental to Lisp. Since they cannot split or gain
6950 access to the parts of an array, an array is considered an atom.
6951 Conversely, the other fundamental function, @code{cons}, can put
6952 together or construct a list, but not an array. (Arrays are handled
6953 by array-specific functions. @xref{Arrays, , Arrays, elisp, The GNU
6954 Emacs Lisp Reference Manual}.)
6957 @section @code{cons}
6958 @findex cons, @r{introduced}
6960 The @code{cons} function constructs lists; it is the inverse of
6961 @code{car} and @code{cdr}. For example, @code{cons} can be used to make
6962 a four element list from the three element list, @code{(fir oak maple)}:
6965 (cons 'pine '(fir oak maple))
6970 After evaluating this list, you will see
6973 (pine fir oak maple)
6977 appear in the echo area. @code{cons} causes the creation of a new
6978 list in which the element is followed by the elements of the original
6981 We often say that ``@code{cons} puts a new element at the beginning of
6982 a list; it attaches or pushes elements onto the list'', but this
6983 phrasing can be misleading, since @code{cons} does not change an
6984 existing list, but creates a new one.
6986 Like @code{car} and @code{cdr}, @code{cons} is non-destructive.
6990 * length:: How to find the length of a list.
6995 @unnumberedsubsec Build a list
6998 @code{cons} must have a list to attach to.@footnote{Actually, you can
6999 @code{cons} an element to an atom to produce a dotted pair. Dotted
7000 pairs are not discussed here; see @ref{Dotted Pair Notation, , Dotted
7001 Pair Notation, elisp, The GNU Emacs Lisp Reference Manual}.} You
7002 cannot start from absolutely nothing. If you are building a list, you
7003 need to provide at least an empty list at the beginning. Here is a
7004 series of @code{cons} expressions that build up a list of flowers. If
7005 you are reading this in Info in GNU Emacs, you can evaluate each of
7006 the expressions in the usual way; the value is printed in this text
7007 after @samp{@result{}}, which you may read as ``evaluates to''.
7011 (cons 'buttercup ())
7012 @result{} (buttercup)
7016 (cons 'daisy '(buttercup))
7017 @result{} (daisy buttercup)
7021 (cons 'violet '(daisy buttercup))
7022 @result{} (violet daisy buttercup)
7026 (cons 'rose '(violet daisy buttercup))
7027 @result{} (rose violet daisy buttercup)
7032 In the first example, the empty list is shown as @code{()} and a list
7033 made up of @code{buttercup} followed by the empty list is constructed.
7034 As you can see, the empty list is not shown in the list that was
7035 constructed. All that you see is @code{(buttercup)}. The empty list is
7036 not counted as an element of a list because there is nothing in an empty
7037 list. Generally speaking, an empty list is invisible.
7039 The second example, @code{(cons 'daisy '(buttercup))} constructs a new,
7040 two element list by putting @code{daisy} in front of @code{buttercup};
7041 and the third example constructs a three element list by putting
7042 @code{violet} in front of @code{daisy} and @code{buttercup}.
7045 @subsection Find the Length of a List: @code{length}
7048 You can find out how many elements there are in a list by using the Lisp
7049 function @code{length}, as in the following examples:
7053 (length '(buttercup))
7058 (length '(daisy buttercup))
7063 (length (cons 'violet '(daisy buttercup)))
7069 In the third example, the @code{cons} function is used to construct a
7070 three element list which is then passed to the @code{length} function as
7074 We can also use @code{length} to count the number of elements in an
7085 As you would expect, the number of elements in an empty list is zero.
7087 An interesting experiment is to find out what happens if you try to find
7088 the length of no list at all; that is, if you try to call @code{length}
7089 without giving it an argument, not even an empty list:
7097 What you see, if you evaluate this, is the error message
7100 Lisp error: (wrong-number-of-arguments length 0)
7104 This means that the function receives the wrong number of
7105 arguments, zero, when it expects some other number of arguments. In
7106 this case, one argument is expected, the argument being a list whose
7107 length the function is measuring. (Note that @emph{one} list is
7108 @emph{one} argument, even if the list has many elements inside it.)
7110 The part of the error message that says @samp{length} is the name of
7114 @code{length} is still a subroutine, but you need C-h f to discover that.
7116 In an earlier version:
7117 This is written with a special notation, @samp{#<subr},
7118 that indicates that the function @code{length} is one of the primitive
7119 functions written in C rather than in Emacs Lisp. (@samp{subr} is an
7120 abbreviation for ``subroutine''.) @xref{What Is a Function, , What Is a
7121 Function?, elisp , The GNU Emacs Lisp Reference Manual}, for more
7126 @section @code{nthcdr}
7129 The @code{nthcdr} function is associated with the @code{cdr} function.
7130 What it does is take the @sc{cdr} of a list repeatedly.
7132 If you take the @sc{cdr} of the list @code{(pine fir
7133 oak maple)}, you will be returned the list @code{(fir oak maple)}. If you
7134 repeat this on what was returned, you will be returned the list
7135 @code{(oak maple)}. (Of course, repeated @sc{cdr}ing on the original
7136 list will just give you the original @sc{cdr} since the function does
7137 not change the list. You need to evaluate the @sc{cdr} of the
7138 @sc{cdr} and so on.) If you continue this, eventually you will be
7139 returned an empty list, which in this case, instead of being shown as
7140 @code{()} is shown as @code{nil}.
7143 For review, here is a series of repeated @sc{cdr}s, the text following
7144 the @samp{@result{}} shows what is returned.
7148 (cdr '(pine fir oak maple))
7149 @result{}(fir oak maple)
7153 (cdr '(fir oak maple))
7154 @result{} (oak maple)
7179 You can also do several @sc{cdr}s without printing the values in
7184 (cdr (cdr '(pine fir oak maple)))
7185 @result{} (oak maple)
7190 In this example, the Lisp interpreter evaluates the innermost list first.
7191 The innermost list is quoted, so it just passes the list as it is to the
7192 innermost @code{cdr}. This @code{cdr} passes a list made up of the
7193 second and subsequent elements of the list to the outermost @code{cdr},
7194 which produces a list composed of the third and subsequent elements of
7195 the original list. In this example, the @code{cdr} function is repeated
7196 and returns a list that consists of the original list without its
7199 The @code{nthcdr} function does the same as repeating the call to
7200 @code{cdr}. In the following example, the argument 2 is passed to the
7201 function @code{nthcdr}, along with the list, and the value returned is
7202 the list without its first two items, which is exactly the same
7203 as repeating @code{cdr} twice on the list:
7207 (nthcdr 2 '(pine fir oak maple))
7208 @result{} (oak maple)
7213 Using the original four element list, we can see what happens when
7214 various numeric arguments are passed to @code{nthcdr}, including 0, 1,
7219 ;; @r{Leave the list as it was.}
7220 (nthcdr 0 '(pine fir oak maple))
7221 @result{} (pine fir oak maple)
7225 ;; @r{Return a copy without the first element.}
7226 (nthcdr 1 '(pine fir oak maple))
7227 @result{} (fir oak maple)
7231 ;; @r{Return a copy of the list without three elements.}
7232 (nthcdr 3 '(pine fir oak maple))
7237 ;; @r{Return a copy lacking all four elements.}
7238 (nthcdr 4 '(pine fir oak maple))
7243 ;; @r{Return a copy lacking all elements.}
7244 (nthcdr 5 '(pine fir oak maple))
7253 The @code{nthcdr} function takes the @sc{cdr} of a list repeatedly.
7254 The @code{nth} function takes the @sc{car} of the result returned by
7255 @code{nthcdr}. It returns the Nth element of the list.
7258 Thus, if it were not defined in C for speed, the definition of
7259 @code{nth} would be:
7264 "Returns the Nth element of LIST.
7265 N counts from zero. If LIST is not that long, nil is returned."
7266 (car (nthcdr n list)))
7271 (Originally, @code{nth} was defined in Emacs Lisp in @file{subr.el},
7272 but its definition was redone in C in the 1980s.)
7274 The @code{nth} function returns a single element of a list.
7275 This can be very convenient.
7277 Note that the elements are numbered from zero, not one. That is to
7278 say, the first element of a list, its @sc{car} is the zeroth element.
7279 This is called ``zero-based'' counting and often bothers people who
7280 are accustomed to the first element in a list being number one, which
7288 (nth 0 '("one" "two" "three"))
7291 (nth 1 '("one" "two" "three"))
7296 It is worth mentioning that @code{nth}, like @code{nthcdr} and
7297 @code{cdr}, does not change the original list---the function is
7298 non-destructive. This is in sharp contrast to the @code{setcar} and
7299 @code{setcdr} functions.
7302 @section @code{setcar}
7305 As you might guess from their names, the @code{setcar} and @code{setcdr}
7306 functions set the @sc{car} or the @sc{cdr} of a list to a new value.
7307 They actually change the original list, unlike @code{car} and @code{cdr}
7308 which leave the original list as it was. One way to find out how this
7309 works is to experiment. We will start with the @code{setcar} function.
7312 First, we can make a list and then set the value of a variable to the
7313 list, using the @code{setq} function. Here is a list of animals:
7316 (setq animals '(antelope giraffe lion tiger))
7320 If you are reading this in Info inside of GNU Emacs, you can evaluate
7321 this expression in the usual fashion, by positioning the cursor after
7322 the expression and typing @kbd{C-x C-e}. (I'm doing this right here
7323 as I write this. This is one of the advantages of having the
7324 interpreter built into the computing environment. Incidentally, when
7325 there is nothing on the line after the final parentheses, such as a
7326 comment, point can be on the next line. Thus, if your cursor is in
7327 the first column of the next line, you do not need to move it.
7328 Indeed, Emacs permits any amount of white space after the final
7332 When we evaluate the variable @code{animals}, we see that it is bound to
7333 the list @code{(antelope giraffe lion tiger)}:
7338 @result{} (antelope giraffe lion tiger)
7343 Put another way, the variable @code{animals} points to the list
7344 @code{(antelope giraffe lion tiger)}.
7346 Next, evaluate the function @code{setcar} while passing it two
7347 arguments, the variable @code{animals} and the quoted symbol
7348 @code{hippopotamus}; this is done by writing the three element list
7349 @code{(setcar animals 'hippopotamus)} and then evaluating it in the
7353 (setcar animals 'hippopotamus)
7358 After evaluating this expression, evaluate the variable @code{animals}
7359 again. You will see that the list of animals has changed:
7364 @result{} (hippopotamus giraffe lion tiger)
7369 The first element on the list, @code{antelope} is replaced by
7370 @code{hippopotamus}.
7372 So we can see that @code{setcar} did not add a new element to the list
7373 as @code{cons} would have; it replaced @code{antelope} with
7374 @code{hippopotamus}; it @emph{changed} the list.
7377 @section @code{setcdr}
7380 The @code{setcdr} function is similar to the @code{setcar} function,
7381 except that the function replaces the second and subsequent elements of
7382 a list rather than the first element.
7384 (To see how to change the last element of a list, look ahead to
7385 @ref{kill-new function, , The @code{kill-new} function}, which uses
7386 the @code{nthcdr} and @code{setcdr} functions.)
7389 To see how this works, set the value of the variable to a list of
7390 domesticated animals by evaluating the following expression:
7393 (setq domesticated-animals '(horse cow sheep goat))
7398 If you now evaluate the list, you will be returned the list
7399 @code{(horse cow sheep goat)}:
7403 domesticated-animals
7404 @result{} (horse cow sheep goat)
7409 Next, evaluate @code{setcdr} with two arguments, the name of the
7410 variable which has a list as its value, and the list to which the
7411 @sc{cdr} of the first list will be set;
7414 (setcdr domesticated-animals '(cat dog))
7418 If you evaluate this expression, the list @code{(cat dog)} will appear
7419 in the echo area. This is the value returned by the function. The
7420 result we are interested in is the ``side effect'', which we can see by
7421 evaluating the variable @code{domesticated-animals}:
7425 domesticated-animals
7426 @result{} (horse cat dog)
7431 Indeed, the list is changed from @code{(horse cow sheep goat)} to
7432 @code{(horse cat dog)}. The @sc{cdr} of the list is changed from
7433 @code{(cow sheep goat)} to @code{(cat dog)}.
7438 Construct a list of four birds by evaluating several expressions with
7439 @code{cons}. Find out what happens when you @code{cons} a list onto
7440 itself. Replace the first element of the list of four birds with a
7441 fish. Replace the rest of that list with a list of other fish.
7443 @node Cutting & Storing Text
7444 @chapter Cutting and Storing Text
7445 @cindex Cutting and storing text
7446 @cindex Storing and cutting text
7447 @cindex Killing text
7448 @cindex Clipping text
7449 @cindex Erasing text
7450 @cindex Deleting text
7452 Whenever you cut or clip text out of a buffer with a ``kill'' command in
7453 GNU Emacs, it is stored in a list and you can bring it back with a
7456 (The use of the word ``kill'' in Emacs for processes which specifically
7457 @emph{do not} destroy the values of the entities is an unfortunate
7458 historical accident. A much more appropriate word would be ``clip'' since
7459 that is what the kill commands do; they clip text out of a buffer and
7460 put it into storage from which it can be brought back. I have often
7461 been tempted to replace globally all occurrences of ``kill'' in the Emacs
7462 sources with ``clip'' and all occurrences of ``killed'' with ``clipped''.)
7465 * Storing Text:: Text is stored in a list.
7466 * zap-to-char:: Cutting out text up to a character.
7467 * kill-region:: Cutting text out of a region.
7468 * copy-region-as-kill:: A definition for copying text.
7469 * Digression into C:: Minor note on C programming language macros.
7470 * defvar:: How to give a variable an initial value.
7471 * cons & search-fwd Review::
7472 * search Exercises::
7477 @unnumberedsec Storing Text in a List
7480 When text is cut out of a buffer, it is stored on a list. Successive
7481 pieces of text are stored on the list successively, so the list might
7485 ("a piece of text" "previous piece")
7490 The function @code{cons} can be used to create a new list from a piece
7491 of text (an ``atom'', to use the jargon) and an existing list, like
7496 (cons "another piece"
7497 '("a piece of text" "previous piece"))
7503 If you evaluate this expression, a list of three elements will appear in
7507 ("another piece" "a piece of text" "previous piece")
7510 With the @code{car} and @code{nthcdr} functions, you can retrieve
7511 whichever piece of text you want. For example, in the following code,
7512 @code{nthcdr 1 @dots{}} returns the list with the first item removed;
7513 and the @code{car} returns the first element of that remainder---the
7514 second element of the original list:
7518 (car (nthcdr 1 '("another piece"
7521 @result{} "a piece of text"
7525 The actual functions in Emacs are more complex than this, of course.
7526 The code for cutting and retrieving text has to be written so that
7527 Emacs can figure out which element in the list you want---the first,
7528 second, third, or whatever. In addition, when you get to the end of
7529 the list, Emacs should give you the first element of the list, rather
7530 than nothing at all.
7532 The list that holds the pieces of text is called the @dfn{kill ring}.
7533 This chapter leads up to a description of the kill ring and how it is
7534 used by first tracing how the @code{zap-to-char} function works. This
7535 function uses (or ``calls'') a function that invokes a function that
7536 manipulates the kill ring. Thus, before reaching the mountains, we
7537 climb the foothills.
7539 A subsequent chapter describes how text that is cut from the buffer is
7540 retrieved. @xref{Yanking, , Yanking Text Back}.
7543 @section @code{zap-to-char}
7546 Let us look at the interactive @code{zap-to-char} function.
7549 * Complete zap-to-char:: The complete implementation.
7550 * zap-to-char interactive:: A three part interactive expression.
7551 * zap-to-char body:: A short overview.
7552 * search-forward:: How to search for a string.
7553 * progn:: The @code{progn} special form.
7554 * Summing up zap-to-char:: Using @code{point} and @code{search-forward}.
7558 @node Complete zap-to-char
7559 @unnumberedsubsec The Complete @code{zap-to-char} Implementation
7562 The @code{zap-to-char} function removes the text in the region between
7563 the location of the cursor (i.e., of point) up to and including the
7564 next occurrence of a specified character. The text that
7565 @code{zap-to-char} removes is put in the kill ring; and it can be
7566 retrieved from the kill ring by typing @kbd{C-y} (@code{yank}). If
7567 the command is given an argument, it removes text through that number
7568 of occurrences. Thus, if the cursor were at the beginning of this
7569 sentence and the character were @samp{s}, @samp{Thus} would be
7570 removed. If the argument were two, @samp{Thus, if the curs} would be
7571 removed, up to and including the @samp{s} in @samp{cursor}.
7573 If the specified character is not found, @code{zap-to-char} will say
7574 ``Search failed'', tell you the character you typed, and not remove
7577 In order to determine how much text to remove, @code{zap-to-char} uses
7578 a search function. Searches are used extensively in code that
7579 manipulates text, and we will focus attention on them as well as on the
7583 @c GNU Emacs version 19
7584 (defun zap-to-char (arg char) ; version 19 implementation
7585 "Kill up to and including ARG'th occurrence of CHAR.
7586 Goes backward if ARG is negative; error if CHAR not found."
7587 (interactive "*p\ncZap to char: ")
7588 (kill-region (point)
7591 (char-to-string char) nil nil arg)
7596 Here is the complete text of the version 22 implementation of the function:
7601 (defun zap-to-char (arg char)
7602 "Kill up to and including ARG'th occurrence of CHAR.
7603 Case is ignored if `case-fold-search' is non-nil in the current buffer.
7604 Goes backward if ARG is negative; error if CHAR not found."
7605 (interactive "p\ncZap to char: ")
7606 (if (char-table-p translation-table-for-input)
7607 (setq char (or (aref translation-table-for-input char) char)))
7608 (kill-region (point) (progn
7609 (search-forward (char-to-string char)
7615 The documentation is thorough. You do need to know the jargon meaning
7616 of the word ``kill''.
7618 @node zap-to-char interactive
7619 @subsection The @code{interactive} Expression
7622 The interactive expression in the @code{zap-to-char} command looks like
7626 (interactive "p\ncZap to char: ")
7629 The part within quotation marks, @code{"p\ncZap to char:@: "}, specifies
7630 two different things. First, and most simply, is the @samp{p}.
7631 This part is separated from the next part by a newline, @samp{\n}.
7632 The @samp{p} means that the first argument to the function will be
7633 passed the value of a ``processed prefix''. The prefix argument is
7634 passed by typing @kbd{C-u} and a number, or @kbd{M-} and a number. If
7635 the function is called interactively without a prefix, 1 is passed to
7638 The second part of @code{"p\ncZap to char:@: "} is
7639 @samp{cZap to char:@: }. In this part, the lower case @samp{c}
7640 indicates that @code{interactive} expects a prompt and that the
7641 argument will be a character. The prompt follows the @samp{c} and is
7642 the string @samp{Zap to char:@: } (with a space after the colon to
7645 What all this does is prepare the arguments to @code{zap-to-char} so they
7646 are of the right type, and give the user a prompt.
7648 In a read-only buffer, the @code{zap-to-char} function copies the text
7649 to the kill ring, but does not remove it. The echo area displays a
7650 message saying that the buffer is read-only. Also, the terminal may
7651 beep or blink at you.
7653 @node zap-to-char body
7654 @subsection The Body of @code{zap-to-char}
7656 The body of the @code{zap-to-char} function contains the code that
7657 kills (that is, removes) the text in the region from the current
7658 position of the cursor up to and including the specified character.
7660 The first part of the code looks like this:
7663 (if (char-table-p translation-table-for-input)
7664 (setq char (or (aref translation-table-for-input char) char)))
7665 (kill-region (point) (progn
7666 (search-forward (char-to-string char) nil nil arg)
7671 @code{char-table-p} is an hitherto unseen function. It determines
7672 whether its argument is a character table. When it is, it sets the
7673 character passed to @code{zap-to-char} to one of them, if that
7674 character exists, or to the character itself. (This becomes important
7675 for certain characters in non-European languages. The @code{aref}
7676 function extracts an element from an array. It is an array-specific
7677 function that is not described in this document. @xref{Arrays, ,
7678 Arrays, elisp, The GNU Emacs Lisp Reference Manual}.)
7681 @code{(point)} is the current position of the cursor.
7683 The next part of the code is an expression using @code{progn}. The body
7684 of the @code{progn} consists of calls to @code{search-forward} and
7687 It is easier to understand how @code{progn} works after learning about
7688 @code{search-forward}, so we will look at @code{search-forward} and
7689 then at @code{progn}.
7691 @node search-forward
7692 @subsection The @code{search-forward} Function
7693 @findex search-forward
7695 The @code{search-forward} function is used to locate the
7696 zapped-for-character in @code{zap-to-char}. If the search is
7697 successful, @code{search-forward} leaves point immediately after the
7698 last character in the target string. (In @code{zap-to-char}, the
7699 target string is just one character long. @code{zap-to-char} uses the
7700 function @code{char-to-string} to ensure that the computer treats that
7701 character as a string.) If the search is backwards,
7702 @code{search-forward} leaves point just before the first character in
7703 the target. Also, @code{search-forward} returns @code{t} for true.
7704 (Moving point is therefore a ``side effect''.)
7707 In @code{zap-to-char}, the @code{search-forward} function looks like this:
7710 (search-forward (char-to-string char) nil nil arg)
7713 The @code{search-forward} function takes four arguments:
7717 The first argument is the target, what is searched for. This must be a
7718 string, such as @samp{"z"}.
7720 As it happens, the argument passed to @code{zap-to-char} is a single
7721 character. Because of the way computers are built, the Lisp
7722 interpreter may treat a single character as being different from a
7723 string of characters. Inside the computer, a single character has a
7724 different electronic format than a string of one character. (A single
7725 character can often be recorded in the computer using exactly one
7726 byte; but a string may be longer, and the computer needs to be ready
7727 for this.) Since the @code{search-forward} function searches for a
7728 string, the character that the @code{zap-to-char} function receives as
7729 its argument must be converted inside the computer from one format to
7730 the other; otherwise the @code{search-forward} function will fail.
7731 The @code{char-to-string} function is used to make this conversion.
7734 The second argument bounds the search; it is specified as a position in
7735 the buffer. In this case, the search can go to the end of the buffer,
7736 so no bound is set and the second argument is @code{nil}.
7739 The third argument tells the function what it should do if the search
7740 fails---it can signal an error (and print a message) or it can return
7741 @code{nil}. A @code{nil} as the third argument causes the function to
7742 signal an error when the search fails.
7745 The fourth argument to @code{search-forward} is the repeat count---how
7746 many occurrences of the string to look for. This argument is optional
7747 and if the function is called without a repeat count, this argument is
7748 passed the value 1. If this argument is negative, the search goes
7753 In template form, a @code{search-forward} expression looks like this:
7757 (search-forward "@var{target-string}"
7758 @var{limit-of-search}
7759 @var{what-to-do-if-search-fails}
7764 We will look at @code{progn} next.
7767 @subsection The @code{progn} Special Form
7770 @code{progn} is a special form that causes each of its arguments to be
7771 evaluated in sequence and then returns the value of the last one. The
7772 preceding expressions are evaluated only for the side effects they
7773 perform. The values produced by them are discarded.
7776 The template for a @code{progn} expression is very simple:
7785 In @code{zap-to-char}, the @code{progn} expression has to do two things:
7786 put point in exactly the right position; and return the location of
7787 point so that @code{kill-region} will know how far to kill to.
7789 The first argument to the @code{progn} is @code{search-forward}. When
7790 @code{search-forward} finds the string, the function leaves point
7791 immediately after the last character in the target string. (In this
7792 case the target string is just one character long.) If the search is
7793 backwards, @code{search-forward} leaves point just before the first
7794 character in the target. The movement of point is a side effect.
7796 The second and last argument to @code{progn} is the expression
7797 @code{(point)}. This expression returns the value of point, which in
7798 this case will be the location to which it has been moved by
7799 @code{search-forward}. (In the source, a line that tells the function
7800 to go to the previous character, if it is going forward, was commented
7801 out in 1999; I don't remember whether that feature or mis-feature was
7802 ever a part of the distributed source.) The value of @code{point} is
7803 returned by the @code{progn} expression and is passed to
7804 @code{kill-region} as @code{kill-region}'s second argument.
7806 @node Summing up zap-to-char
7807 @subsection Summing up @code{zap-to-char}
7809 Now that we have seen how @code{search-forward} and @code{progn} work,
7810 we can see how the @code{zap-to-char} function works as a whole.
7812 The first argument to @code{kill-region} is the position of the cursor
7813 when the @code{zap-to-char} command is given---the value of point at
7814 that time. Within the @code{progn}, the search function then moves
7815 point to just after the zapped-to-character and @code{point} returns the
7816 value of this location. The @code{kill-region} function puts together
7817 these two values of point, the first one as the beginning of the region
7818 and the second one as the end of the region, and removes the region.
7820 The @code{progn} special form is necessary because the
7821 @code{kill-region} command takes two arguments; and it would fail if
7822 @code{search-forward} and @code{point} expressions were written in
7823 sequence as two additional arguments. The @code{progn} expression is
7824 a single argument to @code{kill-region} and returns the one value that
7825 @code{kill-region} needs for its second argument.
7828 @section @code{kill-region}
7831 The @code{zap-to-char} function uses the @code{kill-region} function.
7832 This function clips text from a region and copies that text to
7833 the kill ring, from which it may be retrieved.
7838 (defun kill-region (beg end &optional yank-handler)
7839 "Kill (\"cut\") text between point and mark.
7840 This deletes the text from the buffer and saves it in the kill ring.
7841 The command \\[yank] can retrieve it from there.
7842 \(If you want to kill and then yank immediately, use \\[kill-ring-save].)
7844 If you want to append the killed region to the last killed text,
7845 use \\[append-next-kill] before \\[kill-region].
7847 If the buffer is read-only, Emacs will beep and refrain from deleting
7848 the text, but put the text in the kill ring anyway. This means that
7849 you can use the killing commands to copy text from a read-only buffer.
7851 This is the primitive for programs to kill text (as opposed to deleting it).
7852 Supply two arguments, character positions indicating the stretch of text
7854 Any command that calls this function is a \"kill command\".
7855 If the previous command was also a kill command,
7856 the text killed this time appends to the text killed last time
7857 to make one entry in the kill ring.
7859 In Lisp code, optional third arg YANK-HANDLER, if non-nil,
7860 specifies the yank-handler text property to be set on the killed
7861 text. See `insert-for-yank'."
7862 ;; Pass point first, then mark, because the order matters
7863 ;; when calling kill-append.
7864 (interactive (list (point) (mark)))
7865 (unless (and beg end)
7866 (error "The mark is not set now, so there is no region"))
7868 (let ((string (filter-buffer-substring beg end t)))
7869 (when string ;STRING is nil if BEG = END
7870 ;; Add that string to the kill ring, one way or another.
7871 (if (eq last-command 'kill-region)
7872 (kill-append string (< end beg) yank-handler)
7873 (kill-new string nil yank-handler)))
7874 (when (or string (eq last-command 'kill-region))
7875 (setq this-command 'kill-region))
7877 ((buffer-read-only text-read-only)
7878 ;; The code above failed because the buffer, or some of the characters
7879 ;; in the region, are read-only.
7880 ;; We should beep, in case the user just isn't aware of this.
7881 ;; However, there's no harm in putting
7882 ;; the region's text in the kill ring, anyway.
7883 (copy-region-as-kill beg end)
7884 ;; Set this-command now, so it will be set even if we get an error.
7885 (setq this-command 'kill-region)
7886 ;; This should barf, if appropriate, and give us the correct error.
7887 (if kill-read-only-ok
7888 (progn (message "Read only text copied to kill ring") nil)
7889 ;; Signal an error if the buffer is read-only.
7890 (barf-if-buffer-read-only)
7891 ;; If the buffer isn't read-only, the text is.
7892 (signal 'text-read-only (list (current-buffer)))))))
7895 The Emacs 22 version of that function uses @code{condition-case} and
7896 @code{copy-region-as-kill}, both of which we will explain.
7897 @code{condition-case} is an important special form.
7899 In essence, the @code{kill-region} function calls
7900 @code{condition-case}, which takes three arguments. In this function,
7901 the first argument does nothing. The second argument contains the
7902 code that does the work when all goes well. The third argument
7903 contains the code that is called in the event of an error.
7906 * Complete kill-region:: The function definition.
7907 * condition-case:: Dealing with a problem.
7912 @node Complete kill-region
7913 @unnumberedsubsec The Complete @code{kill-region} Definition
7917 We will go through the @code{condition-case} code in a moment. First,
7918 let us look at the definition of @code{kill-region}, with comments
7924 (defun kill-region (beg end)
7925 "Kill (\"cut\") text between point and mark.
7926 This deletes the text from the buffer and saves it in the kill ring.
7927 The command \\[yank] can retrieve it from there. @dots{} "
7931 ;; @bullet{} Since order matters, pass point first.
7932 (interactive (list (point) (mark)))
7933 ;; @bullet{} And tell us if we cannot cut the text.
7934 ;; 'unless' is an 'if' without a then-part.
7935 (unless (and beg end)
7936 (error "The mark is not set now, so there is no region"))
7940 ;; @bullet{} 'condition-case' takes three arguments.
7941 ;; If the first argument is nil, as it is here,
7942 ;; information about the error signal is not
7943 ;; stored for use by another function.
7948 ;; @bullet{} The second argument to 'condition-case' tells the
7949 ;; Lisp interpreter what to do when all goes well.
7953 ;; It starts with a 'let' function that extracts the string
7954 ;; and tests whether it exists. If so (that is what the
7955 ;; 'when' checks), it calls an 'if' function that determines
7956 ;; whether the previous command was another call to
7957 ;; 'kill-region'; if it was, then the new text is appended to
7958 ;; the previous text; if not, then a different function,
7959 ;; 'kill-new', is called.
7963 ;; The 'kill-append' function concatenates the new string and
7964 ;; the old. The 'kill-new' function inserts text into a new
7965 ;; item in the kill ring.
7969 ;; 'when' is an 'if' without an else-part. The second 'when'
7970 ;; again checks whether the current string exists; in
7971 ;; addition, it checks whether the previous command was
7972 ;; another call to 'kill-region'. If one or the other
7973 ;; condition is true, then it sets the current command to
7974 ;; be 'kill-region'.
7977 (let ((string (filter-buffer-substring beg end t)))
7978 (when string ;STRING is nil if BEG = END
7979 ;; Add that string to the kill ring, one way or another.
7980 (if (eq last-command 'kill-region)
7983 ;; @minus{} 'yank-handler' is an optional argument to
7984 ;; 'kill-region' that tells the 'kill-append' and
7985 ;; 'kill-new' functions how deal with properties
7986 ;; added to the text, such as 'bold' or 'italics'.
7987 (kill-append string (< end beg) yank-handler)
7988 (kill-new string nil yank-handler)))
7989 (when (or string (eq last-command 'kill-region))
7990 (setq this-command 'kill-region))
7995 ;; @bullet{} The third argument to 'condition-case' tells the interpreter
7996 ;; what to do with an error.
7999 ;; The third argument has a conditions part and a body part.
8000 ;; If the conditions are met (in this case,
8001 ;; if text or buffer are read-only)
8002 ;; then the body is executed.
8005 ;; The first part of the third argument is the following:
8006 ((buffer-read-only text-read-only) ;; the if-part
8007 ;; @dots{} the then-part
8008 (copy-region-as-kill beg end)
8011 ;; Next, also as part of the then-part, set this-command, so
8012 ;; it will be set in an error
8013 (setq this-command 'kill-region)
8014 ;; Finally, in the then-part, send a message if you may copy
8015 ;; the text to the kill ring without signaling an error, but
8016 ;; don't if you may not.
8019 (if kill-read-only-ok
8020 (progn (message "Read only text copied to kill ring") nil)
8021 (barf-if-buffer-read-only)
8022 ;; If the buffer isn't read-only, the text is.
8023 (signal 'text-read-only (list (current-buffer)))))
8031 (defun kill-region (beg end)
8032 "Kill between point and mark.
8033 The text is deleted but saved in the kill ring."
8038 ;; 1. 'condition-case' takes three arguments.
8039 ;; If the first argument is nil, as it is here,
8040 ;; information about the error signal is not
8041 ;; stored for use by another function.
8046 ;; 2. The second argument to 'condition-case'
8047 ;; tells the Lisp interpreter what to do when all goes well.
8051 ;; The 'delete-and-extract-region' function usually does the
8052 ;; work. If the beginning and ending of the region are both
8053 ;; the same, then the variable 'string' will be empty, or nil
8054 (let ((string (delete-and-extract-region beg end)))
8058 ;; 'when' is an 'if' clause that cannot take an 'else-part'.
8059 ;; Emacs normally sets the value of 'last-command' to the
8060 ;; previous command.
8063 ;; 'kill-append' concatenates the new string and the old.
8064 ;; 'kill-new' inserts text into a new item in the kill ring.
8066 (if (eq last-command 'kill-region)
8067 ;; if true, prepend string
8068 (kill-append string (< end beg))
8070 (setq this-command 'kill-region))
8074 ;; 3. The third argument to 'condition-case' tells the interpreter
8075 ;; what to do with an error.
8078 ;; The third argument has a conditions part and a body part.
8079 ;; If the conditions are met (in this case,
8080 ;; if text or buffer are read-only)
8081 ;; then the body is executed.
8084 ((buffer-read-only text-read-only) ;; this is the if-part
8086 (copy-region-as-kill beg end)
8089 (if kill-read-only-ok ;; usually this variable is nil
8090 (message "Read only text copied to kill ring")
8091 ;; or else, signal an error if the buffer is read-only;
8092 (barf-if-buffer-read-only)
8093 ;; and, in any case, signal that the text is read-only.
8094 (signal 'text-read-only (list (current-buffer)))))))
8099 @node condition-case
8100 @subsection @code{condition-case}
8101 @findex condition-case
8103 As we have seen earlier (@pxref{Making Errors, , Generate an Error
8104 Message}), when the Emacs Lisp interpreter has trouble evaluating an
8105 expression, it provides you with help; in the jargon, this is called
8106 ``signaling an error''. Usually, the computer stops the program and
8107 shows you a message.
8109 However, some programs undertake complicated actions. They should not
8110 simply stop on an error. In the @code{kill-region} function, the most
8111 likely error is that you will try to kill text that is read-only and
8112 cannot be removed. So the @code{kill-region} function contains code
8113 to handle this circumstance. This code, which makes up the body of
8114 the @code{kill-region} function, is inside of a @code{condition-case}
8118 The template for @code{condition-case} looks like this:
8125 @var{error-handler}@dots{})
8129 The second argument, @var{bodyform}, is straightforward. The
8130 @code{condition-case} special form causes the Lisp interpreter to
8131 evaluate the code in @var{bodyform}. If no error occurs, the special
8132 form returns the code's value and produces the side-effects, if any.
8134 In short, the @var{bodyform} part of a @code{condition-case}
8135 expression determines what should happen when everything works
8138 However, if an error occurs, among its other actions, the function
8139 generating the error signal will define one or more error condition
8142 An error handler is the third argument to @code{condition-case}.
8143 An error handler has two parts, a @var{condition-name} and a
8144 @var{body}. If the @var{condition-name} part of an error handler
8145 matches a condition name generated by an error, then the @var{body}
8146 part of the error handler is run.
8148 As you will expect, the @var{condition-name} part of an error handler
8149 may be either a single condition name or a list of condition names.
8151 Also, a complete @code{condition-case} expression may contain more
8152 than one error handler. When an error occurs, the first applicable
8155 Lastly, the first argument to the @code{condition-case} expression,
8156 the @var{var} argument, is sometimes bound to a variable that
8157 contains information about the error. However, if that argument is
8158 nil, as is the case in @code{kill-region}, that information is
8162 In brief, in the @code{kill-region} function, the code
8163 @code{condition-case} works like this:
8167 @var{If no errors}, @var{run only this code}
8168 @var{but}, @var{if errors}, @var{run this other code}.
8175 copy-region-as-kill is short, 12 lines, and uses
8176 filter-buffer-substring, which is longer, 39 lines
8177 and has delete-and-extract-region in it.
8178 delete-and-extract-region is written in C.
8180 see Initializing a Variable with @code{defvar}
8182 Initializing a Variable with @code{defvar} includes line 8350
8186 @subsection Lisp macro
8190 The part of the @code{condition-case} expression that is evaluated in
8191 the expectation that all goes well has a @code{when}. The code uses
8192 @code{when} to determine whether the @code{string} variable points to
8195 A @code{when} expression is simply a programmers' convenience. It is
8196 an @code{if} without the possibility of an else clause. In your mind,
8197 you can replace @code{when} with @code{if} and understand what goes
8198 on. That is what the Lisp interpreter does.
8200 Technically speaking, @code{when} is a Lisp macro. A Lisp macro
8201 enables you to define new control constructs and other language
8202 features. It tells the interpreter how to compute another Lisp
8203 expression which will in turn compute the value. In this case, the
8204 ``other expression'' is an @code{if} expression.
8206 The @code{kill-region} function definition also has an @code{unless}
8207 macro; it is the converse of @code{when}. The @code{unless} macro is
8208 an @code{if} without a then clause
8210 For more about Lisp macros, see @ref{Macros, , Macros, elisp, The GNU
8211 Emacs Lisp Reference Manual}. The C programming language also
8212 provides macros. These are different, but also useful.
8215 We will briefly look at C macros in
8216 @ref{Digression into C}.
8220 Regarding the @code{when} macro, in the @code{condition-case}
8221 expression, when the string has content, then another conditional
8222 expression is executed. This is an @code{if} with both a then-part
8227 (if (eq last-command 'kill-region)
8228 (kill-append string (< end beg) yank-handler)
8229 (kill-new string nil yank-handler))
8233 The then-part is evaluated if the previous command was another call to
8234 @code{kill-region}; if not, the else-part is evaluated.
8236 @code{yank-handler} is an optional argument to @code{kill-region} that
8237 tells the @code{kill-append} and @code{kill-new} functions how deal
8238 with properties added to the text, such as ``bold'' or ``italics''.
8240 @code{last-command} is a variable that comes with Emacs that we have
8241 not seen before. Normally, whenever a function is executed, Emacs
8242 sets the value of @code{last-command} to the previous command.
8245 In this segment of the definition, the @code{if} expression checks
8246 whether the previous command was @code{kill-region}. If it was,
8249 (kill-append string (< end beg) yank-handler)
8253 concatenates a copy of the newly clipped text to the just previously
8254 clipped text in the kill ring.
8256 @node copy-region-as-kill
8257 @section @code{copy-region-as-kill}
8258 @findex copy-region-as-kill
8261 The @code{copy-region-as-kill} function copies a region of text from a
8262 buffer and (via either @code{kill-append} or @code{kill-new}) saves it
8263 in the @code{kill-ring}.
8265 If you call @code{copy-region-as-kill} immediately after a
8266 @code{kill-region} command, Emacs appends the newly copied text to the
8267 previously copied text. This means that if you yank back the text, you
8268 get it all, from both this and the previous operation. On the other
8269 hand, if some other command precedes the @code{copy-region-as-kill},
8270 the function copies the text into a separate entry in the kill ring.
8273 * Complete copy-region-as-kill:: The complete function definition.
8274 * copy-region-as-kill body:: The body of @code{copy-region-as-kill}.
8278 @node Complete copy-region-as-kill
8279 @unnumberedsubsec The complete @code{copy-region-as-kill} function definition
8283 Here is the complete text of the version 22 @code{copy-region-as-kill}
8288 (defun copy-region-as-kill (beg end)
8289 "Save the region as if killed, but don't kill it.
8290 In Transient Mark mode, deactivate the mark.
8291 If `interprogram-cut-function' is non-nil, also save the text for a window
8292 system cut and paste."
8296 (if (eq last-command 'kill-region)
8297 (kill-append (filter-buffer-substring beg end) (< end beg))
8298 (kill-new (filter-buffer-substring beg end)))
8301 (if transient-mark-mode
8302 (setq deactivate-mark t))
8308 As usual, this function can be divided into its component parts:
8312 (defun copy-region-as-kill (@var{argument-list})
8313 "@var{documentation}@dots{}"
8319 The arguments are @code{beg} and @code{end} and the function is
8320 interactive with @code{"r"}, so the two arguments must refer to the
8321 beginning and end of the region. If you have been reading through this
8322 document from the beginning, understanding these parts of a function is
8323 almost becoming routine.
8325 The documentation is somewhat confusing unless you remember that the
8326 word ``kill'' has a meaning different from usual. The ``Transient Mark''
8327 and @code{interprogram-cut-function} comments explain certain
8330 After you once set a mark, a buffer always contains a region. If you
8331 wish, you can use Transient Mark mode to highlight the region
8332 temporarily. (No one wants to highlight the region all the time, so
8333 Transient Mark mode highlights it only at appropriate times. Many
8334 people turn off Transient Mark mode, so the region is never
8337 Also, a windowing system allows you to copy, cut, and paste among
8338 different programs. In the X windowing system, for example, the
8339 @code{interprogram-cut-function} function is @code{x-select-text},
8340 which works with the windowing system's equivalent of the Emacs kill
8343 The body of the @code{copy-region-as-kill} function starts with an
8344 @code{if} clause. What this clause does is distinguish between two
8345 different situations: whether or not this command is executed
8346 immediately after a previous @code{kill-region} command. In the first
8347 case, the new region is appended to the previously copied text.
8348 Otherwise, it is inserted into the beginning of the kill ring as a
8349 separate piece of text from the previous piece.
8351 The last two lines of the function prevent the region from lighting up
8352 if Transient Mark mode is turned on.
8354 The body of @code{copy-region-as-kill} merits discussion in detail.
8356 @node copy-region-as-kill body
8357 @subsection The Body of @code{copy-region-as-kill}
8359 The @code{copy-region-as-kill} function works in much the same way as
8360 the @code{kill-region} function. Both are written so that two or more
8361 kills in a row combine their text into a single entry. If you yank
8362 back the text from the kill ring, you get it all in one piece.
8363 Moreover, kills that kill forward from the current position of the
8364 cursor are added to the end of the previously copied text and commands
8365 that copy text backwards add it to the beginning of the previously
8366 copied text. This way, the words in the text stay in the proper
8369 Like @code{kill-region}, the @code{copy-region-as-kill} function makes
8370 use of the @code{last-command} variable that keeps track of the
8371 previous Emacs command.
8374 * last-command & this-command::
8375 * kill-append function::
8376 * kill-new function::
8380 @node last-command & this-command
8381 @unnumberedsubsubsec @code{last-command} and @code{this-command}
8384 Normally, whenever a function is executed, Emacs sets the value of
8385 @code{this-command} to the function being executed (which in this case
8386 would be @code{copy-region-as-kill}). At the same time, Emacs sets
8387 the value of @code{last-command} to the previous value of
8388 @code{this-command}.
8390 In the first part of the body of the @code{copy-region-as-kill}
8391 function, an @code{if} expression determines whether the value of
8392 @code{last-command} is @code{kill-region}. If so, the then-part of
8393 the @code{if} expression is evaluated; it uses the @code{kill-append}
8394 function to concatenate the text copied at this call to the function
8395 with the text already in the first element (the @sc{car}) of the kill
8396 ring. On the other hand, if the value of @code{last-command} is not
8397 @code{kill-region}, then the @code{copy-region-as-kill} function
8398 attaches a new element to the kill ring using the @code{kill-new}
8402 The @code{if} expression reads as follows; it uses @code{eq}:
8406 (if (eq last-command 'kill-region)
8408 (kill-append (filter-buffer-substring beg end) (< end beg))
8410 (kill-new (filter-buffer-substring beg end)))
8414 @findex filter-buffer-substring
8415 (The @code{filter-buffer-substring} function returns a filtered
8416 substring of the buffer, if any. Optionally---the arguments are not
8417 here, so neither is done---the function may delete the initial text or
8418 return the text without its properties; this function is a replacement
8419 for the older @code{buffer-substring} function, which came before text
8420 properties were implemented.)
8422 @findex eq @r{(example of use)}
8424 The @code{eq} function tests whether its first argument is the same Lisp
8425 object as its second argument. The @code{eq} function is similar to the
8426 @code{equal} function in that it is used to test for equality, but
8427 differs in that it determines whether two representations are actually
8428 the same object inside the computer, but with different names.
8429 @code{equal} determines whether the structure and contents of two
8430 expressions are the same.
8432 If the previous command was @code{kill-region}, then the Emacs Lisp
8433 interpreter calls the @code{kill-append} function
8435 @node kill-append function
8436 @unnumberedsubsubsec The @code{kill-append} function
8440 The @code{kill-append} function looks like this:
8445 (defun kill-append (string before-p &optional yank-handler)
8446 "Append STRING to the end of the latest kill in the kill ring.
8447 If BEFORE-P is non-nil, prepend STRING to the kill.
8449 (let* ((cur (car kill-ring)))
8450 (kill-new (if before-p (concat string cur) (concat cur string))
8451 (or (= (length cur) 0)
8453 (get-text-property 0 'yank-handler cur)))
8460 (defun kill-append (string before-p)
8461 "Append STRING to the end of the latest kill in the kill ring.
8462 If BEFORE-P is non-nil, prepend STRING to the kill.
8463 If `interprogram-cut-function' is set, pass the resulting kill to
8465 (kill-new (if before-p
8466 (concat string (car kill-ring))
8467 (concat (car kill-ring) string))
8472 The @code{kill-append} function is fairly straightforward. It uses
8473 the @code{kill-new} function, which we will discuss in more detail in
8476 (Also, the function provides an optional argument called
8477 @code{yank-handler}; when invoked, this argument tells the function
8478 how to deal with properties added to the text, such as ``bold'' or
8481 @c !!! bug in GNU Emacs 22 version of kill-append ?
8482 It has a @code{let*} function to set the value of the first element of
8483 the kill ring to @code{cur}. (I do not know why the function does not
8484 use @code{let} instead; only one value is set in the expression.
8485 Perhaps this is a bug that produces no problems?)
8487 Consider the conditional that is one of the two arguments to
8488 @code{kill-new}. It uses @code{concat} to concatenate the new text to
8489 the @sc{car} of the kill ring. Whether it prepends or appends the
8490 text depends on the results of an @code{if} expression:
8494 (if before-p ; @r{if-part}
8495 (concat string cur) ; @r{then-part}
8496 (concat cur string)) ; @r{else-part}
8501 If the region being killed is before the region that was killed in the
8502 last command, then it should be prepended before the material that was
8503 saved in the previous kill; and conversely, if the killed text follows
8504 what was just killed, it should be appended after the previous text.
8505 The @code{if} expression depends on the predicate @code{before-p} to
8506 decide whether the newly saved text should be put before or after the
8507 previously saved text.
8509 The symbol @code{before-p} is the name of one of the arguments to
8510 @code{kill-append}. When the @code{kill-append} function is
8511 evaluated, it is bound to the value returned by evaluating the actual
8512 argument. In this case, this is the expression @code{(< end beg)}.
8513 This expression does not directly determine whether the killed text in
8514 this command is located before or after the kill text of the last
8515 command; what it does is determine whether the value of the variable
8516 @code{end} is less than the value of the variable @code{beg}. If it
8517 is, it means that the user is most likely heading towards the
8518 beginning of the buffer. Also, the result of evaluating the predicate
8519 expression, @code{(< end beg)}, will be true and the text will be
8520 prepended before the previous text. On the other hand, if the value of
8521 the variable @code{end} is greater than the value of the variable
8522 @code{beg}, the text will be appended after the previous text.
8525 When the newly saved text will be prepended, then the string with the new
8526 text will be concatenated before the old text:
8534 But if the text will be appended, it will be concatenated
8538 (concat cur string))
8541 To understand how this works, we first need to review the
8542 @code{concat} function. The @code{concat} function links together or
8543 unites two strings of text. The result is a string. For example:
8547 (concat "abc" "def")
8553 (car '("first element" "second element")))
8554 @result{} "new first element"
8557 '("first element" "second element")) " modified")
8558 @result{} "first element modified"
8562 We can now make sense of @code{kill-append}: it modifies the contents
8563 of the kill ring. The kill ring is a list, each element of which is
8564 saved text. The @code{kill-append} function uses the @code{kill-new}
8565 function which in turn uses the @code{setcar} function.
8567 @node kill-new function
8568 @unnumberedsubsubsec The @code{kill-new} function
8571 @c in GNU Emacs 22, additional documentation to kill-new:
8573 Optional third arguments YANK-HANDLER controls how the STRING is later
8574 inserted into a buffer; see `insert-for-yank' for details.
8575 When a yank handler is specified, STRING must be non-empty (the yank
8576 handler, if non-nil, is stored as a `yank-handler' text property on STRING).
8578 When the yank handler has a non-nil PARAM element, the original STRING
8579 argument is not used by `insert-for-yank'. However, since Lisp code
8580 may access and use elements from the kill ring directly, the STRING
8581 argument should still be a \"useful\" string for such uses."
8584 The @code{kill-new} function looks like this:
8588 (defun kill-new (string &optional replace yank-handler)
8589 "Make STRING the latest kill in the kill ring.
8590 Set `kill-ring-yank-pointer' to point to it.
8592 If `interprogram-cut-function' is non-nil, apply it to STRING.
8593 Optional second argument REPLACE non-nil means that STRING will replace
8594 the front of the kill ring, rather than being added to the list.
8598 (if (> (length string) 0)
8600 (put-text-property 0 (length string)
8601 'yank-handler yank-handler string))
8603 (signal 'args-out-of-range
8604 (list string "yank-handler specified for empty string"))))
8607 (if (fboundp 'menu-bar-update-yank-menu)
8608 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8611 (if (and replace kill-ring)
8612 (setcar kill-ring string)
8613 (push string kill-ring)
8614 (if (> (length kill-ring) kill-ring-max)
8615 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8618 (setq kill-ring-yank-pointer kill-ring)
8619 (if interprogram-cut-function
8620 (funcall interprogram-cut-function string (not replace))))
8625 (defun kill-new (string &optional replace)
8626 "Make STRING the latest kill in the kill ring.
8627 Set the kill-ring-yank pointer to point to it.
8628 If `interprogram-cut-function' is non-nil, apply it to STRING.
8629 Optional second argument REPLACE non-nil means that STRING will replace
8630 the front of the kill ring, rather than being added to the list."
8631 (and (fboundp 'menu-bar-update-yank-menu)
8632 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8633 (if (and replace kill-ring)
8634 (setcar kill-ring string)
8635 (setq kill-ring (cons string kill-ring))
8636 (if (> (length kill-ring) kill-ring-max)
8637 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8638 (setq kill-ring-yank-pointer kill-ring)
8639 (if interprogram-cut-function
8640 (funcall interprogram-cut-function string (not replace))))
8643 (Notice that the function is not interactive.)
8645 As usual, we can look at this function in parts.
8647 The function definition has an optional @code{yank-handler} argument,
8648 which when invoked tells the function how to deal with properties
8649 added to the text, such as ``bold'' or ``italics''. We will skip that.
8652 The first line of the documentation makes sense:
8655 Make STRING the latest kill in the kill ring.
8659 Let's skip over the rest of the documentation for the moment.
8662 Also, let's skip over the initial @code{if} expression and those lines
8663 of code involving @code{menu-bar-update-yank-menu}. We will explain
8667 The critical lines are these:
8671 (if (and replace kill-ring)
8673 (setcar kill-ring string)
8677 (push string kill-ring)
8680 (setq kill-ring (cons string kill-ring))
8681 (if (> (length kill-ring) kill-ring-max)
8682 ;; @r{avoid overly long kill ring}
8683 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8686 (setq kill-ring-yank-pointer kill-ring)
8687 (if interprogram-cut-function
8688 (funcall interprogram-cut-function string (not replace))))
8692 The conditional test is @w{@code{(and replace kill-ring)}}.
8693 This will be true when two conditions are met: the kill ring has
8694 something in it, and the @code{replace} variable is true.
8697 When the @code{kill-append} function sets @code{replace} to be true
8698 and when the kill ring has at least one item in it, the @code{setcar}
8699 expression is executed:
8702 (setcar kill-ring string)
8705 The @code{setcar} function actually changes the first element of the
8706 @code{kill-ring} list to the value of @code{string}. It replaces the
8710 On the other hand, if the kill ring is empty, or replace is false, the
8711 else-part of the condition is executed:
8714 (push string kill-ring)
8719 @code{push} puts its first argument onto the second. It is similar to
8723 (setq kill-ring (cons string kill-ring))
8731 (add-to-list kill-ring string)
8735 When it is false, the expression first constructs a new version of the
8736 kill ring by prepending @code{string} to the existing kill ring as a
8737 new element (that is what the @code{push} does). Then it executes a
8738 second @code{if} clause. This second @code{if} clause keeps the kill
8739 ring from growing too long.
8741 Let's look at these two expressions in order.
8743 The @code{push} line of the else-part sets the new value of the kill
8744 ring to what results from adding the string being killed to the old
8747 We can see how this works with an example.
8753 (setq example-list '("here is a clause" "another clause"))
8758 After evaluating this expression with @kbd{C-x C-e}, you can evaluate
8759 @code{example-list} and see what it returns:
8764 @result{} ("here is a clause" "another clause")
8770 Now, we can add a new element on to this list by evaluating the
8771 following expression:
8772 @findex push, @r{example}
8775 (push "a third clause" example-list)
8780 When we evaluate @code{example-list}, we find its value is:
8785 @result{} ("a third clause" "here is a clause" "another clause")
8790 Thus, the third clause is added to the list by @code{push}.
8793 Now for the second part of the @code{if} clause. This expression
8794 keeps the kill ring from growing too long. It looks like this:
8798 (if (> (length kill-ring) kill-ring-max)
8799 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil))
8803 The code checks whether the length of the kill ring is greater than
8804 the maximum permitted length. This is the value of
8805 @code{kill-ring-max} (which is 60, by default). If the length of the
8806 kill ring is too long, then this code sets the last element of the
8807 kill ring to @code{nil}. It does this by using two functions,
8808 @code{nthcdr} and @code{setcdr}.
8810 We looked at @code{setcdr} earlier (@pxref{setcdr, , @code{setcdr}}).
8811 It sets the @sc{cdr} of a list, just as @code{setcar} sets the
8812 @sc{car} of a list. In this case, however, @code{setcdr} will not be
8813 setting the @sc{cdr} of the whole kill ring; the @code{nthcdr}
8814 function is used to cause it to set the @sc{cdr} of the next to last
8815 element of the kill ring---this means that since the @sc{cdr} of the
8816 next to last element is the last element of the kill ring, it will set
8817 the last element of the kill ring.
8819 @findex nthcdr, @r{example}
8820 The @code{nthcdr} function works by repeatedly taking the @sc{cdr} of a
8821 list---it takes the @sc{cdr} of the @sc{cdr} of the @sc{cdr}
8822 @dots{} It does this @var{N} times and returns the results.
8823 (@xref{nthcdr, , @code{nthcdr}}.)
8825 @findex setcdr, @r{example}
8826 Thus, if we had a four element list that was supposed to be three
8827 elements long, we could set the @sc{cdr} of the next to last element
8828 to @code{nil}, and thereby shorten the list. (If you set the last
8829 element to some other value than @code{nil}, which you could do, then
8830 you would not have shortened the list. @xref{setcdr, ,
8833 You can see shortening by evaluating the following three expressions
8834 in turn. First set the value of @code{trees} to @code{(maple oak pine
8835 birch)}, then set the @sc{cdr} of its second @sc{cdr} to @code{nil}
8836 and then find the value of @code{trees}:
8840 (setq trees '(maple oak pine birch))
8841 @result{} (maple oak pine birch)
8845 (setcdr (nthcdr 2 trees) nil)
8849 @result{} (maple oak pine)
8854 (The value returned by the @code{setcdr} expression is @code{nil} since
8855 that is what the @sc{cdr} is set to.)
8857 To repeat, in @code{kill-new}, the @code{nthcdr} function takes the
8858 @sc{cdr} a number of times that is one less than the maximum permitted
8859 size of the kill ring and @code{setcdr} sets the @sc{cdr} of that
8860 element (which will be the rest of the elements in the kill ring) to
8861 @code{nil}. This prevents the kill ring from growing too long.
8864 The next to last expression in the @code{kill-new} function is
8867 (setq kill-ring-yank-pointer kill-ring)
8870 The @code{kill-ring-yank-pointer} is a global variable that is set to be
8871 the @code{kill-ring}.
8873 Even though the @code{kill-ring-yank-pointer} is called a
8874 @samp{pointer}, it is a variable just like the kill ring. However, the
8875 name has been chosen to help humans understand how the variable is used.
8878 Now, to return to an early expression in the body of the function:
8882 (if (fboundp 'menu-bar-update-yank-menu)
8883 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8888 It starts with an @code{if} expression
8890 In this case, the expression tests first to see whether
8891 @code{menu-bar-update-yank-menu} exists as a function, and if so,
8892 calls it. The @code{fboundp} function returns true if the symbol it
8893 is testing has a function definition that ``is not void''. If the
8894 symbol's function definition were void, we would receive an error
8895 message, as we did when we created errors intentionally (@pxref{Making
8896 Errors, , Generate an Error Message}).
8899 The then-part contains an expression whose first element is the
8900 function @code{and}.
8903 The @code{and} special form evaluates each of its arguments until one
8904 of the arguments returns a value of @code{nil}, in which case the
8905 @code{and} expression returns @code{nil}; however, if none of the
8906 arguments returns a value of @code{nil}, the value resulting from
8907 evaluating the last argument is returned. (Since such a value is not
8908 @code{nil}, it is considered true in Emacs Lisp.) In other words, an
8909 @code{and} expression returns a true value only if all its arguments
8910 are true. (@xref{Second Buffer Related Review}.)
8912 The expression determines whether the second argument to
8913 @code{menu-bar-update-yank-menu} is true or not.
8915 ;; If we're supposed to be extending an existing string, and that
8916 ;; string really is at the front of the menu, then update it in place.
8919 @code{menu-bar-update-yank-menu} is one of the functions that make it
8920 possible to use the ``Select and Paste'' menu in the Edit item of a menu
8921 bar; using a mouse, you can look at the various pieces of text you
8922 have saved and select one piece to paste.
8924 The last expression in the @code{kill-new} function adds the newly
8925 copied string to whatever facility exists for copying and pasting
8926 among different programs running in a windowing system. In the X
8927 Windowing system, for example, the @code{x-select-text} function takes
8928 the string and stores it in memory operated by X@. You can paste the
8929 string in another program, such as an Xterm.
8932 The expression looks like this:
8936 (if interprogram-cut-function
8937 (funcall interprogram-cut-function string (not replace))))
8941 If an @code{interprogram-cut-function} exists, then Emacs executes
8942 @code{funcall}, which in turn calls its first argument as a function
8943 and passes the remaining arguments to it. (Incidentally, as far as I
8944 can see, this @code{if} expression could be replaced by an @code{and}
8945 expression similar to the one in the first part of the function.)
8947 We are not going to discuss windowing systems and other programs
8948 further, but merely note that this is a mechanism that enables GNU
8949 Emacs to work easily and well with other programs.
8951 This code for placing text in the kill ring, either concatenated with
8952 an existing element or as a new element, leads us to the code for
8953 bringing back text that has been cut out of the buffer---the yank
8954 commands. However, before discussing the yank commands, it is better
8955 to learn how lists are implemented in a computer. This will make
8956 clear such mysteries as the use of the term ``pointer''. But before
8957 that, we will digress into C.
8960 @c is this true in Emacs 22? Does not seems to be
8962 (If the @w{@code{(< end beg))}}
8963 expression is true, @code{kill-append} prepends the string to the just
8964 previously clipped text. For a detailed discussion, see
8965 @ref{kill-append function, , The @code{kill-append} function}.)
8967 If you then yank back the text, i.e., ``paste'' it, you get both
8968 pieces of text at once. That way, if you delete two words in a row,
8969 and then yank them back, you get both words, in their proper order,
8970 with one yank. (The @w{@code{(< end beg))}} expression makes sure the
8973 On the other hand, if the previous command is not @code{kill-region},
8974 then the @code{kill-new} function is called, which adds the text to
8975 the kill ring as the latest item, and sets the
8976 @code{kill-ring-yank-pointer} variable to point to it.
8980 @c Evidently, changed for Emacs 22. The zap-to-char command does not
8981 @c use the delete-and-extract-region function
8983 2006 Oct 26, the Digression into C is now OK but should come after
8984 copy-region-as-kill and filter-buffer-substring
8988 copy-region-as-kill is short, 12 lines, and uses
8989 filter-buffer-substring, which is longer, 39 lines
8990 and has delete-and-extract-region in it.
8991 delete-and-extract-region is written in C.
8993 see Initializing a Variable with @code{defvar}
8996 @node Digression into C
8997 @section Digression into C
8998 @findex delete-and-extract-region
8999 @cindex C, a digression into
9000 @cindex Digression into C
9002 The @code{copy-region-as-kill} function (@pxref{copy-region-as-kill, ,
9003 @code{copy-region-as-kill}}) uses the @code{filter-buffer-substring}
9004 function, which in turn uses the @code{delete-and-extract-region}
9005 function. It removes the contents of a region and you cannot get them
9008 Unlike the other code discussed here, the
9009 @code{delete-and-extract-region} function is not written in Emacs
9010 Lisp; it is written in C and is one of the primitives of the GNU Emacs
9011 system. Since it is very simple, I will digress briefly from Lisp and
9014 @c GNU Emacs 24 in src/editfns.c
9015 @c the DEFUN for delete-and-extract-region
9018 Like many of the other Emacs primitives,
9019 @code{delete-and-extract-region} is written as an instance of a C
9020 macro, a macro being a template for code. The complete macro looks
9025 DEFUN ("delete-and-extract-region", Fdelete_and_extract_region,
9026 Sdelete_and_extract_region, 2, 2, 0,
9027 doc: /* Delete the text between START and END and return it. */)
9028 (Lisp_Object start, Lisp_Object end)
9030 validate_region (&start, &end);
9031 if (XINT (start) == XINT (end))
9032 return empty_unibyte_string;
9033 return del_range_1 (XINT (start), XINT (end), 1, 1);
9038 Without going into the details of the macro writing process, let me
9039 point out that this macro starts with the word @code{DEFUN}. The word
9040 @code{DEFUN} was chosen since the code serves the same purpose as
9041 @code{defun} does in Lisp. (The @code{DEFUN} C macro is defined in
9042 @file{emacs/src/lisp.h}.)
9044 The word @code{DEFUN} is followed by seven parts inside of
9049 The first part is the name given to the function in Lisp,
9050 @code{delete-and-extract-region}.
9053 The second part is the name of the function in C,
9054 @code{Fdelete_and_extract_region}. By convention, it starts with
9055 @samp{F}. Since C does not use hyphens in names, underscores are used
9059 The third part is the name for the C constant structure that records
9060 information on this function for internal use. It is the name of the
9061 function in C but begins with an @samp{S} instead of an @samp{F}.
9064 The fourth and fifth parts specify the minimum and maximum number of
9065 arguments the function can have. This function demands exactly 2
9069 The sixth part is nearly like the argument that follows the
9070 @code{interactive} declaration in a function written in Lisp: a letter
9071 followed, perhaps, by a prompt. The only difference from the Lisp is
9072 when the macro is called with no arguments. Then you write a @code{0}
9073 (which is a ``null string''), as in this macro.
9075 If you were to specify arguments, you would place them between
9076 quotation marks. The C macro for @code{goto-char} includes
9077 @code{"NGoto char: "} in this position to indicate that the function
9078 expects a raw prefix, in this case, a numerical location in a buffer,
9079 and provides a prompt.
9082 The seventh part is a documentation string, just like the one for a
9083 function written in Emacs Lisp. This is written as a C comment. (When
9084 you build Emacs, the program @command{lib-src/make-docfile} extracts
9085 these comments and uses them to make the ``real'' documentation.)
9089 In a C macro, the formal parameters come next, with a statement of
9090 what kind of object they are, followed by what might be called the ``body''
9091 of the macro. For @code{delete-and-extract-region} the ``body''
9092 consists of the following four lines:
9096 validate_region (&start, &end);
9097 if (XINT (start) == XINT (end))
9098 return empty_unibyte_string;
9099 return del_range_1 (XINT (start), XINT (end), 1, 1);
9103 The @code{validate_region} function checks whether the values
9104 passed as the beginning and end of the region are the proper type and
9105 are within range. If the beginning and end positions are the same,
9106 then return an empty string.
9108 The @code{del_range_1} function actually deletes the text. It is a
9109 complex function we will not look into. It updates the buffer and
9110 does other things. However, it is worth looking at the two arguments
9111 passed to @code{del_range}. These are @w{@code{XINT (start)}} and
9112 @w{@code{XINT (end)}}.
9114 As far as the C language is concerned, @code{start} and @code{end} are
9115 two integers that mark the beginning and end of the region to be
9116 deleted@footnote{More precisely, and requiring more expert knowledge
9117 to understand, the two integers are of type @code{Lisp_Object}, which can
9118 also be a C union instead of an integer type.}.
9120 In early versions of Emacs, these two numbers were thirty-two bits
9121 long, but the code is slowly being generalized to handle other
9122 lengths. Three of the available bits are used to specify the type of
9123 information; the remaining bits are used as ``content''.
9125 @samp{XINT} is a C macro that extracts the relevant number from the
9126 longer collection of bits; the three other bits are discarded.
9129 The command in @code{delete-and-extract-region} looks like this:
9132 del_range_1 (XINT (start), XINT (end), 1, 1);
9136 It deletes the region between the beginning position, @code{start},
9137 and the ending position, @code{end}.
9139 From the point of view of the person writing Lisp, Emacs is all very
9140 simple; but hidden underneath is a great deal of complexity to make it
9144 @section Initializing a Variable with @code{defvar}
9146 @cindex Initializing a variable
9147 @cindex Variable initialization
9152 copy-region-as-kill is short, 12 lines, and uses
9153 filter-buffer-substring, which is longer, 39 lines
9154 and has delete-and-extract-region in it.
9155 delete-and-extract-region is written in C.
9157 see Initializing a Variable with @code{defvar}
9161 The @code{copy-region-as-kill} function is written in Emacs Lisp. Two
9162 functions within it, @code{kill-append} and @code{kill-new}, copy a
9163 region in a buffer and save it in a variable called the
9164 @code{kill-ring}. This section describes how the @code{kill-ring}
9165 variable is created and initialized using the @code{defvar} special
9168 (Again we note that the term @code{kill-ring} is a misnomer. The text
9169 that is clipped out of the buffer can be brought back; it is not a ring
9170 of corpses, but a ring of resurrectable text.)
9172 In Emacs Lisp, a variable such as the @code{kill-ring} is created and
9173 given an initial value by using the @code{defvar} special form. The
9174 name comes from ``define variable''.
9176 The @code{defvar} special form is similar to @code{setq} in that it sets
9177 the value of a variable. It is unlike @code{setq} in two ways: first,
9178 it only sets the value of the variable if the variable does not already
9179 have a value. If the variable already has a value, @code{defvar} does
9180 not override the existing value. Second, @code{defvar} has a
9181 documentation string.
9183 (There is a related macro, @code{defcustom}, designed for variables
9184 that people customize. It has more features than @code{defvar}.
9185 (@xref{defcustom, , Setting Variables with @code{defcustom}}.)
9188 * See variable current value::
9189 * defvar and asterisk::
9193 @node See variable current value
9194 @unnumberedsubsec Seeing the Current Value of a Variable
9197 You can see the current value of a variable, any variable, by using
9198 the @code{describe-variable} function, which is usually invoked by
9199 typing @kbd{C-h v}. If you type @kbd{C-h v} and then @code{kill-ring}
9200 (followed by @key{RET}) when prompted, you will see what is in your
9201 current kill ring---this may be quite a lot! Conversely, if you have
9202 been doing nothing this Emacs session except read this document, you
9203 may have nothing in it. Also, you will see the documentation for
9209 List of killed text sequences.
9210 Since the kill ring is supposed to interact nicely with cut-and-paste
9211 facilities offered by window systems, use of this variable should
9214 interact nicely with `interprogram-cut-function' and
9215 `interprogram-paste-function'. The functions `kill-new',
9216 `kill-append', and `current-kill' are supposed to implement this
9217 interaction; you may want to use them instead of manipulating the kill
9223 The kill ring is defined by a @code{defvar} in the following way:
9227 (defvar kill-ring nil
9228 "List of killed text sequences.
9234 In this variable definition, the variable is given an initial value of
9235 @code{nil}, which makes sense, since if you have saved nothing, you want
9236 nothing back if you give a @code{yank} command. The documentation
9237 string is written just like the documentation string of a @code{defun}.
9238 As with the documentation string of the @code{defun}, the first line of
9239 the documentation should be a complete sentence, since some commands,
9240 like @code{apropos}, print only the first line of documentation.
9241 Succeeding lines should not be indented; otherwise they look odd when
9242 you use @kbd{C-h v} (@code{describe-variable}).
9244 @node defvar and asterisk
9245 @subsection @code{defvar} and an asterisk
9246 @findex defvar @r{for a user customizable variable}
9247 @findex defvar @r{with an asterisk}
9249 In the past, Emacs used the @code{defvar} special form both for
9250 internal variables that you would not expect a user to change and for
9251 variables that you do expect a user to change. Although you can still
9252 use @code{defvar} for user customizable variables, please use
9253 @code{defcustom} instead, since it provides a path into
9254 the Customization commands. (@xref{defcustom, , Specifying Variables
9255 using @code{defcustom}}.)
9257 When you specified a variable using the @code{defvar} special form,
9258 you could distinguish a variable that a user might want to change from
9259 others by typing an asterisk, @samp{*}, in the first column of its
9260 documentation string. For example:
9264 (defvar shell-command-default-error-buffer nil
9265 "*Buffer name for `shell-command' @dots{} error output.
9270 @findex set-variable
9272 You could (and still can) use the @code{set-variable} command to
9273 change the value of @code{shell-command-default-error-buffer}
9274 temporarily. However, options set using @code{set-variable} are set
9275 only for the duration of your editing session. The new values are not
9276 saved between sessions. Each time Emacs starts, it reads the original
9277 value, unless you change the value within your @file{.emacs} file,
9278 either by setting it manually or by using @code{customize}.
9279 @xref{Emacs Initialization, , Your @file{.emacs} File}.
9281 For me, the major use of the @code{set-variable} command is to suggest
9282 variables that I might want to set in my @file{.emacs} file. There
9283 are now more than 700 such variables, far too many to remember
9284 readily. Fortunately, you can press @key{TAB} after calling the
9285 @code{M-x set-variable} command to see the list of variables.
9286 (@xref{Examining, , Examining and Setting Variables, emacs,
9287 The GNU Emacs Manual}.)
9290 @node cons & search-fwd Review
9293 Here is a brief summary of some recently introduced functions.
9298 @code{car} returns the first element of a list; @code{cdr} returns the
9299 second and subsequent elements of a list.
9306 (car '(1 2 3 4 5 6 7))
9308 (cdr '(1 2 3 4 5 6 7))
9309 @result{} (2 3 4 5 6 7)
9314 @code{cons} constructs a list by prepending its first argument to its
9328 @code{funcall} evaluates its first argument as a function. It passes
9329 its remaining arguments to its first argument.
9332 Return the result of taking @sc{cdr} @var{n} times on a list.
9340 The ``rest of the rest'', as it were.
9347 (nthcdr 3 '(1 2 3 4 5 6 7))
9354 @code{setcar} changes the first element of a list; @code{setcdr}
9355 changes the second and subsequent elements of a list.
9362 (setq triple '(1 2 3))
9369 (setcdr triple '("foo" "bar"))
9372 @result{} (37 "foo" "bar")
9377 Evaluate each argument in sequence and then return the value of the
9390 @item save-restriction
9391 Record whatever narrowing is in effect in the current buffer, if any,
9392 and restore that narrowing after evaluating the arguments.
9394 @item search-forward
9395 Search for a string, and if the string is found, move point. With a
9396 regular expression, use the similar @code{re-search-forward}.
9397 (@xref{Regexp Search, , Regular Expression Searches}, for an
9398 explanation of regular expression patterns and searches.)
9402 @code{search-forward} and @code{re-search-forward} take four
9407 The string or regular expression to search for.
9410 Optionally, the limit of the search.
9413 Optionally, what to do if the search fails, return @code{nil} or an
9417 Optionally, how many times to repeat the search; if negative, the
9418 search goes backwards.
9422 @itemx delete-and-extract-region
9423 @itemx copy-region-as-kill
9425 @code{kill-region} cuts the text between point and mark from the
9426 buffer and stores that text in the kill ring, so you can get it back
9429 @code{copy-region-as-kill} copies the text between point and mark into
9430 the kill ring, from which you can get it by yanking. The function
9431 does not cut or remove the text from the buffer.
9434 @code{delete-and-extract-region} removes the text between point and
9435 mark from the buffer and throws it away. You cannot get it back.
9436 (This is not an interactive command.)
9439 @node search Exercises
9440 @section Searching Exercises
9444 Write an interactive function that searches for a string. If the
9445 search finds the string, leave point after it and display a message
9446 that says ``Found!''. (Do not use @code{search-forward} for the name
9447 of this function; if you do, you will overwrite the existing version of
9448 @code{search-forward} that comes with Emacs. Use a name such as
9449 @code{test-search} instead.)
9452 Write a function that prints the third element of the kill ring in the
9453 echo area, if any; if the kill ring does not contain a third element,
9454 print an appropriate message.
9457 @node List Implementation
9458 @chapter How Lists are Implemented
9459 @cindex Lists in a computer
9461 In Lisp, atoms are recorded in a straightforward fashion; if the
9462 implementation is not straightforward in practice, it is, nonetheless,
9463 straightforward in theory. The atom @samp{rose}, for example, is
9464 recorded as the four contiguous letters @samp{r}, @samp{o}, @samp{s},
9465 @samp{e}. A list, on the other hand, is kept differently. The mechanism
9466 is equally simple, but it takes a moment to get used to the idea. A
9467 list is kept using a series of pairs of pointers. In the series, the
9468 first pointer in each pair points to an atom or to another list, and the
9469 second pointer in each pair points to the next pair, or to the symbol
9470 @code{nil}, which marks the end of the list.
9472 A pointer itself is quite simply the electronic address of what is
9473 pointed to. Hence, a list is kept as a series of electronic addresses.
9476 * Lists diagrammed::
9477 * Symbols as Chest:: Exploring a powerful metaphor.
9482 @node Lists diagrammed
9483 @unnumberedsec Lists diagrammed
9486 For example, the list @code{(rose violet buttercup)} has three elements,
9487 @samp{rose}, @samp{violet}, and @samp{buttercup}. In the computer, the
9488 electronic address of @samp{rose} is recorded in a segment of computer
9489 memory along with the address that gives the electronic address of where
9490 the atom @samp{violet} is located; and that address (the one that tells
9491 where @samp{violet} is located) is kept along with an address that tells
9492 where the address for the atom @samp{buttercup} is located.
9495 This sounds more complicated than it is and is easier seen in a diagram:
9497 @c clear print-postscript-figures
9498 @c !!! cons-cell-diagram #1
9502 ___ ___ ___ ___ ___ ___
9503 |___|___|--> |___|___|--> |___|___|--> nil
9506 --> rose --> violet --> buttercup
9510 @ifset print-postscript-figures
9513 @center @image{cons-1}
9517 @ifclear print-postscript-figures
9521 ___ ___ ___ ___ ___ ___
9522 |___|___|--> |___|___|--> |___|___|--> nil
9525 --> rose --> violet --> buttercup
9532 In the diagram, each box represents a word of computer memory that
9533 holds a Lisp object, usually in the form of a memory address. The boxes,
9534 i.e., the addresses, are in pairs. Each arrow points to what the address
9535 is the address of, either an atom or another pair of addresses. The
9536 first box is the electronic address of @samp{rose} and the arrow points
9537 to @samp{rose}; the second box is the address of the next pair of boxes,
9538 the first part of which is the address of @samp{violet} and the second
9539 part of which is the address of the next pair. The very last box
9540 points to the symbol @code{nil}, which marks the end of the list.
9543 When a variable is set to a list with a function such as @code{setq},
9544 it stores the address of the first box in the variable. Thus,
9545 evaluation of the expression
9548 (setq bouquet '(rose violet buttercup))
9553 creates a situation like this:
9555 @c cons-cell-diagram #2
9561 | ___ ___ ___ ___ ___ ___
9562 --> |___|___|--> |___|___|--> |___|___|--> nil
9565 --> rose --> violet --> buttercup
9569 @ifset print-postscript-figures
9572 @center @image{cons-2}
9576 @ifclear print-postscript-figures
9582 | ___ ___ ___ ___ ___ ___
9583 --> |___|___|--> |___|___|--> |___|___|--> nil
9586 --> rose --> violet --> buttercup
9593 In this example, the symbol @code{bouquet} holds the address of the first
9597 This same list can be illustrated in a different sort of box notation
9600 @c cons-cell-diagram #2a
9606 | -------------- --------------- ----------------
9607 | | car | cdr | | car | cdr | | car | cdr |
9608 -->| rose | o------->| violet | o------->| butter- | nil |
9609 | | | | | | | cup | |
9610 -------------- --------------- ----------------
9614 @ifset print-postscript-figures
9617 @center @image{cons-2a}
9621 @ifclear print-postscript-figures
9627 | -------------- --------------- ----------------
9628 | | car | cdr | | car | cdr | | car | cdr |
9629 -->| rose | o------->| violet | o------->| butter- | nil |
9630 | | | | | | | cup | |
9631 -------------- --------------- ----------------
9637 (Symbols consist of more than pairs of addresses, but the structure of
9638 a symbol is made up of addresses. Indeed, the symbol @code{bouquet}
9639 consists of a group of address-boxes, one of which is the address of
9640 the printed word @samp{bouquet}, a second of which is the address of a
9641 function definition attached to the symbol, if any, a third of which
9642 is the address of the first pair of address-boxes for the list
9643 @code{(rose violet buttercup)}, and so on. Here we are showing that
9644 the symbol's third address-box points to the first pair of
9645 address-boxes for the list.)
9647 If a symbol is set to the @sc{cdr} of a list, the list itself is not
9648 changed; the symbol simply has an address further down the list. (In
9649 the jargon, @sc{car} and @sc{cdr} are ``non-destructive''.) Thus,
9650 evaluation of the following expression
9653 (setq flowers (cdr bouquet))
9660 @c cons-cell-diagram #3
9667 | ___ ___ | ___ ___ ___ ___
9668 --> | | | --> | | | | | |
9669 |___|___|----> |___|___|--> |___|___|--> nil
9672 --> rose --> violet --> buttercup
9677 @ifset print-postscript-figures
9680 @center @image{cons-3}
9684 @ifclear print-postscript-figures
9691 | ___ ___ | ___ ___ ___ ___
9692 --> | | | --> | | | | | |
9693 |___|___|----> |___|___|--> |___|___|--> nil
9696 --> rose --> violet --> buttercup
9704 The value of @code{flowers} is @code{(violet buttercup)}, which is
9705 to say, the symbol @code{flowers} holds the address of the pair of
9706 address-boxes, the first of which holds the address of @code{violet},
9707 and the second of which holds the address of @code{buttercup}.
9709 A pair of address-boxes is called a @dfn{cons cell} or @dfn{dotted
9710 pair}. @xref{Cons Cell Type, , Cons Cell and List Types, elisp, The GNU Emacs Lisp
9711 Reference Manual}, and @ref{Dotted Pair Notation, , Dotted Pair
9712 Notation, elisp, The GNU Emacs Lisp Reference Manual}, for more
9713 information about cons cells and dotted pairs.
9716 The function @code{cons} adds a new pair of addresses to the front of
9717 a series of addresses like that shown above. For example, evaluating
9721 (setq bouquet (cons 'lily bouquet))
9728 @c cons-cell-diagram #4
9735 | ___ ___ ___ ___ | ___ ___ ___ ___
9736 --> | | | | | | --> | | | | | |
9737 |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil
9740 --> lily --> rose --> violet --> buttercup
9745 @ifset print-postscript-figures
9748 @center @image{cons-4}
9752 @ifclear print-postscript-figures
9759 | ___ ___ ___ ___ | ___ ___ ___ ___
9760 --> | | | | | | --> | | | | | |
9761 |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil
9764 --> lily --> rose --> violet --> buttercup
9773 However, this does not change the value of the symbol
9774 @code{flowers}, as you can see by evaluating the following,
9777 (eq (cdr (cdr bouquet)) flowers)
9781 which returns @code{t} for true.
9783 Until it is reset, @code{flowers} still has the value
9784 @code{(violet buttercup)}; that is, it has the address of the cons
9785 cell whose first address is of @code{violet}. Also, this does not
9786 alter any of the pre-existing cons cells; they are all still there.
9788 Thus, in Lisp, to get the @sc{cdr} of a list, you just get the address
9789 of the next cons cell in the series; to get the @sc{car} of a list,
9790 you get the address of the first element of the list; to @code{cons} a
9791 new element on a list, you add a new cons cell to the front of the list.
9792 That is all there is to it! The underlying structure of Lisp is
9795 And what does the last address in a series of cons cells refer to? It
9796 is the address of the empty list, of @code{nil}.
9798 In summary, when a Lisp variable is set to a value, it is provided with
9799 the address of the list to which the variable refers.
9801 @node Symbols as Chest
9802 @section Symbols as a Chest of Drawers
9803 @cindex Symbols as a Chest of Drawers
9804 @cindex Chest of Drawers, metaphor for a symbol
9805 @cindex Drawers, Chest of, metaphor for a symbol
9807 In an earlier section, I suggested that you might imagine a symbol as
9808 being a chest of drawers. The function definition is put in one
9809 drawer, the value in another, and so on. What is put in the drawer
9810 holding the value can be changed without affecting the contents of the
9811 drawer holding the function definition, and vice versa.
9813 Actually, what is put in each drawer is the address of the value or
9814 function definition. It is as if you found an old chest in the attic,
9815 and in one of its drawers you found a map giving you directions to
9816 where the buried treasure lies.
9818 (In addition to its name, symbol definition, and variable value, a
9819 symbol has a ``drawer'' for a @dfn{property list} which can be used to
9820 record other information. Property lists are not discussed here; see
9821 @ref{Property Lists, , Property Lists, elisp, The GNU Emacs Lisp
9825 Here is a fanciful representation:
9827 @c chest-of-drawers diagram
9832 Chest of Drawers Contents of Drawers
9836 ---------------------
9837 | directions to | [map to]
9838 | symbol name | bouquet
9840 +---------------------+
9842 | symbol definition | [none]
9844 +---------------------+
9845 | directions to | [map to]
9846 | variable value | (rose violet buttercup)
9848 +---------------------+
9850 | property list | [not described here]
9852 +---------------------+
9858 @ifset print-postscript-figures
9861 @center @image{drawers}
9865 @ifclear print-postscript-figures
9870 Chest of Drawers Contents of Drawers
9874 ---------------------
9875 | directions to | [map to]
9876 | symbol name | bouquet
9878 +---------------------+
9880 | symbol definition | [none]
9882 +---------------------+
9883 | directions to | [map to]
9884 | variable value | (rose violet buttercup)
9886 +---------------------+
9888 | property list | [not described here]
9890 +---------------------+
9901 Set @code{flowers} to @code{violet} and @code{buttercup}. Cons two
9902 more flowers on to this list and set this new list to
9903 @code{more-flowers}. Set the @sc{car} of @code{flowers} to a fish.
9904 What does the @code{more-flowers} list now contain?
9907 @chapter Yanking Text Back
9909 @cindex Text retrieval
9910 @cindex Retrieving text
9911 @cindex Pasting text
9913 Whenever you cut text out of a buffer with a ``kill'' command in GNU Emacs,
9914 you can bring it back with a ``yank'' command. The text that is cut out of
9915 the buffer is put in the kill ring and the yank commands insert the
9916 appropriate contents of the kill ring back into a buffer (not necessarily
9917 the original buffer).
9919 A simple @kbd{C-y} (@code{yank}) command inserts the first item from
9920 the kill ring into the current buffer. If the @kbd{C-y} command is
9921 followed immediately by @kbd{M-y}, the first element is replaced by
9922 the second element. Successive @kbd{M-y} commands replace the second
9923 element with the third, fourth, or fifth element, and so on. When the
9924 last element in the kill ring is reached, it is replaced by the first
9925 element and the cycle is repeated. (Thus the kill ring is called a
9926 ``ring'' rather than just a ``list''. However, the actual data structure
9927 that holds the text is a list.
9928 @xref{Kill Ring, , Handling the Kill Ring}, for the details of how the
9929 list is handled as a ring.)
9932 * Kill Ring Overview::
9933 * kill-ring-yank-pointer:: The kill ring is a list.
9934 * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable.
9937 @node Kill Ring Overview
9938 @section Kill Ring Overview
9939 @cindex Kill ring overview
9941 The kill ring is a list of textual strings. This is what it looks like:
9944 ("some text" "a different piece of text" "yet more text")
9947 If this were the contents of my kill ring and I pressed @kbd{C-y}, the
9948 string of characters saying @samp{some text} would be inserted in this
9949 buffer where my cursor is located.
9951 The @code{yank} command is also used for duplicating text by copying it.
9952 The copied text is not cut from the buffer, but a copy of it is put on the
9953 kill ring and is inserted by yanking it back.
9955 Three functions are used for bringing text back from the kill ring:
9956 @code{yank}, which is usually bound to @kbd{C-y}; @code{yank-pop},
9957 which is usually bound to @kbd{M-y}; and @code{rotate-yank-pointer},
9958 which is used by the two other functions.
9960 These functions refer to the kill ring through a variable called the
9961 @code{kill-ring-yank-pointer}. Indeed, the insertion code for both the
9962 @code{yank} and @code{yank-pop} functions is:
9965 (insert (car kill-ring-yank-pointer))
9969 (Well, no more. In GNU Emacs 22, the function has been replaced by
9970 @code{insert-for-yank} which calls @code{insert-for-yank-1}
9971 repetitively for each @code{yank-handler} segment. In turn,
9972 @code{insert-for-yank-1} strips text properties from the inserted text
9973 according to @code{yank-excluded-properties}. Otherwise, it is just
9974 like @code{insert}. We will stick with plain @code{insert} since it
9975 is easier to understand.)
9977 To begin to understand how @code{yank} and @code{yank-pop} work, it is
9978 first necessary to look at the @code{kill-ring-yank-pointer} variable.
9980 @node kill-ring-yank-pointer
9981 @section The @code{kill-ring-yank-pointer} Variable
9983 @code{kill-ring-yank-pointer} is a variable, just as @code{kill-ring} is
9984 a variable. It points to something by being bound to the value of what
9985 it points to, like any other Lisp variable.
9988 Thus, if the value of the kill ring is:
9991 ("some text" "a different piece of text" "yet more text")
9996 and the @code{kill-ring-yank-pointer} points to the second clause, the
9997 value of @code{kill-ring-yank-pointer} is:
10000 ("a different piece of text" "yet more text")
10003 As explained in the previous chapter (@pxref{List Implementation}), the
10004 computer does not keep two different copies of the text being pointed to
10005 by both the @code{kill-ring} and the @code{kill-ring-yank-pointer}. The
10006 words ``a different piece of text'' and ``yet more text'' are not
10007 duplicated. Instead, the two Lisp variables point to the same pieces of
10008 text. Here is a diagram:
10010 @c cons-cell-diagram #5
10014 kill-ring kill-ring-yank-pointer
10016 | ___ ___ | ___ ___ ___ ___
10017 ---> | | | --> | | | | | |
10018 |___|___|----> |___|___|--> |___|___|--> nil
10021 | | --> "yet more text"
10023 | --> "a different piece of text"
10030 @ifset print-postscript-figures
10033 @center @image{cons-5}
10037 @ifclear print-postscript-figures
10041 kill-ring kill-ring-yank-pointer
10043 | ___ ___ | ___ ___ ___ ___
10044 ---> | | | --> | | | | | |
10045 |___|___|----> |___|___|--> |___|___|--> nil
10048 | | --> "yet more text"
10050 | --> "a different piece of text
10059 Both the variable @code{kill-ring} and the variable
10060 @code{kill-ring-yank-pointer} are pointers. But the kill ring itself is
10061 usually described as if it were actually what it is composed of. The
10062 @code{kill-ring} is spoken of as if it were the list rather than that it
10063 points to the list. Conversely, the @code{kill-ring-yank-pointer} is
10064 spoken of as pointing to a list.
10066 These two ways of talking about the same thing sound confusing at first but
10067 make sense on reflection. The kill ring is generally thought of as the
10068 complete structure of data that holds the information of what has recently
10069 been cut out of the Emacs buffers. The @code{kill-ring-yank-pointer}
10070 on the other hand, serves to indicate---that is, to ``point to''---that part
10071 of the kill ring of which the first element (the @sc{car}) will be
10075 In GNU Emacs 22, the @code{kill-new} function calls
10077 @code{(setq kill-ring-yank-pointer kill-ring)}
10079 (defun rotate-yank-pointer (arg)
10080 "Rotate the yanking point in the kill ring.
10081 With argument, rotate that many kills forward (or backward, if negative)."
10083 (current-kill arg))
10085 (defun current-kill (n &optional do-not-move)
10086 "Rotate the yanking point by N places, and then return that kill.
10087 If N is zero, `interprogram-paste-function' is set, and calling it
10088 returns a string, then that string is added to the front of the
10089 kill ring and returned as the latest kill.
10090 If optional arg DO-NOT-MOVE is non-nil, then don't actually move the
10091 yanking point; just return the Nth kill forward."
10092 (let ((interprogram-paste (and (= n 0)
10093 interprogram-paste-function
10094 (funcall interprogram-paste-function))))
10095 (if interprogram-paste
10097 ;; Disable the interprogram cut function when we add the new
10098 ;; text to the kill ring, so Emacs doesn't try to own the
10099 ;; selection, with identical text.
10100 (let ((interprogram-cut-function nil))
10101 (kill-new interprogram-paste))
10102 interprogram-paste)
10103 (or kill-ring (error "Kill ring is empty"))
10104 (let ((ARGth-kill-element
10105 (nthcdr (mod (- n (length kill-ring-yank-pointer))
10106 (length kill-ring))
10109 (setq kill-ring-yank-pointer ARGth-kill-element))
10110 (car ARGth-kill-element)))))
10115 @node yank nthcdr Exercises
10116 @section Exercises with @code{yank} and @code{nthcdr}
10120 Using @kbd{C-h v} (@code{describe-variable}), look at the value of
10121 your kill ring. Add several items to your kill ring; look at its
10122 value again. Using @kbd{M-y} (@code{yank-pop)}, move all the way
10123 around the kill ring. How many items were in your kill ring? Find
10124 the value of @code{kill-ring-max}. Was your kill ring full, or could
10125 you have kept more blocks of text within it?
10128 Using @code{nthcdr} and @code{car}, construct a series of expressions
10129 to return the first, second, third, and fourth elements of a list.
10132 @node Loops & Recursion
10133 @chapter Loops and Recursion
10134 @cindex Loops and recursion
10135 @cindex Recursion and loops
10136 @cindex Repetition (loops)
10138 Emacs Lisp has two primary ways to cause an expression, or a series of
10139 expressions, to be evaluated repeatedly: one uses a @code{while}
10140 loop, and the other uses @dfn{recursion}.
10142 Repetition can be very valuable. For example, to move forward four
10143 sentences, you need only write a program that will move forward one
10144 sentence and then repeat the process four times. Since a computer does
10145 not get bored or tired, such repetitive action does not have the
10146 deleterious effects that excessive or the wrong kinds of repetition can
10149 People mostly write Emacs Lisp functions using @code{while} loops and
10150 their kin; but you can use recursion, which provides a very powerful
10151 way to think about and then to solve problems@footnote{You can write
10152 recursive functions to be frugal or wasteful of mental or computer
10153 resources; as it happens, methods that people find easy---that are
10154 frugal of ``mental resources''---sometimes use considerable computer
10155 resources. Emacs was designed to run on machines that we now consider
10156 limited and its default settings are conservative. You may want to
10157 increase the values of @code{max-specpdl-size} and
10158 @code{max-lisp-eval-depth}. In my @file{.emacs} file, I set them to
10159 15 and 30 times their default value.}.
10162 * while:: Causing a stretch of code to repeat.
10164 * Recursion:: Causing a function to call itself.
10165 * Looping exercise::
10169 @section @code{while}
10173 The @code{while} special form tests whether the value returned by
10174 evaluating its first argument is true or false. This is similar to what
10175 the Lisp interpreter does with an @code{if}; what the interpreter does
10176 next, however, is different.
10178 In a @code{while} expression, if the value returned by evaluating the
10179 first argument is false, the Lisp interpreter skips the rest of the
10180 expression (the @dfn{body} of the expression) and does not evaluate it.
10181 However, if the value is true, the Lisp interpreter evaluates the body
10182 of the expression and then again tests whether the first argument to
10183 @code{while} is true or false. If the value returned by evaluating the
10184 first argument is again true, the Lisp interpreter again evaluates the
10185 body of the expression.
10188 The template for a @code{while} expression looks like this:
10192 (while @var{true-or-false-test}
10198 * Looping with while:: Repeat so long as test returns true.
10199 * Loop Example:: A @code{while} loop that uses a list.
10200 * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}.
10201 * Incrementing Loop:: A loop with an incrementing counter.
10202 * Incrementing Loop Details::
10203 * Decrementing Loop:: A loop with a decrementing counter.
10207 @node Looping with while
10208 @unnumberedsubsec Looping with @code{while}
10211 So long as the true-or-false-test of the @code{while} expression
10212 returns a true value when it is evaluated, the body is repeatedly
10213 evaluated. This process is called a loop since the Lisp interpreter
10214 repeats the same thing again and again, like an airplane doing a loop.
10215 When the result of evaluating the true-or-false-test is false, the
10216 Lisp interpreter does not evaluate the rest of the @code{while}
10217 expression and ``exits the loop''.
10219 Clearly, if the value returned by evaluating the first argument to
10220 @code{while} is always true, the body following will be evaluated
10221 again and again @dots{} and again @dots{} forever. Conversely, if the
10222 value returned is never true, the expressions in the body will never
10223 be evaluated. The craft of writing a @code{while} loop consists of
10224 choosing a mechanism such that the true-or-false-test returns true
10225 just the number of times that you want the subsequent expressions to
10226 be evaluated, and then have the test return false.
10228 The value returned by evaluating a @code{while} is the value of the
10229 true-or-false-test. An interesting consequence of this is that a
10230 @code{while} loop that evaluates without error will return @code{nil}
10231 or false regardless of whether it has looped 1 or 100 times or none at
10232 all. A @code{while} expression that evaluates successfully never
10233 returns a true value! What this means is that @code{while} is always
10234 evaluated for its side effects, which is to say, the consequences of
10235 evaluating the expressions within the body of the @code{while} loop.
10236 This makes sense. It is not the mere act of looping that is desired,
10237 but the consequences of what happens when the expressions in the loop
10238 are repeatedly evaluated.
10241 @subsection A @code{while} Loop and a List
10243 A common way to control a @code{while} loop is to test whether a list
10244 has any elements. If it does, the loop is repeated; but if it does not,
10245 the repetition is ended. Since this is an important technique, we will
10246 create a short example to illustrate it.
10248 A simple way to test whether a list has elements is to evaluate the
10249 list: if it has no elements, it is an empty list and will return the
10250 empty list, @code{()}, which is a synonym for @code{nil} or false. On
10251 the other hand, a list with elements will return those elements when it
10252 is evaluated. Since Emacs Lisp considers as true any value that is not
10253 @code{nil}, a list that returns elements will test true in a
10257 For example, you can set the variable @code{empty-list} to @code{nil} by
10258 evaluating the following @code{setq} expression:
10261 (setq empty-list ())
10265 After evaluating the @code{setq} expression, you can evaluate the
10266 variable @code{empty-list} in the usual way, by placing the cursor after
10267 the symbol and typing @kbd{C-x C-e}; @code{nil} will appear in your
10274 On the other hand, if you set a variable to be a list with elements, the
10275 list will appear when you evaluate the variable, as you can see by
10276 evaluating the following two expressions:
10280 (setq animals '(gazelle giraffe lion tiger))
10286 Thus, to create a @code{while} loop that tests whether there are any
10287 items in the list @code{animals}, the first part of the loop will be
10298 When the @code{while} tests its first argument, the variable
10299 @code{animals} is evaluated. It returns a list. So long as the list
10300 has elements, the @code{while} considers the results of the test to be
10301 true; but when the list is empty, it considers the results of the test
10304 To prevent the @code{while} loop from running forever, some mechanism
10305 needs to be provided to empty the list eventually. An oft-used
10306 technique is to have one of the subsequent forms in the @code{while}
10307 expression set the value of the list to be the @sc{cdr} of the list.
10308 Each time the @code{cdr} function is evaluated, the list will be made
10309 shorter, until eventually only the empty list will be left. At this
10310 point, the test of the @code{while} loop will return false, and the
10311 arguments to the @code{while} will no longer be evaluated.
10313 For example, the list of animals bound to the variable @code{animals}
10314 can be set to be the @sc{cdr} of the original list with the
10315 following expression:
10318 (setq animals (cdr animals))
10322 If you have evaluated the previous expressions and then evaluate this
10323 expression, you will see @code{(giraffe lion tiger)} appear in the echo
10324 area. If you evaluate the expression again, @code{(lion tiger)} will
10325 appear in the echo area. If you evaluate it again and yet again,
10326 @code{(tiger)} appears and then the empty list, shown by @code{nil}.
10328 A template for a @code{while} loop that uses the @code{cdr} function
10329 repeatedly to cause the true-or-false-test eventually to test false
10334 (while @var{test-whether-list-is-empty}
10336 @var{set-list-to-cdr-of-list})
10340 This test and use of @code{cdr} can be put together in a function that
10341 goes through a list and prints each element of the list on a line of its
10344 @node print-elements-of-list
10345 @subsection An Example: @code{print-elements-of-list}
10346 @findex print-elements-of-list
10348 The @code{print-elements-of-list} function illustrates a @code{while}
10351 @cindex @file{*scratch*} buffer
10352 The function requires several lines for its output. If you are
10353 reading this in a recent instance of GNU Emacs,
10354 @c GNU Emacs 21, GNU Emacs 22, or a later version,
10355 you can evaluate the following expression inside of Info, as usual.
10357 If you are using an earlier version of Emacs, you need to copy the
10358 necessary expressions to your @file{*scratch*} buffer and evaluate
10359 them there. This is because the echo area had only one line in the
10362 You can copy the expressions by marking the beginning of the region
10363 with @kbd{C-@key{SPC}} (@code{set-mark-command}), moving the cursor to
10364 the end of the region and then copying the region using @kbd{M-w}
10365 (@code{kill-ring-save}, which calls @code{copy-region-as-kill} and
10366 then provides visual feedback). In the @file{*scratch*}
10367 buffer, you can yank the expressions back by typing @kbd{C-y}
10370 After you have copied the expressions to the @file{*scratch*} buffer,
10371 evaluate each expression in turn. Be sure to evaluate the last
10372 expression, @code{(print-elements-of-list animals)}, by typing
10373 @kbd{C-u C-x C-e}, that is, by giving an argument to
10374 @code{eval-last-sexp}. This will cause the result of the evaluation
10375 to be printed in the @file{*scratch*} buffer instead of being printed
10376 in the echo area. (Otherwise you will see something like this in your
10377 echo area: @code{^Jgazelle^J^Jgiraffe^J^Jlion^J^Jtiger^Jnil}, in which
10378 each @samp{^J} stands for a ``newline''.)
10381 In a recent instance of GNU Emacs, you can evaluate these expressions
10382 directly in the Info buffer, and the echo area will grow to show the
10387 (setq animals '(gazelle giraffe lion tiger))
10389 (defun print-elements-of-list (list)
10390 "Print each element of LIST on a line of its own."
10393 (setq list (cdr list))))
10395 (print-elements-of-list animals)
10401 When you evaluate the three expressions in sequence, you will see
10417 Each element of the list is printed on a line of its own (that is what
10418 the function @code{print} does) and then the value returned by the
10419 function is printed. Since the last expression in the function is the
10420 @code{while} loop, and since @code{while} loops always return
10421 @code{nil}, a @code{nil} is printed after the last element of the list.
10423 @node Incrementing Loop
10424 @subsection A Loop with an Incrementing Counter
10426 A loop is not useful unless it stops when it ought. Besides
10427 controlling a loop with a list, a common way of stopping a loop is to
10428 write the first argument as a test that returns false when the correct
10429 number of repetitions are complete. This means that the loop must
10430 have a counter---an expression that counts how many times the loop
10434 @node Incrementing Loop Details
10435 @unnumberedsubsec Details of an Incrementing Loop
10438 The test for a loop with an incrementing counter can be an expression
10439 such as @code{(< count desired-number)} which returns @code{t} for
10440 true if the value of @code{count} is less than the
10441 @code{desired-number} of repetitions and @code{nil} for false if the
10442 value of @code{count} is equal to or is greater than the
10443 @code{desired-number}. The expression that increments the count can
10444 be a simple @code{setq} such as @code{(setq count (1+ count))}, where
10445 @code{1+} is a built-in function in Emacs Lisp that adds 1 to its
10446 argument. (The expression @w{@code{(1+ count)}} has the same result
10447 as @w{@code{(+ count 1)}}, but is easier for a human to read.)
10450 The template for a @code{while} loop controlled by an incrementing
10451 counter looks like this:
10455 @var{set-count-to-initial-value}
10456 (while (< count desired-number) ; @r{true-or-false-test}
10458 (setq count (1+ count))) ; @r{incrementer}
10463 Note that you need to set the initial value of @code{count}; usually it
10467 * Incrementing Example:: Counting pebbles in a triangle.
10468 * Inc Example parts:: The parts of the function definition.
10469 * Inc Example altogether:: Putting the function definition together.
10472 @node Incrementing Example
10473 @unnumberedsubsubsec Example with incrementing counter
10475 Suppose you are playing on the beach and decide to make a triangle of
10476 pebbles, putting one pebble in the first row, two in the second row,
10477 three in the third row and so on, like this:
10495 @bullet{} @bullet{}
10496 @bullet{} @bullet{} @bullet{}
10497 @bullet{} @bullet{} @bullet{} @bullet{}
10504 (About 2500 years ago, Pythagoras and others developed the beginnings of
10505 number theory by considering questions such as this.)
10507 Suppose you want to know how many pebbles you will need to make a
10508 triangle with 7 rows?
10510 Clearly, what you need to do is add up the numbers from 1 to 7. There
10511 are two ways to do this; start with the smallest number, one, and add up
10512 the list in sequence, 1, 2, 3, 4 and so on; or start with the largest
10513 number and add the list going down: 7, 6, 5, 4 and so on. Because both
10514 mechanisms illustrate common ways of writing @code{while} loops, we will
10515 create two examples, one counting up and the other counting down. In
10516 this first example, we will start with 1 and add 2, 3, 4 and so on.
10518 If you are just adding up a short list of numbers, the easiest way to do
10519 it is to add up all the numbers at once. However, if you do not know
10520 ahead of time how many numbers your list will have, or if you want to be
10521 prepared for a very long list, then you need to design your addition so
10522 that what you do is repeat a simple process many times instead of doing
10523 a more complex process once.
10525 For example, instead of adding up all the pebbles all at once, what you
10526 can do is add the number of pebbles in the first row, 1, to the number
10527 in the second row, 2, and then add the total of those two rows to the
10528 third row, 3. Then you can add the number in the fourth row, 4, to the
10529 total of the first three rows; and so on.
10531 The critical characteristic of the process is that each repetitive
10532 action is simple. In this case, at each step we add only two numbers,
10533 the number of pebbles in the row and the total already found. This
10534 process of adding two numbers is repeated again and again until the last
10535 row has been added to the total of all the preceding rows. In a more
10536 complex loop the repetitive action might not be so simple, but it will
10537 be simpler than doing everything all at once.
10539 @node Inc Example parts
10540 @unnumberedsubsubsec The parts of the function definition
10542 The preceding analysis gives us the bones of our function definition:
10543 first, we will need a variable that we can call @code{total} that will
10544 be the total number of pebbles. This will be the value returned by
10547 Second, we know that the function will require an argument: this
10548 argument will be the total number of rows in the triangle. It can be
10549 called @code{number-of-rows}.
10551 Finally, we need a variable to use as a counter. We could call this
10552 variable @code{counter}, but a better name is @code{row-number}. That
10553 is because what the counter does in this function is count rows, and a
10554 program should be written to be as understandable as possible.
10556 When the Lisp interpreter first starts evaluating the expressions in the
10557 function, the value of @code{total} should be set to zero, since we have
10558 not added anything to it. Then the function should add the number of
10559 pebbles in the first row to the total, and then add the number of
10560 pebbles in the second to the total, and then add the number of
10561 pebbles in the third row to the total, and so on, until there are no
10562 more rows left to add.
10564 Both @code{total} and @code{row-number} are used only inside the
10565 function, so they can be declared as local variables with @code{let}
10566 and given initial values. Clearly, the initial value for @code{total}
10567 should be 0. The initial value of @code{row-number} should be 1,
10568 since we start with the first row. This means that the @code{let}
10569 statement will look like this:
10579 After the internal variables are declared and bound to their initial
10580 values, we can begin the @code{while} loop. The expression that serves
10581 as the test should return a value of @code{t} for true so long as the
10582 @code{row-number} is less than or equal to the @code{number-of-rows}.
10583 (If the expression tests true only so long as the row number is less
10584 than the number of rows in the triangle, the last row will never be
10585 added to the total; hence the row number has to be either less than or
10586 equal to the number of rows.)
10589 @findex <= @r{(less than or equal)}
10590 Lisp provides the @code{<=} function that returns true if the value of
10591 its first argument is less than or equal to the value of its second
10592 argument and false otherwise. So the expression that the @code{while}
10593 will evaluate as its test should look like this:
10596 (<= row-number number-of-rows)
10599 The total number of pebbles can be found by repeatedly adding the number
10600 of pebbles in a row to the total already found. Since the number of
10601 pebbles in the row is equal to the row number, the total can be found by
10602 adding the row number to the total. (Clearly, in a more complex
10603 situation, the number of pebbles in the row might be related to the row
10604 number in a more complicated way; if this were the case, the row number
10605 would be replaced by the appropriate expression.)
10608 (setq total (+ total row-number))
10612 What this does is set the new value of @code{total} to be equal to the
10613 sum of adding the number of pebbles in the row to the previous total.
10615 After setting the value of @code{total}, the conditions need to be
10616 established for the next repetition of the loop, if there is one. This
10617 is done by incrementing the value of the @code{row-number} variable,
10618 which serves as a counter. After the @code{row-number} variable has
10619 been incremented, the true-or-false-test at the beginning of the
10620 @code{while} loop tests whether its value is still less than or equal to
10621 the value of the @code{number-of-rows} and if it is, adds the new value
10622 of the @code{row-number} variable to the @code{total} of the previous
10623 repetition of the loop.
10626 The built-in Emacs Lisp function @code{1+} adds 1 to a number, so the
10627 @code{row-number} variable can be incremented with this expression:
10630 (setq row-number (1+ row-number))
10633 @node Inc Example altogether
10634 @unnumberedsubsubsec Putting the function definition together
10636 We have created the parts for the function definition; now we need to
10640 First, the contents of the @code{while} expression:
10644 (while (<= row-number number-of-rows) ; @r{true-or-false-test}
10645 (setq total (+ total row-number))
10646 (setq row-number (1+ row-number))) ; @r{incrementer}
10650 Along with the @code{let} expression varlist, this very nearly
10651 completes the body of the function definition. However, it requires
10652 one final element, the need for which is somewhat subtle.
10654 The final touch is to place the variable @code{total} on a line by
10655 itself after the @code{while} expression. Otherwise, the value returned
10656 by the whole function is the value of the last expression that is
10657 evaluated in the body of the @code{let}, and this is the value
10658 returned by the @code{while}, which is always @code{nil}.
10660 This may not be evident at first sight. It almost looks as if the
10661 incrementing expression is the last expression of the whole function.
10662 But that expression is part of the body of the @code{while}; it is the
10663 last element of the list that starts with the symbol @code{while}.
10664 Moreover, the whole of the @code{while} loop is a list within the body
10668 In outline, the function will look like this:
10672 (defun @var{name-of-function} (@var{argument-list})
10673 "@var{documentation}@dots{}"
10674 (let (@var{varlist})
10675 (while (@var{true-or-false-test})
10676 @var{body-of-while}@dots{} )
10677 @dots{} )) ; @r{Need final expression here.}
10681 The result of evaluating the @code{let} is what is going to be returned
10682 by the @code{defun} since the @code{let} is not embedded within any
10683 containing list, except for the @code{defun} as a whole. However, if
10684 the @code{while} is the last element of the @code{let} expression, the
10685 function will always return @code{nil}. This is not what we want!
10686 Instead, what we want is the value of the variable @code{total}. This
10687 is returned by simply placing the symbol as the last element of the list
10688 starting with @code{let}. It gets evaluated after the preceding
10689 elements of the list are evaluated, which means it gets evaluated after
10690 it has been assigned the correct value for the total.
10692 It may be easier to see this by printing the list starting with
10693 @code{let} all on one line. This format makes it evident that the
10694 @var{varlist} and @code{while} expressions are the second and third
10695 elements of the list starting with @code{let}, and the @code{total} is
10700 (let (@var{varlist}) (while (@var{true-or-false-test}) @var{body-of-while}@dots{} ) total)
10705 Putting everything together, the @code{triangle} function definition
10710 (defun triangle (number-of-rows) ; @r{Version with}
10711 ; @r{ incrementing counter.}
10712 "Add up the number of pebbles in a triangle.
10713 The first row has one pebble, the second row two pebbles,
10714 the third row three pebbles, and so on.
10715 The argument is NUMBER-OF-ROWS."
10720 (while (<= row-number number-of-rows)
10721 (setq total (+ total row-number))
10722 (setq row-number (1+ row-number)))
10728 After you have installed @code{triangle} by evaluating the function, you
10729 can try it out. Here are two examples:
10740 The sum of the first four numbers is 10 and the sum of the first seven
10743 @node Decrementing Loop
10744 @subsection Loop with a Decrementing Counter
10746 Another common way to write a @code{while} loop is to write the test
10747 so that it determines whether a counter is greater than zero. So long
10748 as the counter is greater than zero, the loop is repeated. But when
10749 the counter is equal to or less than zero, the loop is stopped. For
10750 this to work, the counter has to start out greater than zero and then
10751 be made smaller and smaller by a form that is evaluated
10754 The test will be an expression such as @code{(> counter 0)} which
10755 returns @code{t} for true if the value of @code{counter} is greater
10756 than zero, and @code{nil} for false if the value of @code{counter} is
10757 equal to or less than zero. The expression that makes the number
10758 smaller and smaller can be a simple @code{setq} such as @code{(setq
10759 counter (1- counter))}, where @code{1-} is a built-in function in
10760 Emacs Lisp that subtracts 1 from its argument.
10763 The template for a decrementing @code{while} loop looks like this:
10767 (while (> counter 0) ; @r{true-or-false-test}
10769 (setq counter (1- counter))) ; @r{decrementer}
10774 * Decrementing Example:: More pebbles on the beach.
10775 * Dec Example parts:: The parts of the function definition.
10776 * Dec Example altogether:: Putting the function definition together.
10779 @node Decrementing Example
10780 @unnumberedsubsubsec Example with decrementing counter
10782 To illustrate a loop with a decrementing counter, we will rewrite the
10783 @code{triangle} function so the counter decreases to zero.
10785 This is the reverse of the earlier version of the function. In this
10786 case, to find out how many pebbles are needed to make a triangle with
10787 3 rows, add the number of pebbles in the third row, 3, to the number
10788 in the preceding row, 2, and then add the total of those two rows to
10789 the row that precedes them, which is 1.
10791 Likewise, to find the number of pebbles in a triangle with 7 rows, add
10792 the number of pebbles in the seventh row, 7, to the number in the
10793 preceding row, which is 6, and then add the total of those two rows to
10794 the row that precedes them, which is 5, and so on. As in the previous
10795 example, each addition only involves adding two numbers, the total of
10796 the rows already added up and the number of pebbles in the row that is
10797 being added to the total. This process of adding two numbers is
10798 repeated again and again until there are no more pebbles to add.
10800 We know how many pebbles to start with: the number of pebbles in the
10801 last row is equal to the number of rows. If the triangle has seven
10802 rows, the number of pebbles in the last row is 7. Likewise, we know how
10803 many pebbles are in the preceding row: it is one less than the number in
10806 @node Dec Example parts
10807 @unnumberedsubsubsec The parts of the function definition
10809 We start with three variables: the total number of rows in the
10810 triangle; the number of pebbles in a row; and the total number of
10811 pebbles, which is what we want to calculate. These variables can be
10812 named @code{number-of-rows}, @code{number-of-pebbles-in-row}, and
10813 @code{total}, respectively.
10815 Both @code{total} and @code{number-of-pebbles-in-row} are used only
10816 inside the function and are declared with @code{let}. The initial
10817 value of @code{total} should, of course, be zero. However, the
10818 initial value of @code{number-of-pebbles-in-row} should be equal to
10819 the number of rows in the triangle, since the addition will start with
10823 This means that the beginning of the @code{let} expression will look
10829 (number-of-pebbles-in-row number-of-rows))
10834 The total number of pebbles can be found by repeatedly adding the number
10835 of pebbles in a row to the total already found, that is, by repeatedly
10836 evaluating the following expression:
10839 (setq total (+ total number-of-pebbles-in-row))
10843 After the @code{number-of-pebbles-in-row} is added to the @code{total},
10844 the @code{number-of-pebbles-in-row} should be decremented by one, since
10845 the next time the loop repeats, the preceding row will be
10846 added to the total.
10848 The number of pebbles in a preceding row is one less than the number of
10849 pebbles in a row, so the built-in Emacs Lisp function @code{1-} can be
10850 used to compute the number of pebbles in the preceding row. This can be
10851 done with the following expression:
10855 (setq number-of-pebbles-in-row
10856 (1- number-of-pebbles-in-row))
10860 Finally, we know that the @code{while} loop should stop making repeated
10861 additions when there are no pebbles in a row. So the test for
10862 the @code{while} loop is simply:
10865 (while (> number-of-pebbles-in-row 0)
10868 @node Dec Example altogether
10869 @unnumberedsubsubsec Putting the function definition together
10871 We can put these expressions together to create a function definition
10872 that works. However, on examination, we find that one of the local
10873 variables is unneeded!
10876 The function definition looks like this:
10880 ;;; @r{First subtractive version.}
10881 (defun triangle (number-of-rows)
10882 "Add up the number of pebbles in a triangle."
10884 (number-of-pebbles-in-row number-of-rows))
10885 (while (> number-of-pebbles-in-row 0)
10886 (setq total (+ total number-of-pebbles-in-row))
10887 (setq number-of-pebbles-in-row
10888 (1- number-of-pebbles-in-row)))
10893 As written, this function works.
10895 However, we do not need @code{number-of-pebbles-in-row}.
10897 @cindex Argument as local variable
10898 When the @code{triangle} function is evaluated, the symbol
10899 @code{number-of-rows} will be bound to a number, giving it an initial
10900 value. That number can be changed in the body of the function as if
10901 it were a local variable, without any fear that such a change will
10902 effect the value of the variable outside of the function. This is a
10903 very useful characteristic of Lisp; it means that the variable
10904 @code{number-of-rows} can be used anywhere in the function where
10905 @code{number-of-pebbles-in-row} is used.
10908 Here is a second version of the function written a bit more cleanly:
10912 (defun triangle (number) ; @r{Second version.}
10913 "Return sum of numbers 1 through NUMBER inclusive."
10915 (while (> number 0)
10916 (setq total (+ total number))
10917 (setq number (1- number)))
10922 In brief, a properly written @code{while} loop will consist of three parts:
10926 A test that will return false after the loop has repeated itself the
10927 correct number of times.
10930 An expression the evaluation of which will return the value desired
10931 after being repeatedly evaluated.
10934 An expression to change the value passed to the true-or-false-test so
10935 that the test returns false after the loop has repeated itself the right
10939 @node dolist dotimes
10940 @section Save your time: @code{dolist} and @code{dotimes}
10942 In addition to @code{while}, both @code{dolist} and @code{dotimes}
10943 provide for looping. Sometimes these are quicker to write than the
10944 equivalent @code{while} loop. Both are Lisp macros. (@xref{Macros, ,
10945 Macros, elisp, The GNU Emacs Lisp Reference Manual}. )
10947 @code{dolist} works like a @code{while} loop that ``@sc{cdr}s down a
10948 list'': @code{dolist} automatically shortens the list each time it
10949 loops---takes the @sc{cdr} of the list---and binds the @sc{car} of
10950 each shorter version of the list to the first of its arguments.
10952 @code{dotimes} loops a specific number of times: you specify the number.
10960 @unnumberedsubsec The @code{dolist} Macro
10963 Suppose, for example, you want to reverse a list, so that
10964 ``first'' ``second'' ``third'' becomes ``third'' ``second'' ``first''.
10967 In practice, you would use the @code{reverse} function, like this:
10971 (setq animals '(gazelle giraffe lion tiger))
10979 Here is how you could reverse the list using a @code{while} loop:
10983 (setq animals '(gazelle giraffe lion tiger))
10985 (defun reverse-list-with-while (list)
10986 "Using while, reverse the order of LIST."
10987 (let (value) ; make sure list starts empty
10989 (setq value (cons (car list) value))
10990 (setq list (cdr list)))
10993 (reverse-list-with-while animals)
10999 And here is how you could use the @code{dolist} macro:
11003 (setq animals '(gazelle giraffe lion tiger))
11005 (defun reverse-list-with-dolist (list)
11006 "Using dolist, reverse the order of LIST."
11007 (let (value) ; make sure list starts empty
11008 (dolist (element list value)
11009 (setq value (cons element value)))))
11011 (reverse-list-with-dolist animals)
11017 In Info, you can place your cursor after the closing parenthesis of
11018 each expression and type @kbd{C-x C-e}; in each case, you should see
11021 (tiger lion giraffe gazelle)
11027 For this example, the existing @code{reverse} function is obviously best.
11028 The @code{while} loop is just like our first example (@pxref{Loop
11029 Example, , A @code{while} Loop and a List}). The @code{while} first
11030 checks whether the list has elements; if so, it constructs a new list
11031 by adding the first element of the list to the existing list (which in
11032 the first iteration of the loop is @code{nil}). Since the second
11033 element is prepended in front of the first element, and the third
11034 element is prepended in front of the second element, the list is reversed.
11036 In the expression using a @code{while} loop,
11037 the @w{@code{(setq list (cdr list))}}
11038 expression shortens the list, so the @code{while} loop eventually
11039 stops. In addition, it provides the @code{cons} expression with a new
11040 first element by creating a new and shorter list at each repetition of
11043 The @code{dolist} expression does very much the same as the
11044 @code{while} expression, except that the @code{dolist} macro does some
11045 of the work you have to do when writing a @code{while} expression.
11047 Like a @code{while} loop, a @code{dolist} loops. What is different is
11048 that it automatically shortens the list each time it loops---it
11049 ``@sc{cdr}s down the list'' on its own---and it automatically binds
11050 the @sc{car} of each shorter version of the list to the first of its
11053 In the example, the @sc{car} of each shorter version of the list is
11054 referred to using the symbol @samp{element}, the list itself is called
11055 @samp{list}, and the value returned is called @samp{value}. The
11056 remainder of the @code{dolist} expression is the body.
11058 The @code{dolist} expression binds the @sc{car} of each shorter
11059 version of the list to @code{element} and then evaluates the body of
11060 the expression; and repeats the loop. The result is returned in
11064 @unnumberedsubsec The @code{dotimes} Macro
11067 The @code{dotimes} macro is similar to @code{dolist}, except that it
11068 loops a specific number of times.
11070 The first argument to @code{dotimes} is assigned the numbers 0, 1, 2
11071 and so forth each time around the loop, and the value of the third
11072 argument is returned. You need to provide the value of the second
11073 argument, which is how many times the macro loops.
11076 For example, the following binds the numbers from 0 up to, but not
11077 including, the number 3 to the first argument, @var{number}, and then
11078 constructs a list of the three numbers. (The first number is 0, the
11079 second number is 1, and the third number is 2; this makes a total of
11080 three numbers in all, starting with zero as the first number.)
11084 (let (value) ; otherwise a value is a void variable
11085 (dotimes (number 3 value)
11086 (setq value (cons number value))))
11093 @code{dotimes} returns @code{value}, so the way to use
11094 @code{dotimes} is to operate on some expression @var{number} number of
11095 times and then return the result, either as a list or an atom.
11098 Here is an example of a @code{defun} that uses @code{dotimes} to add
11099 up the number of pebbles in a triangle.
11103 (defun triangle-using-dotimes (number-of-rows)
11104 "Using dotimes, add up the number of pebbles in a triangle."
11105 (let ((total 0)) ; otherwise a total is a void variable
11106 (dotimes (number number-of-rows total)
11107 (setq total (+ total (1+ number))))))
11109 (triangle-using-dotimes 4)
11117 A recursive function contains code that tells the Lisp interpreter to
11118 call a program that runs exactly like itself, but with slightly
11119 different arguments. The code runs exactly the same because it has
11120 the same name. However, even though the program has the same name, it
11121 is not the same entity. It is different. In the jargon, it is a
11122 different ``instance''.
11124 Eventually, if the program is written correctly, the ``slightly
11125 different arguments'' will become sufficiently different from the first
11126 arguments that the final instance will stop.
11129 * Building Robots:: Same model, different serial number ...
11130 * Recursive Definition Parts:: Walk until you stop ...
11131 * Recursion with list:: Using a list as the test whether to recurse.
11132 * Recursive triangle function::
11133 * Recursion with cond::
11134 * Recursive Patterns:: Often used templates.
11135 * No Deferment:: Don't store up work ...
11136 * No deferment solution::
11139 @node Building Robots
11140 @subsection Building Robots: Extending the Metaphor
11141 @cindex Building robots
11142 @cindex Robots, building
11144 It is sometimes helpful to think of a running program as a robot that
11145 does a job. In doing its job, a recursive function calls on a second
11146 robot to help it. The second robot is identical to the first in every
11147 way, except that the second robot helps the first and has been
11148 passed different arguments than the first.
11150 In a recursive function, the second robot may call a third; and the
11151 third may call a fourth, and so on. Each of these is a different
11152 entity; but all are clones.
11154 Since each robot has slightly different instructions---the arguments
11155 will differ from one robot to the next---the last robot should know
11158 Let's expand on the metaphor in which a computer program is a robot.
11160 A function definition provides the blueprints for a robot. When you
11161 install a function definition, that is, when you evaluate a
11162 @code{defun} macro, you install the necessary equipment to build
11163 robots. It is as if you were in a factory, setting up an assembly
11164 line. Robots with the same name are built according to the same
11165 blueprints. So they have, as it were, the same ``model number'', but a
11166 different ``serial number''.
11168 We often say that a recursive function ``calls itself''. What we mean
11169 is that the instructions in a recursive function cause the Lisp
11170 interpreter to run a different function that has the same name and
11171 does the same job as the first, but with different arguments.
11173 It is important that the arguments differ from one instance to the
11174 next; otherwise, the process will never stop.
11176 @node Recursive Definition Parts
11177 @subsection The Parts of a Recursive Definition
11178 @cindex Parts of a Recursive Definition
11179 @cindex Recursive Definition Parts
11181 A recursive function typically contains a conditional expression which
11186 A true-or-false-test that determines whether the function is called
11187 again, here called the @dfn{do-again-test}.
11190 The name of the function. When this name is called, a new instance of
11191 the function---a new robot, as it were---is created and told what to do.
11194 An expression that returns a different value each time the function is
11195 called, here called the @dfn{next-step-expression}. Consequently, the
11196 argument (or arguments) passed to the new instance of the function
11197 will be different from that passed to the previous instance. This
11198 causes the conditional expression, the @dfn{do-again-test}, to test
11199 false after the correct number of repetitions.
11202 Recursive functions can be much simpler than any other kind of
11203 function. Indeed, when people first start to use them, they often look
11204 so mysteriously simple as to be incomprehensible. Like riding a
11205 bicycle, reading a recursive function definition takes a certain knack
11206 which is hard at first but then seems simple.
11209 There are several different common recursive patterns. A very simple
11210 pattern looks like this:
11214 (defun @var{name-of-recursive-function} (@var{argument-list})
11215 "@var{documentation}@dots{}"
11216 (if @var{do-again-test}
11218 (@var{name-of-recursive-function}
11219 @var{next-step-expression})))
11223 Each time a recursive function is evaluated, a new instance of it is
11224 created and told what to do. The arguments tell the instance what to do.
11226 An argument is bound to the value of the next-step-expression. Each
11227 instance runs with a different value of the next-step-expression.
11229 The value in the next-step-expression is used in the do-again-test.
11231 The value returned by the next-step-expression is passed to the new
11232 instance of the function, which evaluates it (or some
11233 transmogrification of it) to determine whether to continue or stop.
11234 The next-step-expression is designed so that the do-again-test returns
11235 false when the function should no longer be repeated.
11237 The do-again-test is sometimes called the @dfn{stop condition},
11238 since it stops the repetitions when it tests false.
11240 @node Recursion with list
11241 @subsection Recursion with a List
11243 The example of a @code{while} loop that printed the elements of a list
11244 of numbers can be written recursively. Here is the code, including
11245 an expression to set the value of the variable @code{animals} to a list.
11247 If you are reading this in Info in Emacs, you can evaluate this
11248 expression directly in Info. Otherwise, you must copy the example
11249 to the @file{*scratch*} buffer and evaluate each expression there.
11250 Use @kbd{C-u C-x C-e} to evaluate the
11251 @code{(print-elements-recursively animals)} expression so that the
11252 results are printed in the buffer; otherwise the Lisp interpreter will
11253 try to squeeze the results into the one line of the echo area.
11255 Also, place your cursor immediately after the last closing parenthesis
11256 of the @code{print-elements-recursively} function, before the comment.
11257 Otherwise, the Lisp interpreter will try to evaluate the comment.
11259 @findex print-elements-recursively
11262 (setq animals '(gazelle giraffe lion tiger))
11264 (defun print-elements-recursively (list)
11265 "Print each element of LIST on a line of its own.
11267 (when list ; @r{do-again-test}
11268 (print (car list)) ; @r{body}
11269 (print-elements-recursively ; @r{recursive call}
11270 (cdr list)))) ; @r{next-step-expression}
11272 (print-elements-recursively animals)
11276 The @code{print-elements-recursively} function first tests whether
11277 there is any content in the list; if there is, the function prints the
11278 first element of the list, the @sc{car} of the list. Then the
11279 function ``invokes itself'', but gives itself as its argument, not the
11280 whole list, but the second and subsequent elements of the list, the
11281 @sc{cdr} of the list.
11283 Put another way, if the list is not empty, the function invokes
11284 another instance of code that is similar to the initial code, but is a
11285 different thread of execution, with different arguments than the first
11288 Put in yet another way, if the list is not empty, the first robot
11289 assembles a second robot and tells it what to do; the second robot is
11290 a different individual from the first, but is the same model.
11292 When the second evaluation occurs, the @code{when} expression is
11293 evaluated and if true, prints the first element of the list it
11294 receives as its argument (which is the second element of the original
11295 list). Then the function ``calls itself'' with the @sc{cdr} of the list
11296 it is invoked with, which (the second time around) is the @sc{cdr} of
11297 the @sc{cdr} of the original list.
11299 Note that although we say that the function ``calls itself'', what we
11300 mean is that the Lisp interpreter assembles and instructs a new
11301 instance of the program. The new instance is a clone of the first,
11302 but is a separate individual.
11304 Each time the function ``invokes itself'', it invokes itself on a
11305 shorter version of the original list. It creates a new instance that
11306 works on a shorter list.
11308 Eventually, the function invokes itself on an empty list. It creates
11309 a new instance whose argument is @code{nil}. The conditional expression
11310 tests the value of @code{list}. Since the value of @code{list} is
11311 @code{nil}, the @code{when} expression tests false so the then-part is
11312 not evaluated. The function as a whole then returns @code{nil}.
11315 When you evaluate the expression @code{(print-elements-recursively
11316 animals)} in the @file{*scratch*} buffer, you see this result:
11332 @node Recursive triangle function
11333 @subsection Recursion in Place of a Counter
11334 @findex triangle-recursively
11337 The @code{triangle} function described in a previous section can also
11338 be written recursively. It looks like this:
11342 (defun triangle-recursively (number)
11343 "Return the sum of the numbers 1 through NUMBER inclusive.
11345 (if (= number 1) ; @r{do-again-test}
11347 (+ number ; @r{else-part}
11348 (triangle-recursively ; @r{recursive call}
11349 (1- number))))) ; @r{next-step-expression}
11351 (triangle-recursively 7)
11356 You can install this function by evaluating it and then try it by
11357 evaluating @code{(triangle-recursively 7)}. (Remember to put your
11358 cursor immediately after the last parenthesis of the function
11359 definition, before the comment.) The function evaluates to 28.
11361 To understand how this function works, let's consider what happens in the
11362 various cases when the function is passed 1, 2, 3, or 4 as the value of
11366 * Recursive Example arg of 1 or 2::
11367 * Recursive Example arg of 3 or 4::
11371 @node Recursive Example arg of 1 or 2
11372 @unnumberedsubsubsec An argument of 1 or 2
11375 First, what happens if the value of the argument is 1?
11377 The function has an @code{if} expression after the documentation
11378 string. It tests whether the value of @code{number} is equal to 1; if
11379 so, Emacs evaluates the then-part of the @code{if} expression, which
11380 returns the number 1 as the value of the function. (A triangle with
11381 one row has one pebble in it.)
11383 Suppose, however, that the value of the argument is 2. In this case,
11384 Emacs evaluates the else-part of the @code{if} expression.
11387 The else-part consists of an addition, the recursive call to
11388 @code{triangle-recursively} and a decrementing action; and it looks like
11392 (+ number (triangle-recursively (1- number)))
11395 When Emacs evaluates this expression, the innermost expression is
11396 evaluated first; then the other parts in sequence. Here are the steps
11400 @item Step 1 @w{ } Evaluate the innermost expression.
11402 The innermost expression is @code{(1- number)} so Emacs decrements the
11403 value of @code{number} from 2 to 1.
11405 @item Step 2 @w{ } Evaluate the @code{triangle-recursively} function.
11407 The Lisp interpreter creates an individual instance of
11408 @code{triangle-recursively}. It does not matter that this function is
11409 contained within itself. Emacs passes the result Step 1 as the
11410 argument used by this instance of the @code{triangle-recursively}
11413 In this case, Emacs evaluates @code{triangle-recursively} with an
11414 argument of 1. This means that this evaluation of
11415 @code{triangle-recursively} returns 1.
11417 @item Step 3 @w{ } Evaluate the value of @code{number}.
11419 The variable @code{number} is the second element of the list that
11420 starts with @code{+}; its value is 2.
11422 @item Step 4 @w{ } Evaluate the @code{+} expression.
11424 The @code{+} expression receives two arguments, the first
11425 from the evaluation of @code{number} (Step 3) and the second from the
11426 evaluation of @code{triangle-recursively} (Step 2).
11428 The result of the addition is the sum of 2 plus 1, and the number 3 is
11429 returned, which is correct. A triangle with two rows has three
11433 @node Recursive Example arg of 3 or 4
11434 @unnumberedsubsubsec An argument of 3 or 4
11436 Suppose that @code{triangle-recursively} is called with an argument of
11440 @item Step 1 @w{ } Evaluate the do-again-test.
11442 The @code{if} expression is evaluated first. This is the do-again
11443 test and returns false, so the else-part of the @code{if} expression
11444 is evaluated. (Note that in this example, the do-again-test causes
11445 the function to call itself when it tests false, not when it tests
11448 @item Step 2 @w{ } Evaluate the innermost expression of the else-part.
11450 The innermost expression of the else-part is evaluated, which decrements
11451 3 to 2. This is the next-step-expression.
11453 @item Step 3 @w{ } Evaluate the @code{triangle-recursively} function.
11455 The number 2 is passed to the @code{triangle-recursively} function.
11457 We already know what happens when Emacs evaluates @code{triangle-recursively} with
11458 an argument of 2. After going through the sequence of actions described
11459 earlier, it returns a value of 3. So that is what will happen here.
11461 @item Step 4 @w{ } Evaluate the addition.
11463 3 will be passed as an argument to the addition and will be added to the
11464 number with which the function was called, which is 3.
11468 The value returned by the function as a whole will be 6.
11470 Now that we know what will happen when @code{triangle-recursively} is
11471 called with an argument of 3, it is evident what will happen if it is
11472 called with an argument of 4:
11476 In the recursive call, the evaluation of
11479 (triangle-recursively (1- 4))
11484 will return the value of evaluating
11487 (triangle-recursively 3)
11491 which is 6 and this value will be added to 4 by the addition in the
11496 The value returned by the function as a whole will be 10.
11498 Each time @code{triangle-recursively} is evaluated, it evaluates a
11499 version of itself---a different instance of itself---with a smaller
11500 argument, until the argument is small enough so that it does not
11503 Note that this particular design for a recursive function
11504 requires that operations be deferred.
11506 Before @code{(triangle-recursively 7)} can calculate its answer, it
11507 must call @code{(triangle-recursively 6)}; and before
11508 @code{(triangle-recursively 6)} can calculate its answer, it must call
11509 @code{(triangle-recursively 5)}; and so on. That is to say, the
11510 calculation that @code{(triangle-recursively 7)} makes must be
11511 deferred until @code{(triangle-recursively 6)} makes its calculation;
11512 and @code{(triangle-recursively 6)} must defer until
11513 @code{(triangle-recursively 5)} completes; and so on.
11515 If each of these instances of @code{triangle-recursively} are thought
11516 of as different robots, the first robot must wait for the second to
11517 complete its job, which must wait until the third completes, and so
11520 There is a way around this kind of waiting, which we will discuss in
11521 @ref{No Deferment, , Recursion without Deferments}.
11523 @node Recursion with cond
11524 @subsection Recursion Example Using @code{cond}
11527 The version of @code{triangle-recursively} described earlier is written
11528 with the @code{if} special form. It can also be written using another
11529 special form called @code{cond}. The name of the special form
11530 @code{cond} is an abbreviation of the word @samp{conditional}.
11532 Although the @code{cond} special form is not used as often in the
11533 Emacs Lisp sources as @code{if}, it is used often enough to justify
11537 The template for a @code{cond} expression looks like this:
11547 where the @var{body} is a series of lists.
11550 Written out more fully, the template looks like this:
11555 (@var{first-true-or-false-test} @var{first-consequent})
11556 (@var{second-true-or-false-test} @var{second-consequent})
11557 (@var{third-true-or-false-test} @var{third-consequent})
11562 When the Lisp interpreter evaluates the @code{cond} expression, it
11563 evaluates the first element (the @sc{car} or true-or-false-test) of
11564 the first expression in a series of expressions within the body of the
11567 If the true-or-false-test returns @code{nil} the rest of that
11568 expression, the consequent, is skipped and the true-or-false-test of the
11569 next expression is evaluated. When an expression is found whose
11570 true-or-false-test returns a value that is not @code{nil}, the
11571 consequent of that expression is evaluated. The consequent can be one
11572 or more expressions. If the consequent consists of more than one
11573 expression, the expressions are evaluated in sequence and the value of
11574 the last one is returned. If the expression does not have a consequent,
11575 the value of the true-or-false-test is returned.
11577 If none of the true-or-false-tests test true, the @code{cond} expression
11578 returns @code{nil}.
11581 Written using @code{cond}, the @code{triangle} function looks like this:
11585 (defun triangle-using-cond (number)
11586 (cond ((<= number 0) 0)
11589 (+ number (triangle-using-cond (1- number))))))
11594 In this example, the @code{cond} returns 0 if the number is less than or
11595 equal to 0, it returns 1 if the number is 1 and it evaluates @code{(+
11596 number (triangle-using-cond (1- number)))} if the number is greater than
11599 @node Recursive Patterns
11600 @subsection Recursive Patterns
11601 @cindex Recursive Patterns
11603 Here are three common recursive patterns. Each involves a list.
11604 Recursion does not need to involve lists, but Lisp is designed for lists
11605 and this provides a sense of its primal capabilities.
11614 @unnumberedsubsubsec Recursive Pattern: @emph{every}
11615 @cindex Every, type of recursive pattern
11616 @cindex Recursive pattern - every
11618 In the @code{every} recursive pattern, an action is performed on every
11622 The basic pattern is:
11626 If a list be empty, return @code{nil}.
11628 Else, act on the beginning of the list (the @sc{car} of the list)
11631 through a recursive call by the function on the rest (the
11632 @sc{cdr}) of the list,
11634 and, optionally, combine the acted-on element, using @code{cons},
11635 with the results of acting on the rest.
11644 (defun square-each (numbers-list)
11645 "Square each of a NUMBERS LIST, recursively."
11646 (if (not numbers-list) ; do-again-test
11649 (* (car numbers-list) (car numbers-list))
11650 (square-each (cdr numbers-list))))) ; next-step-expression
11654 (square-each '(1 2 3))
11661 If @code{numbers-list} is empty, do nothing. But if it has content,
11662 construct a list combining the square of the first number in the list
11663 with the result of the recursive call.
11665 (The example follows the pattern exactly: @code{nil} is returned if
11666 the numbers' list is empty. In practice, you would write the
11667 conditional so it carries out the action when the numbers' list is not
11670 The @code{print-elements-recursively} function (@pxref{Recursion with
11671 list, , Recursion with a List}) is another example of an @code{every}
11672 pattern, except in this case, rather than bring the results together
11673 using @code{cons}, we print each element of output.
11676 The @code{print-elements-recursively} function looks like this:
11680 (setq animals '(gazelle giraffe lion tiger))
11684 (defun print-elements-recursively (list)
11685 "Print each element of LIST on a line of its own.
11687 (when list ; @r{do-again-test}
11688 (print (car list)) ; @r{body}
11689 (print-elements-recursively ; @r{recursive call}
11690 (cdr list)))) ; @r{next-step-expression}
11692 (print-elements-recursively animals)
11697 The pattern for @code{print-elements-recursively} is:
11701 When the list is empty, do nothing.
11703 But when the list has at least one element,
11706 act on the beginning of the list (the @sc{car} of the list),
11708 and make a recursive call on the rest (the @sc{cdr}) of the list.
11713 @unnumberedsubsubsec Recursive Pattern: @emph{accumulate}
11714 @cindex Accumulate, type of recursive pattern
11715 @cindex Recursive pattern - accumulate
11717 Another recursive pattern is called the @code{accumulate} pattern. In
11718 the @code{accumulate} recursive pattern, an action is performed on
11719 every element of a list and the result of that action is accumulated
11720 with the results of performing the action on the other elements.
11722 This is very like the ``every'' pattern using @code{cons}, except that
11723 @code{cons} is not used, but some other combiner.
11730 If a list be empty, return zero or some other constant.
11732 Else, act on the beginning of the list (the @sc{car} of the list),
11735 and combine that acted-on element, using @code{+} or
11736 some other combining function, with
11738 a recursive call by the function on the rest (the @sc{cdr}) of the list.
11743 Here is an example:
11747 (defun add-elements (numbers-list)
11748 "Add the elements of NUMBERS-LIST together."
11749 (if (not numbers-list)
11751 (+ (car numbers-list) (add-elements (cdr numbers-list)))))
11755 (add-elements '(1 2 3 4))
11760 @xref{Files List, , Making a List of Files}, for an example of the
11761 accumulate pattern.
11764 @unnumberedsubsubsec Recursive Pattern: @emph{keep}
11765 @cindex Keep, type of recursive pattern
11766 @cindex Recursive pattern - keep
11768 A third recursive pattern is called the @code{keep} pattern.
11769 In the @code{keep} recursive pattern, each element of a list is tested;
11770 the element is acted on and the results are kept only if the element
11773 Again, this is very like the ``every'' pattern, except the element is
11774 skipped unless it meets a criterion.
11777 The pattern has three parts:
11781 If a list be empty, return @code{nil}.
11783 Else, if the beginning of the list (the @sc{car} of the list) passes
11787 act on that element and combine it, using @code{cons} with
11789 a recursive call by the function on the rest (the @sc{cdr}) of the list.
11792 Otherwise, if the beginning of the list (the @sc{car} of the list) fails
11796 skip on that element,
11798 and, recursively call the function on the rest (the @sc{cdr}) of the list.
11803 Here is an example that uses @code{cond}:
11807 (defun keep-three-letter-words (word-list)
11808 "Keep three letter words in WORD-LIST."
11810 ;; First do-again-test: stop-condition
11811 ((not word-list) nil)
11813 ;; Second do-again-test: when to act
11814 ((eq 3 (length (symbol-name (car word-list))))
11815 ;; combine acted-on element with recursive call on shorter list
11816 (cons (car word-list) (keep-three-letter-words (cdr word-list))))
11818 ;; Third do-again-test: when to skip element;
11819 ;; recursively call shorter list with next-step expression
11820 (t (keep-three-letter-words (cdr word-list)))))
11824 (keep-three-letter-words '(one two three four five six))
11825 @result{} (one two six)
11829 It goes without saying that you need not use @code{nil} as the test for
11830 when to stop; and you can, of course, combine these patterns.
11833 @subsection Recursion without Deferments
11834 @cindex Deferment in recursion
11835 @cindex Recursion without Deferments
11837 Let's consider again what happens with the @code{triangle-recursively}
11838 function. We will find that the intermediate calculations are
11839 deferred until all can be done.
11842 Here is the function definition:
11846 (defun triangle-recursively (number)
11847 "Return the sum of the numbers 1 through NUMBER inclusive.
11849 (if (= number 1) ; @r{do-again-test}
11851 (+ number ; @r{else-part}
11852 (triangle-recursively ; @r{recursive call}
11853 (1- number))))) ; @r{next-step-expression}
11857 What happens when we call this function with a argument of 7?
11859 The first instance of the @code{triangle-recursively} function adds
11860 the number 7 to the value returned by a second instance of
11861 @code{triangle-recursively}, an instance that has been passed an
11862 argument of 6. That is to say, the first calculation is:
11865 (+ 7 (triangle-recursively 6))
11869 The first instance of @code{triangle-recursively}---you may want to
11870 think of it as a little robot---cannot complete its job. It must hand
11871 off the calculation for @code{(triangle-recursively 6)} to a second
11872 instance of the program, to a second robot. This second individual is
11873 completely different from the first one; it is, in the jargon, a
11874 ``different instantiation''. Or, put another way, it is a different
11875 robot. It is the same model as the first; it calculates triangle
11876 numbers recursively; but it has a different serial number.
11878 And what does @code{(triangle-recursively 6)} return? It returns the
11879 number 6 added to the value returned by evaluating
11880 @code{triangle-recursively} with an argument of 5. Using the robot
11881 metaphor, it asks yet another robot to help it.
11887 (+ 7 6 (triangle-recursively 5))
11891 And what happens next?
11894 (+ 7 6 5 (triangle-recursively 4))
11897 Each time @code{triangle-recursively} is called, except for the last
11898 time, it creates another instance of the program---another robot---and
11899 asks it to make a calculation.
11902 Eventually, the full addition is set up and performed:
11908 This design for the function defers the calculation of the first step
11909 until the second can be done, and defers that until the third can be
11910 done, and so on. Each deferment means the computer must remember what
11911 is being waited on. This is not a problem when there are only a few
11912 steps, as in this example. But it can be a problem when there are
11915 @node No deferment solution
11916 @subsection No Deferment Solution
11917 @cindex No deferment solution
11918 @cindex Solution without deferment
11920 The solution to the problem of deferred operations is to write in a
11921 manner that does not defer operations@footnote{The phrase @dfn{tail
11922 recursive} is used to describe such a process, one that uses
11923 ``constant space''.}. This requires
11924 writing to a different pattern, often one that involves writing two
11925 function definitions, an ``initialization'' function and a ``helper''
11928 The ``initialization'' function sets up the job; the ``helper'' function
11932 Here are the two function definitions for adding up numbers. They are
11933 so simple, I find them hard to understand.
11937 (defun triangle-initialization (number)
11938 "Return the sum of the numbers 1 through NUMBER inclusive.
11939 This is the `initialization' component of a two function
11940 duo that uses recursion."
11941 (triangle-recursive-helper 0 0 number))
11947 (defun triangle-recursive-helper (sum counter number)
11948 "Return SUM, using COUNTER, through NUMBER inclusive.
11949 This is the `helper' component of a two function duo
11950 that uses recursion."
11951 (if (> counter number)
11953 (triangle-recursive-helper (+ sum counter) ; @r{sum}
11954 (1+ counter) ; @r{counter}
11955 number))) ; @r{number}
11960 Install both function definitions by evaluating them, then call
11961 @code{triangle-initialization} with 2 rows:
11965 (triangle-initialization 2)
11970 The ``initialization'' function calls the first instance of the ``helper''
11971 function with three arguments: zero, zero, and a number which is the
11972 number of rows in the triangle.
11974 The first two arguments passed to the ``helper'' function are
11975 initialization values. These values are changed when
11976 @code{triangle-recursive-helper} invokes new instances.@footnote{The
11977 jargon is mildly confusing: @code{triangle-recursive-helper} uses a
11978 process that is iterative in a procedure that is recursive. The
11979 process is called iterative because the computer need only record the
11980 three values, @code{sum}, @code{counter}, and @code{number}; the
11981 procedure is recursive because the function ``calls itself''. On the
11982 other hand, both the process and the procedure used by
11983 @code{triangle-recursively} are called recursive. The word
11984 ``recursive'' has different meanings in the two contexts.}
11986 Let's see what happens when we have a triangle that has one row. (This
11987 triangle will have one pebble in it!)
11990 @code{triangle-initialization} will call its helper with
11991 the arguments @w{@code{0 0 1}}. That function will run the conditional
11992 test whether @code{(> counter number)}:
12000 and find that the result is false, so it will invoke
12001 the else-part of the @code{if} clause:
12005 (triangle-recursive-helper
12006 (+ sum counter) ; @r{sum plus counter} @result{} @r{sum}
12007 (1+ counter) ; @r{increment counter} @result{} @r{counter}
12008 number) ; @r{number stays the same}
12014 which will first compute:
12018 (triangle-recursive-helper (+ 0 0) ; @r{sum}
12019 (1+ 0) ; @r{counter}
12023 (triangle-recursive-helper 0 1 1)
12027 Again, @code{(> counter number)} will be false, so again, the Lisp
12028 interpreter will evaluate @code{triangle-recursive-helper}, creating a
12029 new instance with new arguments.
12032 This new instance will be;
12036 (triangle-recursive-helper
12037 (+ sum counter) ; @r{sum plus counter} @result{} @r{sum}
12038 (1+ counter) ; @r{increment counter} @result{} @r{counter}
12039 number) ; @r{number stays the same}
12043 (triangle-recursive-helper 1 2 1)
12047 In this case, the @code{(> counter number)} test will be true! So the
12048 instance will return the value of the sum, which will be 1, as
12051 Now, let's pass @code{triangle-initialization} an argument
12052 of 2, to find out how many pebbles there are in a triangle with two rows.
12054 That function calls @code{(triangle-recursive-helper 0 0 2)}.
12057 In stages, the instances called will be:
12061 @r{sum counter number}
12062 (triangle-recursive-helper 0 1 2)
12064 (triangle-recursive-helper 1 2 2)
12066 (triangle-recursive-helper 3 3 2)
12070 When the last instance is called, the @code{(> counter number)} test
12071 will be true, so the instance will return the value of @code{sum},
12074 This kind of pattern helps when you are writing functions that can use
12075 many resources in a computer.
12078 @node Looping exercise
12079 @section Looping Exercise
12083 Write a function similar to @code{triangle} in which each row has a
12084 value which is the square of the row number. Use a @code{while} loop.
12087 Write a function similar to @code{triangle} that multiplies instead of
12091 Rewrite these two functions recursively. Rewrite these functions
12094 @c comma in printed title causes problem in Info cross reference
12096 Write a function for Texinfo mode that creates an index entry at the
12097 beginning of a paragraph for every @samp{@@dfn} within the paragraph.
12098 (In a Texinfo file, @samp{@@dfn} marks a definition. This book is
12099 written in Texinfo.)
12101 Many of the functions you will need are described in two of the
12102 previous chapters, @ref{Cutting & Storing Text, , Cutting and Storing
12103 Text}, and @ref{Yanking, , Yanking Text Back}. If you use
12104 @code{forward-paragraph} to put the index entry at the beginning of
12105 the paragraph, you will have to use @w{@kbd{C-h f}}
12106 (@code{describe-function}) to find out how to make the command go
12109 For more information, see
12111 @ref{Indicating, , Indicating Definitions, texinfo}.
12114 @ref{Indicating, , Indicating, texinfo, Texinfo Manual}, which goes to
12115 a Texinfo manual in the current directory. Or, if you are on the
12117 @uref{http://www.gnu.org/software/texinfo/manual/texinfo/}
12120 ``Indicating Definitions, Commands, etc.''@: in @cite{Texinfo, The GNU
12121 Documentation Format}.
12125 @node Regexp Search
12126 @chapter Regular Expression Searches
12127 @cindex Searches, illustrating
12128 @cindex Regular expression searches
12129 @cindex Patterns, searching for
12130 @cindex Motion by sentence and paragraph
12131 @cindex Sentences, movement by
12132 @cindex Paragraphs, movement by
12134 Regular expression searches are used extensively in GNU Emacs. The
12135 two functions, @code{forward-sentence} and @code{forward-paragraph},
12136 illustrate these searches well. They use regular expressions to find
12137 where to move point. The phrase ``regular expression'' is often written
12140 Regular expression searches are described in @ref{Regexp Search, ,
12141 Regular Expression Search, emacs, The GNU Emacs Manual}, as well as in
12142 @ref{Regular Expressions, , , elisp, The GNU Emacs Lisp Reference
12143 Manual}. In writing this chapter, I am presuming that you have at
12144 least a mild acquaintance with them. The major point to remember is
12145 that regular expressions permit you to search for patterns as well as
12146 for literal strings of characters. For example, the code in
12147 @code{forward-sentence} searches for the pattern of possible
12148 characters that could mark the end of a sentence, and moves point to
12151 Before looking at the code for the @code{forward-sentence} function, it
12152 is worth considering what the pattern that marks the end of a sentence
12153 must be. The pattern is discussed in the next section; following that
12154 is a description of the regular expression search function,
12155 @code{re-search-forward}. The @code{forward-sentence} function
12156 is described in the section following. Finally, the
12157 @code{forward-paragraph} function is described in the last section of
12158 this chapter. @code{forward-paragraph} is a complex function that
12159 introduces several new features.
12162 * sentence-end:: The regular expression for @code{sentence-end}.
12163 * re-search-forward:: Very similar to @code{search-forward}.
12164 * forward-sentence:: A straightforward example of regexp search.
12165 * forward-paragraph:: A somewhat complex example.
12166 * etags:: How to create your own @file{TAGS} table.
12168 * re-search Exercises::
12172 @section The Regular Expression for @code{sentence-end}
12173 @findex sentence-end
12175 The symbol @code{sentence-end} is bound to the pattern that marks the
12176 end of a sentence. What should this regular expression be?
12178 Clearly, a sentence may be ended by a period, a question mark, or an
12179 exclamation mark. Indeed, in English, only clauses that end with one
12180 of those three characters should be considered the end of a sentence.
12181 This means that the pattern should include the character set:
12187 However, we do not want @code{forward-sentence} merely to jump to a
12188 period, a question mark, or an exclamation mark, because such a character
12189 might be used in the middle of a sentence. A period, for example, is
12190 used after abbreviations. So other information is needed.
12192 According to convention, you type two spaces after every sentence, but
12193 only one space after a period, a question mark, or an exclamation mark in
12194 the body of a sentence. So a period, a question mark, or an exclamation
12195 mark followed by two spaces is a good indicator of an end of sentence.
12196 However, in a file, the two spaces may instead be a tab or the end of a
12197 line. This means that the regular expression should include these three
12198 items as alternatives.
12201 This group of alternatives will look like this:
12212 Here, @samp{$} indicates the end of the line, and I have pointed out
12213 where the tab and two spaces are inserted in the expression. Both are
12214 inserted by putting the actual characters into the expression.
12216 Two backslashes, @samp{\\}, are required before the parentheses and
12217 vertical bars: the first backslash quotes the following backslash in
12218 Emacs; and the second indicates that the following character, the
12219 parenthesis or the vertical bar, is special.
12222 Also, a sentence may be followed by one or more carriage returns, like
12233 Like tabs and spaces, a carriage return is inserted into a regular
12234 expression by inserting it literally. The asterisk indicates that the
12235 @key{RET} is repeated zero or more times.
12237 But a sentence end does not consist only of a period, a question mark or
12238 an exclamation mark followed by appropriate space: a closing quotation
12239 mark or a closing brace of some kind may precede the space. Indeed more
12240 than one such mark or brace may precede the space. These require a
12241 expression that looks like this:
12247 In this expression, the first @samp{]} is the first character in the
12248 expression; the second character is @samp{"}, which is preceded by a
12249 @samp{\} to tell Emacs the @samp{"} is @emph{not} special. The last
12250 three characters are @samp{'}, @samp{)}, and @samp{@}}.
12252 All this suggests what the regular expression pattern for matching the
12253 end of a sentence should be; and, indeed, if we evaluate
12254 @code{sentence-end} we find that it returns the following value:
12259 @result{} "[.?!][]\"')@}]*\\($\\| \\| \\)[
12265 (Well, not in GNU Emacs 22; that is because of an effort to make the
12266 process simpler and to handle more glyphs and languages. When the
12267 value of @code{sentence-end} is @code{nil}, then use the value defined
12268 by the function @code{sentence-end}. (Here is a use of the difference
12269 between a value and a function in Emacs Lisp.) The function returns a
12270 value constructed from the variables @code{sentence-end-base},
12271 @code{sentence-end-double-space}, @code{sentence-end-without-period},
12272 and @code{sentence-end-without-space}. The critical variable is
12273 @code{sentence-end-base}; its global value is similar to the one
12274 described above but it also contains two additional quotation marks.
12275 These have differing degrees of curliness. The
12276 @code{sentence-end-without-period} variable, when true, tells Emacs
12277 that a sentence may end without a period, such as text in Thai.)
12281 (Note that here the @key{TAB}, two spaces, and @key{RET} are shown
12282 literally in the pattern.)
12284 This regular expression can be deciphered as follows:
12288 The first part of the pattern is the three characters, a period, a question
12289 mark and an exclamation mark, within square brackets. The pattern must
12290 begin with one or other of these characters.
12293 The second part of the pattern is the group of closing braces and
12294 quotation marks, which can appear zero or more times. These may follow
12295 the period, question mark or exclamation mark. In a regular expression,
12296 the backslash, @samp{\}, followed by the double quotation mark,
12297 @samp{"}, indicates the class of string-quote characters. Usually, the
12298 double quotation mark is the only character in this class. The
12299 asterisk, @samp{*}, indicates that the items in the previous group (the
12300 group surrounded by square brackets, @samp{[]}) may be repeated zero or
12303 @item \\($\\| \\| \\)
12304 The third part of the pattern is one or other of: either the end of a
12305 line, or two blank spaces, or a tab. The double back-slashes are used
12306 to prevent Emacs from reading the parentheses and vertical bars as part
12307 of the search pattern; the parentheses are used to mark the group and
12308 the vertical bars are used to indicated that the patterns to either side
12309 of them are alternatives. The dollar sign is used to indicate the end
12310 of a line and both the two spaces and the tab are each inserted as is to
12311 indicate what they are.
12314 Finally, the last part of the pattern indicates that the end of the line
12315 or the whitespace following the period, question mark or exclamation
12316 mark may, but need not, be followed by one or more carriage returns. In
12317 the pattern, the carriage return is inserted as an actual carriage
12318 return between square brackets but here it is shown as @key{RET}.
12322 @node re-search-forward
12323 @section The @code{re-search-forward} Function
12324 @findex re-search-forward
12326 The @code{re-search-forward} function is very like the
12327 @code{search-forward} function. (@xref{search-forward, , The
12328 @code{search-forward} Function}.)
12330 @code{re-search-forward} searches for a regular expression. If the
12331 search is successful, it leaves point immediately after the last
12332 character in the target. If the search is backwards, it leaves point
12333 just before the first character in the target. You may tell
12334 @code{re-search-forward} to return @code{t} for true. (Moving point
12335 is therefore a ``side effect''.)
12337 Like @code{search-forward}, the @code{re-search-forward} function takes
12342 The first argument is the regular expression that the function searches
12343 for. The regular expression will be a string between quotation marks.
12346 The optional second argument limits how far the function will search; it is a
12347 bound, which is specified as a position in the buffer.
12350 The optional third argument specifies how the function responds to
12351 failure: @code{nil} as the third argument causes the function to
12352 signal an error (and print a message) when the search fails; any other
12353 value causes it to return @code{nil} if the search fails and @code{t}
12354 if the search succeeds.
12357 The optional fourth argument is the repeat count. A negative repeat
12358 count causes @code{re-search-forward} to search backwards.
12362 The template for @code{re-search-forward} looks like this:
12366 (re-search-forward "@var{regular-expression}"
12367 @var{limit-of-search}
12368 @var{what-to-do-if-search-fails}
12369 @var{repeat-count})
12373 The second, third, and fourth arguments are optional. However, if you
12374 want to pass a value to either or both of the last two arguments, you
12375 must also pass a value to all the preceding arguments. Otherwise, the
12376 Lisp interpreter will mistake which argument you are passing the value
12380 In the @code{forward-sentence} function, the regular expression will be
12381 the value of the variable @code{sentence-end}. In simple form, that is:
12385 "[.?!][]\"')@}]*\\($\\| \\| \\)[
12391 The limit of the search will be the end of the paragraph (since a
12392 sentence cannot go beyond a paragraph). If the search fails, the
12393 function will return @code{nil}; and the repeat count will be provided
12394 by the argument to the @code{forward-sentence} function.
12396 @node forward-sentence
12397 @section @code{forward-sentence}
12398 @findex forward-sentence
12400 The command to move the cursor forward a sentence is a straightforward
12401 illustration of how to use regular expression searches in Emacs Lisp.
12402 Indeed, the function looks longer and more complicated than it is; this
12403 is because the function is designed to go backwards as well as forwards;
12404 and, optionally, over more than one sentence. The function is usually
12405 bound to the key command @kbd{M-e}.
12408 * Complete forward-sentence::
12409 * fwd-sentence while loops:: Two @code{while} loops.
12410 * fwd-sentence re-search:: A regular expression search.
12414 @node Complete forward-sentence
12415 @unnumberedsubsec Complete @code{forward-sentence} function definition
12419 Here is the code for @code{forward-sentence}:
12424 (defun forward-sentence (&optional arg)
12425 "Move forward to next `sentence-end'. With argument, repeat.
12426 With negative argument, move backward repeatedly to `sentence-beginning'.
12428 The variable `sentence-end' is a regular expression that matches ends of
12429 sentences. Also, every paragraph boundary terminates sentences as well."
12433 (or arg (setq arg 1))
12434 (let ((opoint (point))
12435 (sentence-end (sentence-end)))
12437 (let ((pos (point))
12438 (par-beg (save-excursion (start-of-paragraph-text) (point))))
12439 (if (and (re-search-backward sentence-end par-beg t)
12440 (or (< (match-end 0) pos)
12441 (re-search-backward sentence-end par-beg t)))
12442 (goto-char (match-end 0))
12443 (goto-char par-beg)))
12444 (setq arg (1+ arg)))
12448 (let ((par-end (save-excursion (end-of-paragraph-text) (point))))
12449 (if (re-search-forward sentence-end par-end t)
12450 (skip-chars-backward " \t\n")
12451 (goto-char par-end)))
12452 (setq arg (1- arg)))
12453 (constrain-to-field nil opoint t)))
12461 (defun forward-sentence (&optional arg)
12462 "Move forward to next sentence-end. With argument, repeat.
12463 With negative argument, move backward repeatedly to sentence-beginning.
12464 Sentence ends are identified by the value of sentence-end
12465 treated as a regular expression. Also, every paragraph boundary
12466 terminates sentences as well."
12470 (or arg (setq arg 1))
12473 (save-excursion (start-of-paragraph-text) (point))))
12474 (if (re-search-backward
12475 (concat sentence-end "[^ \t\n]") par-beg t)
12476 (goto-char (1- (match-end 0)))
12477 (goto-char par-beg)))
12478 (setq arg (1+ arg)))
12481 (save-excursion (end-of-paragraph-text) (point))))
12482 (if (re-search-forward sentence-end par-end t)
12483 (skip-chars-backward " \t\n")
12484 (goto-char par-end)))
12485 (setq arg (1- arg))))
12490 The function looks long at first sight and it is best to look at its
12491 skeleton first, and then its muscle. The way to see the skeleton is to
12492 look at the expressions that start in the left-most columns:
12496 (defun forward-sentence (&optional arg)
12497 "@var{documentation}@dots{}"
12499 (or arg (setq arg 1))
12500 (let ((opoint (point)) (sentence-end (sentence-end)))
12502 (let ((pos (point))
12503 (par-beg (save-excursion (start-of-paragraph-text) (point))))
12504 @var{rest-of-body-of-while-loop-when-going-backwards}
12506 (let ((par-end (save-excursion (end-of-paragraph-text) (point))))
12507 @var{rest-of-body-of-while-loop-when-going-forwards}
12508 @var{handle-forms-and-equivalent}
12512 This looks much simpler! The function definition consists of
12513 documentation, an @code{interactive} expression, an @code{or}
12514 expression, a @code{let} expression, and @code{while} loops.
12516 Let's look at each of these parts in turn.
12518 We note that the documentation is thorough and understandable.
12520 The function has an @code{interactive "p"} declaration. This means
12521 that the processed prefix argument, if any, is passed to the
12522 function as its argument. (This will be a number.) If the function
12523 is not passed an argument (it is optional) then the argument
12524 @code{arg} will be bound to 1.
12526 When @code{forward-sentence} is called non-interactively without an
12527 argument, @code{arg} is bound to @code{nil}. The @code{or} expression
12528 handles this. What it does is either leave the value of @code{arg} as
12529 it is, but only if @code{arg} is bound to a value; or it sets the
12530 value of @code{arg} to 1, in the case when @code{arg} is bound to
12533 Next is a @code{let}. That specifies the values of two local
12534 variables, @code{point} and @code{sentence-end}. The local value of
12535 point, from before the search, is used in the
12536 @code{constrain-to-field} function which handles forms and
12537 equivalents. The @code{sentence-end} variable is set by the
12538 @code{sentence-end} function.
12540 @node fwd-sentence while loops
12541 @unnumberedsubsec The @code{while} loops
12543 Two @code{while} loops follow. The first @code{while} has a
12544 true-or-false-test that tests true if the prefix argument for
12545 @code{forward-sentence} is a negative number. This is for going
12546 backwards. The body of this loop is similar to the body of the second
12547 @code{while} clause, but it is not exactly the same. We will skip
12548 this @code{while} loop and concentrate on the second @code{while}
12552 The second @code{while} loop is for moving point forward. Its skeleton
12557 (while (> arg 0) ; @r{true-or-false-test}
12559 (if (@var{true-or-false-test})
12562 (setq arg (1- arg)))) ; @code{while} @r{loop decrementer}
12566 The @code{while} loop is of the decrementing kind.
12567 (@xref{Decrementing Loop, , A Loop with a Decrementing Counter}.) It
12568 has a true-or-false-test that tests true so long as the counter (in
12569 this case, the variable @code{arg}) is greater than zero; and it has a
12570 decrementer that subtracts 1 from the value of the counter every time
12573 If no prefix argument is given to @code{forward-sentence}, which is
12574 the most common way the command is used, this @code{while} loop will
12575 run once, since the value of @code{arg} will be 1.
12577 The body of the @code{while} loop consists of a @code{let} expression,
12578 which creates and binds a local variable, and has, as its body, an
12579 @code{if} expression.
12582 The body of the @code{while} loop looks like this:
12587 (save-excursion (end-of-paragraph-text) (point))))
12588 (if (re-search-forward sentence-end par-end t)
12589 (skip-chars-backward " \t\n")
12590 (goto-char par-end)))
12594 The @code{let} expression creates and binds the local variable
12595 @code{par-end}. As we shall see, this local variable is designed to
12596 provide a bound or limit to the regular expression search. If the
12597 search fails to find a proper sentence ending in the paragraph, it will
12598 stop on reaching the end of the paragraph.
12600 But first, let us examine how @code{par-end} is bound to the value of
12601 the end of the paragraph. What happens is that the @code{let} sets the
12602 value of @code{par-end} to the value returned when the Lisp interpreter
12603 evaluates the expression
12607 (save-excursion (end-of-paragraph-text) (point))
12612 In this expression, @code{(end-of-paragraph-text)} moves point to the
12613 end of the paragraph, @code{(point)} returns the value of point, and then
12614 @code{save-excursion} restores point to its original position. Thus,
12615 the @code{let} binds @code{par-end} to the value returned by the
12616 @code{save-excursion} expression, which is the position of the end of
12617 the paragraph. (The @code{end-of-paragraph-text} function uses
12618 @code{forward-paragraph}, which we will discuss shortly.)
12621 Emacs next evaluates the body of the @code{let}, which is an @code{if}
12622 expression that looks like this:
12626 (if (re-search-forward sentence-end par-end t) ; @r{if-part}
12627 (skip-chars-backward " \t\n") ; @r{then-part}
12628 (goto-char par-end))) ; @r{else-part}
12632 The @code{if} tests whether its first argument is true and if so,
12633 evaluates its then-part; otherwise, the Emacs Lisp interpreter
12634 evaluates the else-part. The true-or-false-test of the @code{if}
12635 expression is the regular expression search.
12637 It may seem odd to have what looks like the ``real work'' of
12638 the @code{forward-sentence} function buried here, but this is a common
12639 way this kind of operation is carried out in Lisp.
12641 @node fwd-sentence re-search
12642 @unnumberedsubsec The regular expression search
12644 The @code{re-search-forward} function searches for the end of the
12645 sentence, that is, for the pattern defined by the @code{sentence-end}
12646 regular expression. If the pattern is found---if the end of the sentence is
12647 found---then the @code{re-search-forward} function does two things:
12651 The @code{re-search-forward} function carries out a side effect, which
12652 is to move point to the end of the occurrence found.
12655 The @code{re-search-forward} function returns a value of true. This is
12656 the value received by the @code{if}, and means that the search was
12661 The side effect, the movement of point, is completed before the
12662 @code{if} function is handed the value returned by the successful
12663 conclusion of the search.
12665 When the @code{if} function receives the value of true from a successful
12666 call to @code{re-search-forward}, the @code{if} evaluates the then-part,
12667 which is the expression @code{(skip-chars-backward " \t\n")}. This
12668 expression moves backwards over any blank spaces, tabs or carriage
12669 returns until a printed character is found and then leaves point after
12670 the character. Since point has already been moved to the end of the
12671 pattern that marks the end of the sentence, this action leaves point
12672 right after the closing printed character of the sentence, which is
12675 On the other hand, if the @code{re-search-forward} function fails to
12676 find a pattern marking the end of the sentence, the function returns
12677 false. The false then causes the @code{if} to evaluate its third
12678 argument, which is @code{(goto-char par-end)}: it moves point to the
12679 end of the paragraph.
12681 (And if the text is in a form or equivalent, and point may not move
12682 fully, then the @code{constrain-to-field} function comes into play.)
12684 Regular expression searches are exceptionally useful and the pattern
12685 illustrated by @code{re-search-forward}, in which the search is the
12686 test of an @code{if} expression, is handy. You will see or write code
12687 incorporating this pattern often.
12689 @node forward-paragraph
12690 @section @code{forward-paragraph}: a Goldmine of Functions
12691 @findex forward-paragraph
12695 (defun forward-paragraph (&optional arg)
12696 "Move forward to end of paragraph.
12697 With argument ARG, do it ARG times;
12698 a negative argument ARG = -N means move backward N paragraphs.
12700 A line which `paragraph-start' matches either separates paragraphs
12701 \(if `paragraph-separate' matches it also) or is the first line of a paragraph.
12702 A paragraph end is the beginning of a line which is not part of the paragraph
12703 to which the end of the previous line belongs, or the end of the buffer.
12704 Returns the count of paragraphs left to move."
12706 (or arg (setq arg 1))
12707 (let* ((opoint (point))
12708 (fill-prefix-regexp
12709 (and fill-prefix (not (equal fill-prefix ""))
12710 (not paragraph-ignore-fill-prefix)
12711 (regexp-quote fill-prefix)))
12712 ;; Remove ^ from paragraph-start and paragraph-sep if they are there.
12713 ;; These regexps shouldn't be anchored, because we look for them
12714 ;; starting at the left-margin. This allows paragraph commands to
12715 ;; work normally with indented text.
12716 ;; This hack will not find problem cases like "whatever\\|^something".
12717 (parstart (if (and (not (equal "" paragraph-start))
12718 (equal ?^ (aref paragraph-start 0)))
12719 (substring paragraph-start 1)
12721 (parsep (if (and (not (equal "" paragraph-separate))
12722 (equal ?^ (aref paragraph-separate 0)))
12723 (substring paragraph-separate 1)
12724 paragraph-separate))
12726 (if fill-prefix-regexp
12727 (concat parsep "\\|"
12728 fill-prefix-regexp "[ \t]*$")
12730 ;; This is used for searching.
12731 (sp-parstart (concat "^[ \t]*\\(?:" parstart "\\|" parsep "\\)"))
12733 (while (and (< arg 0) (not (bobp)))
12734 (if (and (not (looking-at parsep))
12735 (re-search-backward "^\n" (max (1- (point)) (point-min)) t)
12736 (looking-at parsep))
12737 (setq arg (1+ arg))
12738 (setq start (point))
12739 ;; Move back over paragraph-separating lines.
12740 (forward-char -1) (beginning-of-line)
12741 (while (and (not (bobp))
12742 (progn (move-to-left-margin)
12743 (looking-at parsep)))
12747 (setq arg (1+ arg))
12748 ;; Go to end of the previous (non-separating) line.
12750 ;; Search back for line that starts or separates paragraphs.
12751 (if (if fill-prefix-regexp
12752 ;; There is a fill prefix; it overrides parstart.
12753 (let (multiple-lines)
12754 (while (and (progn (beginning-of-line) (not (bobp)))
12755 (progn (move-to-left-margin)
12756 (not (looking-at parsep)))
12757 (looking-at fill-prefix-regexp))
12758 (unless (= (point) start)
12759 (setq multiple-lines t))
12761 (move-to-left-margin)
12762 ;; This deleted code caused a long hanging-indent line
12763 ;; not to be filled together with the following lines.
12764 ;; ;; Don't move back over a line before the paragraph
12765 ;; ;; which doesn't start with fill-prefix
12766 ;; ;; unless that is the only line we've moved over.
12767 ;; (and (not (looking-at fill-prefix-regexp))
12769 ;; (forward-line 1))
12771 (while (and (re-search-backward sp-parstart nil 1)
12772 (setq found-start t)
12773 ;; Found a candidate, but need to check if it is a
12775 (progn (setq start (point))
12776 (move-to-left-margin)
12777 (not (looking-at parsep)))
12778 (not (and (looking-at parstart)
12779 (or (not use-hard-newlines)
12782 (1- start) 'hard)))))
12783 (setq found-start nil)
12788 ;; Move forward over paragraph separators.
12789 ;; We know this cannot reach the place we started
12790 ;; because we know we moved back over a non-separator.
12791 (while (and (not (eobp))
12792 (progn (move-to-left-margin)
12793 (looking-at parsep)))
12795 ;; If line before paragraph is just margin, back up to there.
12797 (if (> (current-column) (current-left-margin))
12799 (skip-chars-backward " \t")
12801 (forward-line 1))))
12802 ;; No starter or separator line => use buffer beg.
12803 (goto-char (point-min))))))
12805 (while (and (> arg 0) (not (eobp)))
12806 ;; Move forward over separator lines...
12807 (while (and (not (eobp))
12808 (progn (move-to-left-margin) (not (eobp)))
12809 (looking-at parsep))
12811 (unless (eobp) (setq arg (1- arg)))
12812 ;; ... and one more line.
12814 (if fill-prefix-regexp
12815 ;; There is a fill prefix; it overrides parstart.
12816 (while (and (not (eobp))
12817 (progn (move-to-left-margin) (not (eobp)))
12818 (not (looking-at parsep))
12819 (looking-at fill-prefix-regexp))
12821 (while (and (re-search-forward sp-parstart nil 1)
12822 (progn (setq start (match-beginning 0))
12825 (progn (move-to-left-margin)
12826 (not (looking-at parsep)))
12827 (or (not (looking-at parstart))
12828 (and use-hard-newlines
12829 (not (get-text-property (1- start) 'hard)))))
12831 (if (< (point) (point-max))
12832 (goto-char start))))
12833 (constrain-to-field nil opoint t)
12834 ;; Return the number of steps that could not be done.
12838 The @code{forward-paragraph} function moves point forward to the end
12839 of the paragraph. It is usually bound to @kbd{M-@}} and makes use of a
12840 number of functions that are important in themselves, including
12841 @code{let*}, @code{match-beginning}, and @code{looking-at}.
12843 The function definition for @code{forward-paragraph} is considerably
12844 longer than the function definition for @code{forward-sentence}
12845 because it works with a paragraph, each line of which may begin with a
12848 A fill prefix consists of a string of characters that are repeated at
12849 the beginning of each line. For example, in Lisp code, it is a
12850 convention to start each line of a paragraph-long comment with
12851 @samp{;;; }. In Text mode, four blank spaces make up another common
12852 fill prefix, creating an indented paragraph. (@xref{Fill Prefix, , ,
12853 emacs, The GNU Emacs Manual}, for more information about fill
12856 The existence of a fill prefix means that in addition to being able to
12857 find the end of a paragraph whose lines begin on the left-most
12858 column, the @code{forward-paragraph} function must be able to find the
12859 end of a paragraph when all or many of the lines in the buffer begin
12860 with the fill prefix.
12862 Moreover, it is sometimes practical to ignore a fill prefix that
12863 exists, especially when blank lines separate paragraphs.
12864 This is an added complication.
12867 * forward-paragraph in brief:: Key parts of the function definition.
12868 * fwd-para let:: The @code{let*} expression.
12869 * fwd-para while:: The forward motion @code{while} loop.
12873 @node forward-paragraph in brief
12874 @unnumberedsubsec Shortened @code{forward-paragraph} function definition
12877 Rather than print all of the @code{forward-paragraph} function, we
12878 will only print parts of it. Read without preparation, the function
12882 In outline, the function looks like this:
12886 (defun forward-paragraph (&optional arg)
12887 "@var{documentation}@dots{}"
12889 (or arg (setq arg 1))
12892 (while (and (< arg 0) (not (bobp))) ; @r{backward-moving-code}
12894 (while (and (> arg 0) (not (eobp))) ; @r{forward-moving-code}
12899 The first parts of the function are routine: the function's argument
12900 list consists of one optional argument. Documentation follows.
12902 The lower case @samp{p} in the @code{interactive} declaration means
12903 that the processed prefix argument, if any, is passed to the function.
12904 This will be a number, and is the repeat count of how many paragraphs
12905 point will move. The @code{or} expression in the next line handles
12906 the common case when no argument is passed to the function, which occurs
12907 if the function is called from other code rather than interactively.
12908 This case was described earlier. (@xref{forward-sentence, The
12909 @code{forward-sentence} function}.) Now we reach the end of the
12910 familiar part of this function.
12913 @unnumberedsubsec The @code{let*} expression
12915 The next line of the @code{forward-paragraph} function begins a
12916 @code{let*} expression. This is a different than @code{let}. The
12917 symbol is @code{let*} not @code{let}.
12919 The @code{let*} special form is like @code{let} except that Emacs sets
12920 each variable in sequence, one after another, and variables in the
12921 latter part of the varlist can make use of the values to which Emacs
12922 set variables in the earlier part of the varlist.
12925 ( refappend save-excursion, , code save-excursion in code append-to-buffer .)
12928 (@ref{append save-excursion, , @code{save-excursion} in @code{append-to-buffer}}.)
12930 In the @code{let*} expression in this function, Emacs binds a total of
12931 seven variables: @code{opoint}, @code{fill-prefix-regexp},
12932 @code{parstart}, @code{parsep}, @code{sp-parstart}, @code{start}, and
12933 @code{found-start}.
12935 The variable @code{parsep} appears twice, first, to remove instances
12936 of @samp{^}, and second, to handle fill prefixes.
12938 The variable @code{opoint} is just the value of @code{point}. As you
12939 can guess, it is used in a @code{constrain-to-field} expression, just
12940 as in @code{forward-sentence}.
12942 The variable @code{fill-prefix-regexp} is set to the value returned by
12943 evaluating the following list:
12948 (not (equal fill-prefix ""))
12949 (not paragraph-ignore-fill-prefix)
12950 (regexp-quote fill-prefix))
12955 This is an expression whose first element is the @code{and} special form.
12957 As we learned earlier (@pxref{kill-new function, , The @code{kill-new}
12958 function}), the @code{and} special form evaluates each of its
12959 arguments until one of the arguments returns a value of @code{nil}, in
12960 which case the @code{and} expression returns @code{nil}; however, if
12961 none of the arguments returns a value of @code{nil}, the value
12962 resulting from evaluating the last argument is returned. (Since such
12963 a value is not @code{nil}, it is considered true in Lisp.) In other
12964 words, an @code{and} expression returns a true value only if all its
12965 arguments are true.
12968 In this case, the variable @code{fill-prefix-regexp} is bound to a
12969 non-@code{nil} value only if the following four expressions produce a
12970 true (i.e., a non-@code{nil}) value when they are evaluated; otherwise,
12971 @code{fill-prefix-regexp} is bound to @code{nil}.
12975 When this variable is evaluated, the value of the fill prefix, if any,
12976 is returned. If there is no fill prefix, this variable returns
12979 @item (not (equal fill-prefix "")
12980 This expression checks whether an existing fill prefix is an empty
12981 string, that is, a string with no characters in it. An empty string is
12982 not a useful fill prefix.
12984 @item (not paragraph-ignore-fill-prefix)
12985 This expression returns @code{nil} if the variable
12986 @code{paragraph-ignore-fill-prefix} has been turned on by being set to a
12987 true value such as @code{t}.
12989 @item (regexp-quote fill-prefix)
12990 This is the last argument to the @code{and} special form. If all the
12991 arguments to the @code{and} are true, the value resulting from
12992 evaluating this expression will be returned by the @code{and} expression
12993 and bound to the variable @code{fill-prefix-regexp},
12996 @findex regexp-quote
12998 The result of evaluating this @code{and} expression successfully is that
12999 @code{fill-prefix-regexp} will be bound to the value of
13000 @code{fill-prefix} as modified by the @code{regexp-quote} function.
13001 What @code{regexp-quote} does is read a string and return a regular
13002 expression that will exactly match the string and match nothing else.
13003 This means that @code{fill-prefix-regexp} will be set to a value that
13004 will exactly match the fill prefix if the fill prefix exists.
13005 Otherwise, the variable will be set to @code{nil}.
13007 The next two local variables in the @code{let*} expression are
13008 designed to remove instances of @samp{^} from @code{parstart} and
13009 @code{parsep}, the local variables which indicate the paragraph start
13010 and the paragraph separator. The next expression sets @code{parsep}
13011 again. That is to handle fill prefixes.
13013 This is the setting that requires the definition call @code{let*}
13014 rather than @code{let}. The true-or-false-test for the @code{if}
13015 depends on whether the variable @code{fill-prefix-regexp} evaluates to
13016 @code{nil} or some other value.
13018 If @code{fill-prefix-regexp} does not have a value, Emacs evaluates
13019 the else-part of the @code{if} expression and binds @code{parsep} to
13020 its local value. (@code{parsep} is a regular expression that matches
13021 what separates paragraphs.)
13023 But if @code{fill-prefix-regexp} does have a value, Emacs evaluates
13024 the then-part of the @code{if} expression and binds @code{parsep} to a
13025 regular expression that includes the @code{fill-prefix-regexp} as part
13028 Specifically, @code{parsep} is set to the original value of the
13029 paragraph separate regular expression concatenated with an alternative
13030 expression that consists of the @code{fill-prefix-regexp} followed by
13031 optional whitespace to the end of the line. The whitespace is defined
13032 by @w{@code{"[ \t]*$"}}.) The @samp{\\|} defines this portion of the
13033 regexp as an alternative to @code{parsep}.
13035 According to a comment in the code, the next local variable,
13036 @code{sp-parstart}, is used for searching, and then the final two,
13037 @code{start} and @code{found-start}, are set to @code{nil}.
13039 Now we get into the body of the @code{let*}. The first part of the body
13040 of the @code{let*} deals with the case when the function is given a
13041 negative argument and is therefore moving backwards. We will skip this
13044 @node fwd-para while
13045 @unnumberedsubsec The forward motion @code{while} loop
13047 The second part of the body of the @code{let*} deals with forward
13048 motion. It is a @code{while} loop that repeats itself so long as the
13049 value of @code{arg} is greater than zero. In the most common use of
13050 the function, the value of the argument is 1, so the body of the
13051 @code{while} loop is evaluated exactly once, and the cursor moves
13052 forward one paragraph.
13055 (while (and (> arg 0) (not (eobp)))
13057 ;; Move forward over separator lines...
13058 (while (and (not (eobp))
13059 (progn (move-to-left-margin) (not (eobp)))
13060 (looking-at parsep))
13062 (unless (eobp) (setq arg (1- arg)))
13063 ;; ... and one more line.
13066 (if fill-prefix-regexp
13067 ;; There is a fill prefix; it overrides parstart.
13068 (while (and (not (eobp))
13069 (progn (move-to-left-margin) (not (eobp)))
13070 (not (looking-at parsep))
13071 (looking-at fill-prefix-regexp))
13074 (while (and (re-search-forward sp-parstart nil 1)
13075 (progn (setq start (match-beginning 0))
13078 (progn (move-to-left-margin)
13079 (not (looking-at parsep)))
13080 (or (not (looking-at parstart))
13081 (and use-hard-newlines
13082 (not (get-text-property (1- start) 'hard)))))
13085 (if (< (point) (point-max))
13086 (goto-char start))))
13089 This part handles three situations: when point is between paragraphs,
13090 when there is a fill prefix and when there is no fill prefix.
13093 The @code{while} loop looks like this:
13097 ;; @r{going forwards and not at the end of the buffer}
13098 (while (and (> arg 0) (not (eobp)))
13100 ;; @r{between paragraphs}
13101 ;; Move forward over separator lines...
13102 (while (and (not (eobp))
13103 (progn (move-to-left-margin) (not (eobp)))
13104 (looking-at parsep))
13106 ;; @r{This decrements the loop}
13107 (unless (eobp) (setq arg (1- arg)))
13108 ;; ... and one more line.
13113 (if fill-prefix-regexp
13114 ;; There is a fill prefix; it overrides parstart;
13115 ;; we go forward line by line
13116 (while (and (not (eobp))
13117 (progn (move-to-left-margin) (not (eobp)))
13118 (not (looking-at parsep))
13119 (looking-at fill-prefix-regexp))
13124 ;; There is no fill prefix;
13125 ;; we go forward character by character
13126 (while (and (re-search-forward sp-parstart nil 1)
13127 (progn (setq start (match-beginning 0))
13130 (progn (move-to-left-margin)
13131 (not (looking-at parsep)))
13132 (or (not (looking-at parstart))
13133 (and use-hard-newlines
13134 (not (get-text-property (1- start) 'hard)))))
13139 ;; and if there is no fill prefix and if we are not at the end,
13140 ;; go to whatever was found in the regular expression search
13142 (if (< (point) (point-max))
13143 (goto-char start))))
13148 We can see that this is a decrementing counter @code{while} loop,
13149 using the expression @code{(setq arg (1- arg))} as the decrementer.
13150 That expression is not far from the @code{while}, but is hidden in
13151 another Lisp macro, an @code{unless} macro. Unless we are at the end
13152 of the buffer---that is what the @code{eobp} function determines; it
13153 is an abbreviation of @samp{End Of Buffer P}---we decrease the value
13154 of @code{arg} by one.
13156 (If we are at the end of the buffer, we cannot go forward any more and
13157 the next loop of the @code{while} expression will test false since the
13158 test is an @code{and} with @code{(not (eobp))}. The @code{not}
13159 function means exactly as you expect; it is another name for
13160 @code{null}, a function that returns true when its argument is false.)
13162 Interestingly, the loop count is not decremented until we leave the
13163 space between paragraphs, unless we come to the end of buffer or stop
13164 seeing the local value of the paragraph separator.
13166 That second @code{while} also has a @code{(move-to-left-margin)}
13167 expression. The function is self-explanatory. It is inside a
13168 @code{progn} expression and not the last element of its body, so it is
13169 only invoked for its side effect, which is to move point to the left
13170 margin of the current line.
13173 The @code{looking-at} function is also self-explanatory; it returns
13174 true if the text after point matches the regular expression given as
13177 The rest of the body of the loop looks difficult at first, but makes
13178 sense as you come to understand it.
13181 First consider what happens if there is a fill prefix:
13185 (if fill-prefix-regexp
13186 ;; There is a fill prefix; it overrides parstart;
13187 ;; we go forward line by line
13188 (while (and (not (eobp))
13189 (progn (move-to-left-margin) (not (eobp)))
13190 (not (looking-at parsep))
13191 (looking-at fill-prefix-regexp))
13197 This expression moves point forward line by line so long
13198 as four conditions are true:
13202 Point is not at the end of the buffer.
13205 We can move to the left margin of the text and are
13206 not at the end of the buffer.
13209 The text following point does not separate paragraphs.
13212 The pattern following point is the fill prefix regular expression.
13215 The last condition may be puzzling, until you remember that point was
13216 moved to the beginning of the line early in the @code{forward-paragraph}
13217 function. This means that if the text has a fill prefix, the
13218 @code{looking-at} function will see it.
13221 Consider what happens when there is no fill prefix.
13225 (while (and (re-search-forward sp-parstart nil 1)
13226 (progn (setq start (match-beginning 0))
13229 (progn (move-to-left-margin)
13230 (not (looking-at parsep)))
13231 (or (not (looking-at parstart))
13232 (and use-hard-newlines
13233 (not (get-text-property (1- start) 'hard)))))
13239 This @code{while} loop has us searching forward for
13240 @code{sp-parstart}, which is the combination of possible whitespace
13241 with the local value of the start of a paragraph or of a paragraph
13242 separator. (The latter two are within an expression starting
13243 @code{\(?:} so that they are not referenced by the
13244 @code{match-beginning} function.)
13247 The two expressions,
13251 (setq start (match-beginning 0))
13257 mean go to the start of the text matched by the regular expression
13260 The @code{(match-beginning 0)} expression is new. It returns a number
13261 specifying the location of the start of the text that was matched by
13264 The @code{match-beginning} function is used here because of a
13265 characteristic of a forward search: a successful forward search,
13266 regardless of whether it is a plain search or a regular expression
13267 search, moves point to the end of the text that is found. In this
13268 case, a successful search moves point to the end of the pattern for
13269 @code{sp-parstart}.
13271 However, we want to put point at the end of the current paragraph, not
13272 somewhere else. Indeed, since the search possibly includes the
13273 paragraph separator, point may end up at the beginning of the next one
13274 unless we use an expression that includes @code{match-beginning}.
13276 @findex match-beginning
13277 When given an argument of 0, @code{match-beginning} returns the
13278 position that is the start of the text matched by the most recent
13279 search. In this case, the most recent search looks for
13280 @code{sp-parstart}. The @code{(match-beginning 0)} expression returns
13281 the beginning position of that pattern, rather than the end position
13284 (Incidentally, when passed a positive number as an argument, the
13285 @code{match-beginning} function returns the location of point at that
13286 parenthesized expression in the last search unless that parenthesized
13287 expression begins with @code{\(?:}. I don't know why @code{\(?:}
13288 appears here since the argument is 0.)
13291 The last expression when there is no fill prefix is
13295 (if (< (point) (point-max))
13296 (goto-char start))))
13301 This says that if there is no fill prefix and if we are not at the
13302 end, point should move to the beginning of whatever was found by the
13303 regular expression search for @code{sp-parstart}.
13305 The full definition for the @code{forward-paragraph} function not only
13306 includes code for going forwards, but also code for going backwards.
13308 If you are reading this inside of GNU Emacs and you want to see the
13309 whole function, you can type @kbd{C-h f} (@code{describe-function})
13310 and the name of the function. This gives you the function
13311 documentation and the name of the library containing the function's
13312 source. Place point over the name of the library and press the RET
13313 key; you will be taken directly to the source. (Be sure to install
13314 your sources! Without them, you are like a person who tries to drive
13315 a car with his eyes shut!)
13318 @section Create Your Own @file{TAGS} File
13320 @cindex @file{TAGS} file, create own
13322 Besides @kbd{C-h f} (@code{describe-function}), another way to see the
13323 source of a function is to type @kbd{M-.} (@code{find-tag}) and the
13324 name of the function when prompted for it. This is a good habit to
13325 get into. The @kbd{M-.} (@code{find-tag}) command takes you directly
13326 to the source for a function, variable, or node. The function depends
13327 on tags tables to tell it where to go.
13329 If the @code{find-tag} function first asks you for the name of a
13330 @file{TAGS} table, give it the name of a @file{TAGS} file such as
13331 @file{/usr/local/src/emacs/src/TAGS}. (The exact path to your
13332 @file{TAGS} file depends on how your copy of Emacs was installed. I
13333 just told you the location that provides both my C and my Emacs Lisp
13336 You can also create your own @file{TAGS} file for directories that
13339 You often need to build and install tags tables yourself. They are
13340 not built automatically. A tags table is called a @file{TAGS} file;
13341 the name is in upper case letters.
13343 You can create a @file{TAGS} file by calling the @code{etags} program
13344 that comes as a part of the Emacs distribution. Usually, @code{etags}
13345 is compiled and installed when Emacs is built. (@code{etags} is not
13346 an Emacs Lisp function or a part of Emacs; it is a C program.)
13349 To create a @file{TAGS} file, first switch to the directory in which
13350 you want to create the file. In Emacs you can do this with the
13351 @kbd{M-x cd} command, or by visiting a file in the directory, or by
13352 listing the directory with @kbd{C-x d} (@code{dired}). Then run the
13353 compile command, with @w{@code{etags *.el}} as the command to execute
13356 M-x compile RET etags *.el RET
13360 to create a @file{TAGS} file for Emacs Lisp.
13362 For example, if you have a large number of files in your
13363 @file{~/emacs} directory, as I do---I have 137 @file{.el} files in it,
13364 of which I load 12---you can create a @file{TAGS} file for the Emacs
13365 Lisp files in that directory.
13368 The @code{etags} program takes all the usual shell ``wildcards''. For
13369 example, if you have two directories for which you want a single
13370 @file{TAGS} file, type @w{@code{etags *.el ../elisp/*.el}}, where
13371 @file{../elisp/} is the second directory:
13374 M-x compile RET etags *.el ../elisp/*.el RET
13381 M-x compile RET etags --help RET
13385 to see a list of the options accepted by @code{etags} as well as a
13386 list of supported languages.
13388 The @code{etags} program handles more than 20 languages, including
13389 Emacs Lisp, Common Lisp, Scheme, C, C++, Ada, Fortran, HTML, Java,
13390 LaTeX, Pascal, Perl, PostScript, Python, TeX, Texinfo, makefiles, and
13391 most assemblers. The program has no switches for specifying the
13392 language; it recognizes the language in an input file according to its
13393 file name and contents.
13395 @file{etags} is very helpful when you are writing code yourself and
13396 want to refer back to functions you have already written. Just run
13397 @code{etags} again at intervals as you write new functions, so they
13398 become part of the @file{TAGS} file.
13400 If you think an appropriate @file{TAGS} file already exists for what
13401 you want, but do not know where it is, you can use the @code{locate}
13402 program to attempt to find it.
13404 Type @w{@kbd{M-x locate @key{RET} TAGS @key{RET}}} and Emacs will list
13405 for you the full path names of all your @file{TAGS} files. On my
13406 system, this command lists 34 @file{TAGS} files. On the other hand, a
13407 ``plain vanilla'' system I recently installed did not contain any
13410 If the tags table you want has been created, you can use the @code{M-x
13411 visit-tags-table} command to specify it. Otherwise, you will need to
13412 create the tag table yourself and then use @code{M-x
13415 @subsubheading Building Tags in the Emacs sources
13416 @cindex Building Tags in the Emacs sources
13417 @cindex Tags in the Emacs sources
13420 The GNU Emacs sources come with a @file{Makefile} that contains a
13421 sophisticated @code{etags} command that creates, collects, and merges
13422 tags tables from all over the Emacs sources and puts the information
13423 into one @file{TAGS} file in the @file{src/} directory. (The
13424 @file{src/} directory is below the top level of your Emacs directory.)
13427 To build this @file{TAGS} file, go to the top level of your Emacs
13428 source directory and run the compile command @code{make tags}:
13431 M-x compile RET make tags RET
13435 (The @code{make tags} command works well with the GNU Emacs sources,
13436 as well as with some other source packages.)
13438 For more information, see @ref{Tags, , Tag Tables, emacs, The GNU Emacs
13441 @node Regexp Review
13444 Here is a brief summary of some recently introduced functions.
13448 Repeatedly evaluate the body of the expression so long as the first
13449 element of the body tests true. Then return @code{nil}. (The
13450 expression is evaluated only for its side effects.)
13459 (insert (format "foo is %d.\n" foo))
13460 (setq foo (1- foo))))
13462 @result{} foo is 2.
13469 (The @code{insert} function inserts its arguments at point; the
13470 @code{format} function returns a string formatted from its arguments
13471 the way @code{message} formats its arguments; @code{\n} produces a new
13474 @item re-search-forward
13475 Search for a pattern, and if the pattern is found, move point to rest
13479 Takes four arguments, like @code{search-forward}:
13483 A regular expression that specifies the pattern to search for.
13484 (Remember to put quotation marks around this argument!)
13487 Optionally, the limit of the search.
13490 Optionally, what to do if the search fails, return @code{nil} or an
13494 Optionally, how many times to repeat the search; if negative, the
13495 search goes backwards.
13499 Bind some variables locally to particular values,
13500 and then evaluate the remaining arguments, returning the value of the
13501 last one. While binding the local variables, use the local values of
13502 variables bound earlier, if any.
13511 (message "'bar' is %d." bar))
13512 @result{} 'bar' is 21.
13516 @item match-beginning
13517 Return the position of the start of the text found by the last regular
13521 Return @code{t} for true if the text after point matches the argument,
13522 which should be a regular expression.
13525 Return @code{t} for true if point is at the end of the accessible part
13526 of a buffer. The end of the accessible part is the end of the buffer
13527 if the buffer is not narrowed; it is the end of the narrowed part if
13528 the buffer is narrowed.
13532 @node re-search Exercises
13533 @section Exercises with @code{re-search-forward}
13537 Write a function to search for a regular expression that matches two
13538 or more blank lines in sequence.
13541 Write a function to search for duplicated words, such as ``the the''.
13542 @xref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs
13543 Manual}, for information on how to write a regexp (a regular
13544 expression) to match a string that is composed of two identical
13545 halves. You can devise several regexps; some are better than others.
13546 The function I use is described in an appendix, along with several
13547 regexps. @xref{the-the, , @code{the-the} Duplicated Words Function}.
13550 @node Counting Words
13551 @chapter Counting via Repetition and Regexps
13552 @cindex Repetition for word counting
13553 @cindex Regular expressions for word counting
13555 Repetition and regular expression searches are powerful tools that you
13556 often use when you write code in Emacs Lisp. This chapter illustrates
13557 the use of regular expression searches through the construction of
13558 word count commands using @code{while} loops and recursion.
13561 * Why Count Words::
13562 * @value{COUNT-WORDS}:: Use a regexp, but find a problem.
13563 * recursive-count-words:: Start with case of no words in region.
13564 * Counting Exercise::
13568 @node Why Count Words
13569 @unnumberedsec Counting words
13572 The standard Emacs distribution contains functions for counting the
13573 number of lines and words within a region.
13575 Certain types of writing ask you to count words. Thus, if you write
13576 an essay, you may be limited to 800 words; if you write a novel, you
13577 may discipline yourself to write 1000 words a day. It seems odd, but
13578 for a long time, Emacs lacked a word count command. Perhaps people used
13579 Emacs mostly for code or types of documentation that did not require
13580 word counts; or perhaps they restricted themselves to the operating
13581 system word count command, @code{wc}. Alternatively, people may have
13582 followed the publishers' convention and computed a word count by
13583 dividing the number of characters in a document by five.
13585 There are many ways to implement a command to count words. Here are
13586 some examples, which you may wish to compare with the standard Emacs
13587 command, @code{count-words-region}.
13589 @node @value{COUNT-WORDS}
13590 @section The @code{@value{COUNT-WORDS}} Function
13591 @findex @value{COUNT-WORDS}
13593 A word count command could count words in a line, paragraph, region,
13594 or buffer. What should the command cover? You could design the
13595 command to count the number of words in a complete buffer. However,
13596 the Emacs tradition encourages flexibility---you may want to count
13597 words in just a section, rather than all of a buffer. So it makes
13598 more sense to design the command to count the number of words in a
13599 region. Once you have a command to count words in a region, you can,
13600 if you wish, count words in a whole buffer by marking it with
13601 @w{@kbd{C-x h}} (@code{mark-whole-buffer}).
13603 Clearly, counting words is a repetitive act: starting from the
13604 beginning of the region, you count the first word, then the second
13605 word, then the third word, and so on, until you reach the end of the
13606 region. This means that word counting is ideally suited to recursion
13607 or to a @code{while} loop.
13610 * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop.
13611 * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}.
13615 @node Design @value{COUNT-WORDS}
13616 @unnumberedsubsec Designing @code{@value{COUNT-WORDS}}
13619 First, we will implement the word count command with a @code{while}
13620 loop, then with recursion. The command will, of course, be
13624 The template for an interactive function definition is, as always:
13628 (defun @var{name-of-function} (@var{argument-list})
13629 "@var{documentation}@dots{}"
13630 (@var{interactive-expression}@dots{})
13635 What we need to do is fill in the slots.
13637 The name of the function should be self-explanatory and similar to the
13638 existing @code{count-lines-region} name. This makes the name easier
13639 to remember. @code{count-words-region} is the obvious choice. Since
13640 that name is now used for the standard Emacs command to count words, we
13641 will name our implementation @code{@value{COUNT-WORDS}}.
13643 The function counts words within a region. This means that the
13644 argument list must contain symbols that are bound to the two
13645 positions, the beginning and end of the region. These two positions
13646 can be called @samp{beginning} and @samp{end} respectively. The first
13647 line of the documentation should be a single sentence, since that is
13648 all that is printed as documentation by a command such as
13649 @code{apropos}. The interactive expression will be of the form
13650 @samp{(interactive "r")}, since that will cause Emacs to pass the
13651 beginning and end of the region to the function's argument list. All
13654 The body of the function needs to be written to do three tasks:
13655 first, to set up conditions under which the @code{while} loop can
13656 count words, second, to run the @code{while} loop, and third, to send
13657 a message to the user.
13659 When a user calls @code{@value{COUNT-WORDS}}, point may be at the
13660 beginning or the end of the region. However, the counting process
13661 must start at the beginning of the region. This means we will want
13662 to put point there if it is not already there. Executing
13663 @code{(goto-char beginning)} ensures this. Of course, we will want to
13664 return point to its expected position when the function finishes its
13665 work. For this reason, the body must be enclosed in a
13666 @code{save-excursion} expression.
13668 The central part of the body of the function consists of a
13669 @code{while} loop in which one expression jumps point forward word by
13670 word, and another expression counts those jumps. The true-or-false-test
13671 of the @code{while} loop should test true so long as point should jump
13672 forward, and false when point is at the end of the region.
13674 We could use @code{(forward-word 1)} as the expression for moving point
13675 forward word by word, but it is easier to see what Emacs identifies as a
13676 ``word'' if we use a regular expression search.
13678 A regular expression search that finds the pattern for which it is
13679 searching leaves point after the last character matched. This means
13680 that a succession of successful word searches will move point forward
13683 As a practical matter, we want the regular expression search to jump
13684 over whitespace and punctuation between words as well as over the
13685 words themselves. A regexp that refuses to jump over interword
13686 whitespace would never jump more than one word! This means that
13687 the regexp should include the whitespace and punctuation that follows
13688 a word, if any, as well as the word itself. (A word may end a buffer
13689 and not have any following whitespace or punctuation, so that part of
13690 the regexp must be optional.)
13692 Thus, what we want for the regexp is a pattern defining one or more
13693 word constituent characters followed, optionally, by one or more
13694 characters that are not word constituents. The regular expression for
13702 The buffer's syntax table determines which characters are and are not
13703 word constituents. For more information about syntax,
13704 @pxref{Syntax Tables, , Syntax Tables, elisp, The GNU Emacs Lisp
13708 The search expression looks like this:
13711 (re-search-forward "\\w+\\W*")
13715 (Note that paired backslashes precede the @samp{w} and @samp{W}. A
13716 single backslash has special meaning to the Emacs Lisp interpreter.
13717 It indicates that the following character is interpreted differently
13718 than usual. For example, the two characters, @samp{\n}, stand for
13719 @samp{newline}, rather than for a backslash followed by @samp{n}. Two
13720 backslashes in a row stand for an ordinary, ``unspecial'' backslash, so
13721 Emacs Lisp interpreter ends of seeing a single backslash followed by a
13722 letter. So it discovers the letter is special.)
13724 We need a counter to count how many words there are; this variable
13725 must first be set to 0 and then incremented each time Emacs goes
13726 around the @code{while} loop. The incrementing expression is simply:
13729 (setq count (1+ count))
13732 Finally, we want to tell the user how many words there are in the
13733 region. The @code{message} function is intended for presenting this
13734 kind of information to the user. The message has to be phrased so
13735 that it reads properly regardless of how many words there are in the
13736 region: we don't want to say that ``there are 1 words in the region''.
13737 The conflict between singular and plural is ungrammatical. We can
13738 solve this problem by using a conditional expression that evaluates
13739 different messages depending on the number of words in the region.
13740 There are three possibilities: no words in the region, one word in the
13741 region, and more than one word. This means that the @code{cond}
13742 special form is appropriate.
13745 All this leads to the following function definition:
13749 ;;; @r{First version; has bugs!}
13750 (defun @value{COUNT-WORDS} (beginning end)
13751 "Print number of words in the region.
13752 Words are defined as at least one word-constituent
13753 character followed by at least one character that
13754 is not a word-constituent. The buffer's syntax
13755 table determines which characters these are."
13757 (message "Counting words in region ... ")
13761 ;;; @r{1. Set up appropriate conditions.}
13763 (goto-char beginning)
13768 ;;; @r{2. Run the} while @r{loop.}
13769 (while (< (point) end)
13770 (re-search-forward "\\w+\\W*")
13771 (setq count (1+ count)))
13775 ;;; @r{3. Send a message to the user.}
13776 (cond ((zerop count)
13778 "The region does NOT have any words."))
13781 "The region has 1 word."))
13784 "The region has %d words." count))))))
13789 As written, the function works, but not in all circumstances.
13791 @node Whitespace Bug
13792 @subsection The Whitespace Bug in @code{@value{COUNT-WORDS}}
13794 The @code{@value{COUNT-WORDS}} command described in the preceding
13795 section has two bugs, or rather, one bug with two manifestations.
13796 First, if you mark a region containing only whitespace in the middle
13797 of some text, the @code{@value{COUNT-WORDS}} command tells you that the
13798 region contains one word! Second, if you mark a region containing
13799 only whitespace at the end of the buffer or the accessible portion of
13800 a narrowed buffer, the command displays an error message that looks
13804 Search failed: "\\w+\\W*"
13807 If you are reading this in Info in GNU Emacs, you can test for these
13810 First, evaluate the function in the usual manner to install it.
13812 Here is a copy of the definition. Place your cursor after the closing
13813 parenthesis and type @kbd{C-x C-e} to install it.
13817 ;; @r{First version; has bugs!}
13818 (defun @value{COUNT-WORDS} (beginning end)
13819 "Print number of words in the region.
13820 Words are defined as at least one word-constituent character followed
13821 by at least one character that is not a word-constituent. The buffer's
13822 syntax table determines which characters these are."
13826 (message "Counting words in region ... ")
13830 ;;; @r{1. Set up appropriate conditions.}
13832 (goto-char beginning)
13837 ;;; @r{2. Run the} while @r{loop.}
13838 (while (< (point) end)
13839 (re-search-forward "\\w+\\W*")
13840 (setq count (1+ count)))
13844 ;;; @r{3. Send a message to the user.}
13845 (cond ((zerop count)
13846 (message "The region does NOT have any words."))
13847 ((= 1 count) (message "The region has 1 word."))
13848 (t (message "The region has %d words." count))))))
13854 If you wish, you can also install this keybinding by evaluating it:
13857 (global-set-key "\C-c=" '@value{COUNT-WORDS})
13860 To conduct the first test, set mark and point to the beginning and end
13861 of the following line and then type @kbd{C-c =} (or @kbd{M-x
13862 @value{COUNT-WORDS}} if you have not bound @kbd{C-c =}):
13869 Emacs will tell you, correctly, that the region has three words.
13871 Repeat the test, but place mark at the beginning of the line and place
13872 point just @emph{before} the word @samp{one}. Again type the command
13873 @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}). Emacs should tell you
13874 that the region has no words, since it is composed only of the
13875 whitespace at the beginning of the line. But instead Emacs tells you
13876 that the region has one word!
13878 For the third test, copy the sample line to the end of the
13879 @file{*scratch*} buffer and then type several spaces at the end of the
13880 line. Place mark right after the word @samp{three} and point at the
13881 end of line. (The end of the line will be the end of the buffer.)
13882 Type @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}) as you did before.
13883 Again, Emacs should tell you that the region has no words, since it is
13884 composed only of the whitespace at the end of the line. Instead,
13885 Emacs displays an error message saying @samp{Search failed}.
13887 The two bugs stem from the same problem.
13889 Consider the first manifestation of the bug, in which the command
13890 tells you that the whitespace at the beginning of the line contains
13891 one word. What happens is this: The @code{M-x @value{COUNT-WORDS}}
13892 command moves point to the beginning of the region. The @code{while}
13893 tests whether the value of point is smaller than the value of
13894 @code{end}, which it is. Consequently, the regular expression search
13895 looks for and finds the first word. It leaves point after the word.
13896 @code{count} is set to one. The @code{while} loop repeats; but this
13897 time the value of point is larger than the value of @code{end}, the
13898 loop is exited; and the function displays a message saying the number
13899 of words in the region is one. In brief, the regular expression
13900 search looks for and finds the word even though it is outside
13903 In the second manifestation of the bug, the region is whitespace at
13904 the end of the buffer. Emacs says @samp{Search failed}. What happens
13905 is that the true-or-false-test in the @code{while} loop tests true, so
13906 the search expression is executed. But since there are no more words
13907 in the buffer, the search fails.
13909 In both manifestations of the bug, the search extends or attempts to
13910 extend outside of the region.
13912 The solution is to limit the search to the region---this is a fairly
13913 simple action, but as you may have come to expect, it is not quite as
13914 simple as you might think.
13916 As we have seen, the @code{re-search-forward} function takes a search
13917 pattern as its first argument. But in addition to this first,
13918 mandatory argument, it accepts three optional arguments. The optional
13919 second argument bounds the search. The optional third argument, if
13920 @code{t}, causes the function to return @code{nil} rather than signal
13921 an error if the search fails. The optional fourth argument is a
13922 repeat count. (In Emacs, you can see a function's documentation by
13923 typing @kbd{C-h f}, the name of the function, and then @key{RET}.)
13925 In the @code{@value{COUNT-WORDS}} definition, the value of the end of
13926 the region is held by the variable @code{end} which is passed as an
13927 argument to the function. Thus, we can add @code{end} as an argument
13928 to the regular expression search expression:
13931 (re-search-forward "\\w+\\W*" end)
13934 However, if you make only this change to the @code{@value{COUNT-WORDS}}
13935 definition and then test the new version of the definition on a
13936 stretch of whitespace, you will receive an error message saying
13937 @samp{Search failed}.
13939 What happens is this: the search is limited to the region, and fails
13940 as you expect because there are no word-constituent characters in the
13941 region. Since it fails, we receive an error message. But we do not
13942 want to receive an error message in this case; we want to receive the
13943 message that "The region does NOT have any words."
13945 The solution to this problem is to provide @code{re-search-forward}
13946 with a third argument of @code{t}, which causes the function to return
13947 @code{nil} rather than signal an error if the search fails.
13949 However, if you make this change and try it, you will see the message
13950 ``Counting words in region ... '' and @dots{} you will keep on seeing
13951 that message @dots{}, until you type @kbd{C-g} (@code{keyboard-quit}).
13953 Here is what happens: the search is limited to the region, as before,
13954 and it fails because there are no word-constituent characters in the
13955 region, as expected. Consequently, the @code{re-search-forward}
13956 expression returns @code{nil}. It does nothing else. In particular,
13957 it does not move point, which it does as a side effect if it finds the
13958 search target. After the @code{re-search-forward} expression returns
13959 @code{nil}, the next expression in the @code{while} loop is evaluated.
13960 This expression increments the count. Then the loop repeats. The
13961 true-or-false-test tests true because the value of point is still less
13962 than the value of end, since the @code{re-search-forward} expression
13963 did not move point. @dots{} and the cycle repeats @dots{}
13965 The @code{@value{COUNT-WORDS}} definition requires yet another
13966 modification, to cause the true-or-false-test of the @code{while} loop
13967 to test false if the search fails. Put another way, there are two
13968 conditions that must be satisfied in the true-or-false-test before the
13969 word count variable is incremented: point must still be within the
13970 region and the search expression must have found a word to count.
13972 Since both the first condition and the second condition must be true
13973 together, the two expressions, the region test and the search
13974 expression, can be joined with an @code{and} special form and embedded in
13975 the @code{while} loop as the true-or-false-test, like this:
13978 (and (< (point) end) (re-search-forward "\\w+\\W*" end t))
13981 @c colon in printed section title causes problem in Info cross reference
13982 @c also trouble with an overfull hbox
13985 (For information about @code{and}, see
13986 @ref{kill-new function, , The @code{kill-new} function}.)
13990 (@xref{kill-new function, , The @code{kill-new} function}, for
13991 information about @code{and}.)
13994 The @code{re-search-forward} expression returns @code{t} if the search
13995 succeeds and as a side effect moves point. Consequently, as words are
13996 found, point is moved through the region. When the search expression
13997 fails to find another word, or when point reaches the end of the
13998 region, the true-or-false-test tests false, the @code{while} loop
13999 exits, and the @code{@value{COUNT-WORDS}} function displays one or
14000 other of its messages.
14002 After incorporating these final changes, the @code{@value{COUNT-WORDS}}
14003 works without bugs (or at least, without bugs that I have found!).
14004 Here is what it looks like:
14008 ;;; @r{Final version:} @code{while}
14009 (defun @value{COUNT-WORDS} (beginning end)
14010 "Print number of words in the region."
14012 (message "Counting words in region ... ")
14016 ;;; @r{1. Set up appropriate conditions.}
14019 (goto-char beginning)
14023 ;;; @r{2. Run the} while @r{loop.}
14024 (while (and (< (point) end)
14025 (re-search-forward "\\w+\\W*" end t))
14026 (setq count (1+ count)))
14030 ;;; @r{3. Send a message to the user.}
14031 (cond ((zerop count)
14033 "The region does NOT have any words."))
14036 "The region has 1 word."))
14039 "The region has %d words." count))))))
14043 @node recursive-count-words
14044 @section Count Words Recursively
14045 @cindex Count words recursively
14046 @cindex Recursively counting words
14047 @cindex Words, counted recursively
14049 You can write the function for counting words recursively as well as
14050 with a @code{while} loop. Let's see how this is done.
14052 First, we need to recognize that the @code{@value{COUNT-WORDS}}
14053 function has three jobs: it sets up the appropriate conditions for
14054 counting to occur; it counts the words in the region; and it sends a
14055 message to the user telling how many words there are.
14057 If we write a single recursive function to do everything, we will
14058 receive a message for every recursive call. If the region contains 13
14059 words, we will receive thirteen messages, one right after the other.
14060 We don't want this! Instead, we must write two functions to do the
14061 job, one of which (the recursive function) will be used inside of the
14062 other. One function will set up the conditions and display the
14063 message; the other will return the word count.
14065 Let us start with the function that causes the message to be displayed.
14066 We can continue to call this @code{@value{COUNT-WORDS}}.
14068 This is the function that the user will call. It will be interactive.
14069 Indeed, it will be similar to our previous versions of this
14070 function, except that it will call @code{recursive-count-words} to
14071 determine how many words are in the region.
14074 We can readily construct a template for this function, based on our
14079 ;; @r{Recursive version; uses regular expression search}
14080 (defun @value{COUNT-WORDS} (beginning end)
14081 "@var{documentation}@dots{}"
14082 (@var{interactive-expression}@dots{})
14086 ;;; @r{1. Set up appropriate conditions.}
14087 (@var{explanatory message})
14088 (@var{set-up functions}@dots{}
14092 ;;; @r{2. Count the words.}
14093 @var{recursive call}
14097 ;;; @r{3. Send a message to the user.}
14098 @var{message providing word count}))
14102 The definition looks straightforward, except that somehow the count
14103 returned by the recursive call must be passed to the message
14104 displaying the word count. A little thought suggests that this can be
14105 done by making use of a @code{let} expression: we can bind a variable
14106 in the varlist of a @code{let} expression to the number of words in
14107 the region, as returned by the recursive call; and then the
14108 @code{cond} expression, using binding, can display the value to the
14111 Often, one thinks of the binding within a @code{let} expression as
14112 somehow secondary to the ``primary'' work of a function. But in this
14113 case, what you might consider the ``primary'' job of the function,
14114 counting words, is done within the @code{let} expression.
14117 Using @code{let}, the function definition looks like this:
14121 (defun @value{COUNT-WORDS} (beginning end)
14122 "Print number of words in the region."
14127 ;;; @r{1. Set up appropriate conditions.}
14128 (message "Counting words in region ... ")
14130 (goto-char beginning)
14134 ;;; @r{2. Count the words.}
14135 (let ((count (recursive-count-words end)))
14139 ;;; @r{3. Send a message to the user.}
14140 (cond ((zerop count)
14142 "The region does NOT have any words."))
14145 "The region has 1 word."))
14148 "The region has %d words." count))))))
14152 Next, we need to write the recursive counting function.
14154 A recursive function has at least three parts: the ``do-again-test'', the
14155 ``next-step-expression'', and the recursive call.
14157 The do-again-test determines whether the function will or will not be
14158 called again. Since we are counting words in a region and can use a
14159 function that moves point forward for every word, the do-again-test
14160 can check whether point is still within the region. The do-again-test
14161 should find the value of point and determine whether point is before,
14162 at, or after the value of the end of the region. We can use the
14163 @code{point} function to locate point. Clearly, we must pass the
14164 value of the end of the region to the recursive counting function as an
14167 In addition, the do-again-test should also test whether the search finds a
14168 word. If it does not, the function should not call itself again.
14170 The next-step-expression changes a value so that when the recursive
14171 function is supposed to stop calling itself, it stops. More
14172 precisely, the next-step-expression changes a value so that at the
14173 right time, the do-again-test stops the recursive function from
14174 calling itself again. In this case, the next-step-expression can be
14175 the expression that moves point forward, word by word.
14177 The third part of a recursive function is the recursive call.
14179 Somewhere, also, we also need a part that does the ``work'' of the
14180 function, a part that does the counting. A vital part!
14183 But already, we have an outline of the recursive counting function:
14187 (defun recursive-count-words (region-end)
14188 "@var{documentation}@dots{}"
14189 @var{do-again-test}
14190 @var{next-step-expression}
14191 @var{recursive call})
14195 Now we need to fill in the slots. Let's start with the simplest cases
14196 first: if point is at or beyond the end of the region, there cannot
14197 be any words in the region, so the function should return zero.
14198 Likewise, if the search fails, there are no words to count, so the
14199 function should return zero.
14201 On the other hand, if point is within the region and the search
14202 succeeds, the function should call itself again.
14205 Thus, the do-again-test should look like this:
14209 (and (< (point) region-end)
14210 (re-search-forward "\\w+\\W*" region-end t))
14214 Note that the search expression is part of the do-again-test---the
14215 function returns @code{t} if its search succeeds and @code{nil} if it
14216 fails. (@xref{Whitespace Bug, , The Whitespace Bug in
14217 @code{@value{COUNT-WORDS}}}, for an explanation of how
14218 @code{re-search-forward} works.)
14220 The do-again-test is the true-or-false test of an @code{if} clause.
14221 Clearly, if the do-again-test succeeds, the then-part of the @code{if}
14222 clause should call the function again; but if it fails, the else-part
14223 should return zero since either point is outside the region or the
14224 search failed because there were no words to find.
14226 But before considering the recursive call, we need to consider the
14227 next-step-expression. What is it? Interestingly, it is the search
14228 part of the do-again-test.
14230 In addition to returning @code{t} or @code{nil} for the
14231 do-again-test, @code{re-search-forward} moves point forward as a side
14232 effect of a successful search. This is the action that changes the
14233 value of point so that the recursive function stops calling itself
14234 when point completes its movement through the region. Consequently,
14235 the @code{re-search-forward} expression is the next-step-expression.
14238 In outline, then, the body of the @code{recursive-count-words}
14239 function looks like this:
14243 (if @var{do-again-test-and-next-step-combined}
14245 @var{recursive-call-returning-count}
14251 How to incorporate the mechanism that counts?
14253 If you are not used to writing recursive functions, a question like
14254 this can be troublesome. But it can and should be approached
14257 We know that the counting mechanism should be associated in some way
14258 with the recursive call. Indeed, since the next-step-expression moves
14259 point forward by one word, and since a recursive call is made for
14260 each word, the counting mechanism must be an expression that adds one
14261 to the value returned by a call to @code{recursive-count-words}.
14264 Consider several cases:
14268 If there are two words in the region, the function should return
14269 a value resulting from adding one to the value returned when it counts
14270 the first word, plus the number returned when it counts the remaining
14271 words in the region, which in this case is one.
14274 If there is one word in the region, the function should return
14275 a value resulting from adding one to the value returned when it counts
14276 that word, plus the number returned when it counts the remaining
14277 words in the region, which in this case is zero.
14280 If there are no words in the region, the function should return zero.
14283 From the sketch we can see that the else-part of the @code{if} returns
14284 zero for the case of no words. This means that the then-part of the
14285 @code{if} must return a value resulting from adding one to the value
14286 returned from a count of the remaining words.
14289 The expression will look like this, where @code{1+} is a function that
14290 adds one to its argument.
14293 (1+ (recursive-count-words region-end))
14297 The whole @code{recursive-count-words} function will then look like
14302 (defun recursive-count-words (region-end)
14303 "@var{documentation}@dots{}"
14305 ;;; @r{1. do-again-test}
14306 (if (and (< (point) region-end)
14307 (re-search-forward "\\w+\\W*" region-end t))
14311 ;;; @r{2. then-part: the recursive call}
14312 (1+ (recursive-count-words region-end))
14314 ;;; @r{3. else-part}
14320 Let's examine how this works:
14322 If there are no words in the region, the else part of the @code{if}
14323 expression is evaluated and consequently the function returns zero.
14325 If there is one word in the region, the value of point is less than
14326 the value of @code{region-end} and the search succeeds. In this case,
14327 the true-or-false-test of the @code{if} expression tests true, and the
14328 then-part of the @code{if} expression is evaluated. The counting
14329 expression is evaluated. This expression returns a value (which will
14330 be the value returned by the whole function) that is the sum of one
14331 added to the value returned by a recursive call.
14333 Meanwhile, the next-step-expression has caused point to jump over the
14334 first (and in this case only) word in the region. This means that
14335 when @code{(recursive-count-words region-end)} is evaluated a second
14336 time, as a result of the recursive call, the value of point will be
14337 equal to or greater than the value of region end. So this time,
14338 @code{recursive-count-words} will return zero. The zero will be added
14339 to one, and the original evaluation of @code{recursive-count-words}
14340 will return one plus zero, which is one, which is the correct amount.
14342 Clearly, if there are two words in the region, the first call to
14343 @code{recursive-count-words} returns one added to the value returned
14344 by calling @code{recursive-count-words} on a region containing the
14345 remaining word---that is, it adds one to one, producing two, which is
14346 the correct amount.
14348 Similarly, if there are three words in the region, the first call to
14349 @code{recursive-count-words} returns one added to the value returned
14350 by calling @code{recursive-count-words} on a region containing the
14351 remaining two words---and so on and so on.
14355 With full documentation the two functions look like this:
14359 The recursive function:
14361 @findex recursive-count-words
14364 (defun recursive-count-words (region-end)
14365 "Number of words between point and REGION-END."
14369 ;;; @r{1. do-again-test}
14370 (if (and (< (point) region-end)
14371 (re-search-forward "\\w+\\W*" region-end t))
14375 ;;; @r{2. then-part: the recursive call}
14376 (1+ (recursive-count-words region-end))
14378 ;;; @r{3. else-part}
14389 ;;; @r{Recursive version}
14390 (defun @value{COUNT-WORDS} (beginning end)
14391 "Print number of words in the region.
14395 Words are defined as at least one word-constituent
14396 character followed by at least one character that is
14397 not a word-constituent. The buffer's syntax table
14398 determines which characters these are."
14402 (message "Counting words in region ... ")
14404 (goto-char beginning)
14405 (let ((count (recursive-count-words end)))
14408 (cond ((zerop count)
14410 "The region does NOT have any words."))
14414 (message "The region has 1 word."))
14417 "The region has %d words." count))))))
14421 @node Counting Exercise
14422 @section Exercise: Counting Punctuation
14424 Using a @code{while} loop, write a function to count the number of
14425 punctuation marks in a region---period, comma, semicolon, colon,
14426 exclamation mark, and question mark. Do the same using recursion.
14428 @node Words in a defun
14429 @chapter Counting Words in a @code{defun}
14430 @cindex Counting words in a @code{defun}
14431 @cindex Word counting in a @code{defun}
14433 Our next project is to count the number of words in a function
14434 definition. Clearly, this can be done using some variant of
14435 @code{@value{COUNT-WORDS}}. @xref{Counting Words, , Counting via
14436 Repetition and Regexps}. If we are just going to count the words in
14437 one definition, it is easy enough to mark the definition with the
14438 @kbd{C-M-h} (@code{mark-defun}) command, and then call
14439 @code{@value{COUNT-WORDS}}.
14441 However, I am more ambitious: I want to count the words and symbols in
14442 every definition in the Emacs sources and then print a graph that
14443 shows how many functions there are of each length: how many contain 40
14444 to 49 words or symbols, how many contain 50 to 59 words or symbols,
14445 and so on. I have often been curious how long a typical function is,
14446 and this will tell.
14449 * Divide and Conquer::
14450 * Words and Symbols:: What to count?
14451 * Syntax:: What constitutes a word or symbol?
14452 * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}.
14453 * Several defuns:: Counting several defuns in a file.
14454 * Find a File:: Do you want to look at a file?
14455 * lengths-list-file:: A list of the lengths of many definitions.
14456 * Several files:: Counting in definitions in different files.
14457 * Several files recursively:: Recursively counting in different files.
14458 * Prepare the data:: Prepare the data for display in a graph.
14462 @node Divide and Conquer
14463 @unnumberedsec Divide and Conquer
14466 Described in one phrase, the histogram project is daunting; but
14467 divided into numerous small steps, each of which we can take one at a
14468 time, the project becomes less fearsome. Let us consider what the
14473 First, write a function to count the words in one definition. This
14474 includes the problem of handling symbols as well as words.
14477 Second, write a function to list the numbers of words in each function
14478 in a file. This function can use the @code{count-words-in-defun}
14482 Third, write a function to list the numbers of words in each function
14483 in each of several files. This entails automatically finding the
14484 various files, switching to them, and counting the words in the
14485 definitions within them.
14488 Fourth, write a function to convert the list of numbers that we
14489 created in step three to a form that will be suitable for printing as
14493 Fifth, write a function to print the results as a graph.
14496 This is quite a project! But if we take each step slowly, it will not
14499 @node Words and Symbols
14500 @section What to Count?
14501 @cindex Words and symbols in defun
14503 When we first start thinking about how to count the words in a
14504 function definition, the first question is (or ought to be) what are
14505 we going to count? When we speak of ``words'' with respect to a Lisp
14506 function definition, we are actually speaking, in large part, of
14507 ``symbols''. For example, the following @code{multiply-by-seven}
14508 function contains the five symbols @code{defun},
14509 @code{multiply-by-seven}, @code{number}, @code{*}, and @code{7}. In
14510 addition, in the documentation string, it contains the four words
14511 @samp{Multiply}, @samp{NUMBER}, @samp{by}, and @samp{seven}. The
14512 symbol @samp{number} is repeated, so the definition contains a total
14513 of ten words and symbols.
14517 (defun multiply-by-seven (number)
14518 "Multiply NUMBER by seven."
14524 However, if we mark the @code{multiply-by-seven} definition with
14525 @kbd{C-M-h} (@code{mark-defun}), and then call
14526 @code{@value{COUNT-WORDS}} on it, we will find that
14527 @code{@value{COUNT-WORDS}} claims the definition has eleven words, not
14528 ten! Something is wrong!
14530 The problem is twofold: @code{@value{COUNT-WORDS}} does not count the
14531 @samp{*} as a word, and it counts the single symbol,
14532 @code{multiply-by-seven}, as containing three words. The hyphens are
14533 treated as if they were interword spaces rather than intraword
14534 connectors: @samp{multiply-by-seven} is counted as if it were written
14535 @samp{multiply by seven}.
14537 The cause of this confusion is the regular expression search within
14538 the @code{@value{COUNT-WORDS}} definition that moves point forward word
14539 by word. In the canonical version of @code{@value{COUNT-WORDS}}, the
14547 This regular expression is a pattern defining one or more word
14548 constituent characters possibly followed by one or more characters
14549 that are not word constituents. What is meant by ``word constituent
14550 characters'' brings us to the issue of syntax, which is worth a section
14554 @section What Constitutes a Word or Symbol?
14555 @cindex Syntax categories and tables
14557 Emacs treats different characters as belonging to different
14558 @dfn{syntax categories}. For example, the regular expression,
14559 @samp{\\w+}, is a pattern specifying one or more @emph{word
14560 constituent} characters. Word constituent characters are members of
14561 one syntax category. Other syntax categories include the class of
14562 punctuation characters, such as the period and the comma, and the
14563 class of whitespace characters, such as the blank space and the tab
14564 character. (For more information, @pxref{Syntax Tables, , Syntax
14565 Tables, elisp, The GNU Emacs Lisp Reference Manual}.)
14567 Syntax tables specify which characters belong to which categories.
14568 Usually, a hyphen is not specified as a ``word constituent character''.
14569 Instead, it is specified as being in the ``class of characters that are
14570 part of symbol names but not words.'' This means that the
14571 @code{@value{COUNT-WORDS}} function treats it in the same way it treats
14572 an interword white space, which is why @code{@value{COUNT-WORDS}}
14573 counts @samp{multiply-by-seven} as three words.
14575 There are two ways to cause Emacs to count @samp{multiply-by-seven} as
14576 one symbol: modify the syntax table or modify the regular expression.
14578 We could redefine a hyphen as a word constituent character by
14579 modifying the syntax table that Emacs keeps for each mode. This
14580 action would serve our purpose, except that a hyphen is merely the
14581 most common character within symbols that is not typically a word
14582 constituent character; there are others, too.
14584 Alternatively, we can redefine the regexp used in the
14585 @code{@value{COUNT-WORDS}} definition so as to include symbols. This
14586 procedure has the merit of clarity, but the task is a little tricky.
14589 The first part is simple enough: the pattern must match ``at least one
14590 character that is a word or symbol constituent''. Thus:
14593 "\\(\\w\\|\\s_\\)+"
14597 The @samp{\\(} is the first part of the grouping construct that
14598 includes the @samp{\\w} and the @samp{\\s_} as alternatives, separated
14599 by the @samp{\\|}. The @samp{\\w} matches any word-constituent
14600 character and the @samp{\\s_} matches any character that is part of a
14601 symbol name but not a word-constituent character. The @samp{+}
14602 following the group indicates that the word or symbol constituent
14603 characters must be matched at least once.
14605 However, the second part of the regexp is more difficult to design.
14606 What we want is to follow the first part with ``optionally one or more
14607 characters that are not constituents of a word or symbol''. At first,
14608 I thought I could define this with the following:
14611 "\\(\\W\\|\\S_\\)*"
14615 The upper case @samp{W} and @samp{S} match characters that are
14616 @emph{not} word or symbol constituents. Unfortunately, this
14617 expression matches any character that is either not a word constituent
14618 or not a symbol constituent. This matches any character!
14620 I then noticed that every word or symbol in my test region was
14621 followed by white space (blank space, tab, or newline). So I tried
14622 placing a pattern to match one or more blank spaces after the pattern
14623 for one or more word or symbol constituents. This failed, too. Words
14624 and symbols are often separated by whitespace, but in actual code
14625 parentheses may follow symbols and punctuation may follow words. So
14626 finally, I designed a pattern in which the word or symbol constituents
14627 are followed optionally by characters that are not white space and
14628 then followed optionally by white space.
14631 Here is the full regular expression:
14634 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
14637 @node count-words-in-defun
14638 @section The @code{count-words-in-defun} Function
14639 @cindex Counting words in a @code{defun}
14641 We have seen that there are several ways to write a
14642 @code{count-words-region} function. To write a
14643 @code{count-words-in-defun}, we need merely adapt one of these
14646 The version that uses a @code{while} loop is easy to understand, so I
14647 am going to adapt that. Because @code{count-words-in-defun} will be
14648 part of a more complex program, it need not be interactive and it need
14649 not display a message but just return the count. These considerations
14650 simplify the definition a little.
14652 On the other hand, @code{count-words-in-defun} will be used within a
14653 buffer that contains function definitions. Consequently, it is
14654 reasonable to ask that the function determine whether it is called
14655 when point is within a function definition, and if it is, to return
14656 the count for that definition. This adds complexity to the
14657 definition, but saves us from needing to pass arguments to the
14661 These considerations lead us to prepare the following template:
14665 (defun count-words-in-defun ()
14666 "@var{documentation}@dots{}"
14667 (@var{set up}@dots{}
14668 (@var{while loop}@dots{})
14669 @var{return count})
14674 As usual, our job is to fill in the slots.
14678 We are presuming that this function will be called within a buffer
14679 containing function definitions. Point will either be within a
14680 function definition or not. For @code{count-words-in-defun} to work,
14681 point must move to the beginning of the definition, a counter must
14682 start at zero, and the counting loop must stop when point reaches the
14683 end of the definition.
14685 The @code{beginning-of-defun} function searches backwards for an
14686 opening delimiter such as a @samp{(} at the beginning of a line, and
14687 moves point to that position, or else to the limit of the search. In
14688 practice, this means that @code{beginning-of-defun} moves point to the
14689 beginning of an enclosing or preceding function definition, or else to
14690 the beginning of the buffer. We can use @code{beginning-of-defun} to
14691 place point where we wish to start.
14693 The @code{while} loop requires a counter to keep track of the words or
14694 symbols being counted. A @code{let} expression can be used to create
14695 a local variable for this purpose, and bind it to an initial value of zero.
14697 The @code{end-of-defun} function works like @code{beginning-of-defun}
14698 except that it moves point to the end of the definition.
14699 @code{end-of-defun} can be used as part of an expression that
14700 determines the position of the end of the definition.
14702 The set up for @code{count-words-in-defun} takes shape rapidly: first
14703 we move point to the beginning of the definition, then we create a
14704 local variable to hold the count, and finally, we record the position
14705 of the end of the definition so the @code{while} loop will know when to stop
14709 The code looks like this:
14713 (beginning-of-defun)
14715 (end (save-excursion (end-of-defun) (point))))
14720 The code is simple. The only slight complication is likely to concern
14721 @code{end}: it is bound to the position of the end of the definition
14722 by a @code{save-excursion} expression that returns the value of point
14723 after @code{end-of-defun} temporarily moves it to the end of the
14726 The second part of the @code{count-words-in-defun}, after the set up,
14727 is the @code{while} loop.
14729 The loop must contain an expression that jumps point forward word by
14730 word and symbol by symbol, and another expression that counts the
14731 jumps. The true-or-false-test for the @code{while} loop should test
14732 true so long as point should jump forward, and false when point is at
14733 the end of the definition. We have already redefined the regular
14734 expression for this, so the loop is straightforward:
14738 (while (and (< (point) end)
14740 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*" end t))
14741 (setq count (1+ count)))
14745 The third part of the function definition returns the count of words
14746 and symbols. This part is the last expression within the body of the
14747 @code{let} expression, and can be, very simply, the local variable
14748 @code{count}, which when evaluated returns the count.
14751 Put together, the @code{count-words-in-defun} definition looks like this:
14753 @findex count-words-in-defun
14756 (defun count-words-in-defun ()
14757 "Return the number of words and symbols in a defun."
14758 (beginning-of-defun)
14760 (end (save-excursion (end-of-defun) (point))))
14764 (and (< (point) end)
14766 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
14768 (setq count (1+ count)))
14773 How to test this? The function is not interactive, but it is easy to
14774 put a wrapper around the function to make it interactive; we can use
14775 almost the same code as for the recursive version of
14776 @code{@value{COUNT-WORDS}}:
14780 ;;; @r{Interactive version.}
14781 (defun count-words-defun ()
14782 "Number of words and symbols in a function definition."
14785 "Counting words and symbols in function definition ... ")
14788 (let ((count (count-words-in-defun)))
14792 "The definition does NOT have any words or symbols."))
14797 "The definition has 1 word or symbol."))
14800 "The definition has %d words or symbols." count)))))
14806 Let's re-use @kbd{C-c =} as a convenient keybinding:
14809 (global-set-key "\C-c=" 'count-words-defun)
14812 Now we can try out @code{count-words-defun}: install both
14813 @code{count-words-in-defun} and @code{count-words-defun}, and set the
14814 keybinding, and then place the cursor within the following definition:
14818 (defun multiply-by-seven (number)
14819 "Multiply NUMBER by seven."
14826 Success! The definition has 10 words and symbols.
14828 The next problem is to count the numbers of words and symbols in
14829 several definitions within a single file.
14831 @node Several defuns
14832 @section Count Several @code{defuns} Within a File
14834 A file such as @file{simple.el} may have a hundred or more function
14835 definitions within it. Our long term goal is to collect statistics on
14836 many files, but as a first step, our immediate goal is to collect
14837 statistics on one file.
14839 The information will be a series of numbers, each number being the
14840 length of a function definition. We can store the numbers in a list.
14842 We know that we will want to incorporate the information regarding one
14843 file with information about many other files; this means that the
14844 function for counting definition lengths within one file need only
14845 return the list of lengths. It need not and should not display any
14848 The word count commands contain one expression to jump point forward
14849 word by word and another expression to count the jumps. The function
14850 to return the lengths of definitions can be designed to work the same
14851 way, with one expression to jump point forward definition by
14852 definition and another expression to construct the lengths' list.
14854 This statement of the problem makes it elementary to write the
14855 function definition. Clearly, we will start the count at the
14856 beginning of the file, so the first command will be @code{(goto-char
14857 (point-min))}. Next, we start the @code{while} loop; and the
14858 true-or-false test of the loop can be a regular expression search for
14859 the next function definition---so long as the search succeeds, point
14860 is moved forward and then the body of the loop is evaluated. The body
14861 needs an expression that constructs the lengths' list. @code{cons},
14862 the list construction command, can be used to create the list. That
14863 is almost all there is to it.
14866 Here is what this fragment of code looks like:
14870 (goto-char (point-min))
14871 (while (re-search-forward "^(defun" nil t)
14873 (cons (count-words-in-defun) lengths-list)))
14877 What we have left out is the mechanism for finding the file that
14878 contains the function definitions.
14880 In previous examples, we either used this, the Info file, or we
14881 switched back and forth to some other buffer, such as the
14882 @file{*scratch*} buffer.
14884 Finding a file is a new process that we have not yet discussed.
14887 @section Find a File
14888 @cindex Find a File
14890 To find a file in Emacs, you use the @kbd{C-x C-f} (@code{find-file})
14891 command. This command is almost, but not quite right for the lengths
14895 Let's look at the source for @code{find-file}:
14899 (defun find-file (filename)
14900 "Edit file FILENAME.
14901 Switch to a buffer visiting file FILENAME,
14902 creating one if none already exists."
14903 (interactive "FFind file: ")
14904 (switch-to-buffer (find-file-noselect filename)))
14909 (The most recent version of the @code{find-file} function definition
14910 permits you to specify optional wildcards to visit multiple files; that
14911 makes the definition more complex and we will not discuss it here,
14912 since it is not relevant. You can see its source using either
14913 @kbd{M-.} (@code{find-tag}) or @kbd{C-h f} (@code{describe-function}).)
14917 (defun find-file (filename &optional wildcards)
14918 "Edit file FILENAME.
14919 Switch to a buffer visiting file FILENAME,
14920 creating one if none already exists.
14921 Interactively, the default if you just type RET is the current directory,
14922 but the visited file name is available through the minibuffer history:
14923 type M-n to pull it into the minibuffer.
14925 Interactively, or if WILDCARDS is non-nil in a call from Lisp,
14926 expand wildcards (if any) and visit multiple files. You can
14927 suppress wildcard expansion by setting `find-file-wildcards' to nil.
14929 To visit a file without any kind of conversion and without
14930 automatically choosing a major mode, use \\[find-file-literally]."
14931 (interactive (find-file-read-args "Find file: " nil))
14932 (let ((value (find-file-noselect filename nil nil wildcards)))
14934 (mapcar 'switch-to-buffer (nreverse value))
14935 (switch-to-buffer value))))
14938 The definition I am showing possesses short but complete documentation
14939 and an interactive specification that prompts you for a file name when
14940 you use the command interactively. The body of the definition
14941 contains two functions, @code{find-file-noselect} and
14942 @code{switch-to-buffer}.
14944 According to its documentation as shown by @kbd{C-h f} (the
14945 @code{describe-function} command), the @code{find-file-noselect}
14946 function reads the named file into a buffer and returns the buffer.
14947 (Its most recent version includes an optional wildcards argument,
14948 too, as well as another to read a file literally and an other you
14949 suppress warning messages. These optional arguments are irrelevant.)
14951 However, the @code{find-file-noselect} function does not select the
14952 buffer in which it puts the file. Emacs does not switch its attention
14953 (or yours if you are using @code{find-file-noselect}) to the selected
14954 buffer. That is what @code{switch-to-buffer} does: it switches the
14955 buffer to which Emacs attention is directed; and it switches the
14956 buffer displayed in the window to the new buffer. We have discussed
14957 buffer switching elsewhere. (@xref{Switching Buffers}.)
14959 In this histogram project, we do not need to display each file on the
14960 screen as the program determines the length of each definition within
14961 it. Instead of employing @code{switch-to-buffer}, we can work with
14962 @code{set-buffer}, which redirects the attention of the computer
14963 program to a different buffer but does not redisplay it on the screen.
14964 So instead of calling on @code{find-file} to do the job, we must write
14965 our own expression.
14967 The task is easy: use @code{find-file-noselect} and @code{set-buffer}.
14969 @node lengths-list-file
14970 @section @code{lengths-list-file} in Detail
14972 The core of the @code{lengths-list-file} function is a @code{while}
14973 loop containing a function to move point forward ``defun by defun'' and
14974 a function to count the number of words and symbols in each defun.
14975 This core must be surrounded by functions that do various other tasks,
14976 including finding the file, and ensuring that point starts out at the
14977 beginning of the file. The function definition looks like this:
14978 @findex lengths-list-file
14982 (defun lengths-list-file (filename)
14983 "Return list of definitions' lengths within FILE.
14984 The returned list is a list of numbers.
14985 Each number is the number of words or
14986 symbols in one function definition."
14989 (message "Working on '%s' ... " filename)
14991 (let ((buffer (find-file-noselect filename))
14993 (set-buffer buffer)
14994 (setq buffer-read-only t)
14996 (goto-char (point-min))
14997 (while (re-search-forward "^(defun" nil t)
14999 (cons (count-words-in-defun) lengths-list)))
15000 (kill-buffer buffer)
15006 The function is passed one argument, the name of the file on which it
15007 will work. It has four lines of documentation, but no interactive
15008 specification. Since people worry that a computer is broken if they
15009 don't see anything going on, the first line of the body is a
15012 The next line contains a @code{save-excursion} that returns Emacs's
15013 attention to the current buffer when the function completes. This is
15014 useful in case you embed this function in another function that
15015 presumes point is restored to the original buffer.
15017 In the varlist of the @code{let} expression, Emacs finds the file and
15018 binds the local variable @code{buffer} to the buffer containing the
15019 file. At the same time, Emacs creates @code{lengths-list} as a local
15022 Next, Emacs switches its attention to the buffer.
15024 In the following line, Emacs makes the buffer read-only. Ideally,
15025 this line is not necessary. None of the functions for counting words
15026 and symbols in a function definition should change the buffer.
15027 Besides, the buffer is not going to be saved, even if it were changed.
15028 This line is entirely the consequence of great, perhaps excessive,
15029 caution. The reason for the caution is that this function and those
15030 it calls work on the sources for Emacs and it is inconvenient if they
15031 are inadvertently modified. It goes without saying that I did not
15032 realize a need for this line until an experiment went awry and started
15033 to modify my Emacs source files @dots{}
15035 Next comes a call to widen the buffer if it is narrowed. This
15036 function is usually not needed---Emacs creates a fresh buffer if none
15037 already exists; but if a buffer visiting the file already exists Emacs
15038 returns that one. In this case, the buffer may be narrowed and must
15039 be widened. If we wanted to be fully ``user-friendly'', we would
15040 arrange to save the restriction and the location of point, but we
15043 The @code{(goto-char (point-min))} expression moves point to the
15044 beginning of the buffer.
15046 Then comes a @code{while} loop in which the ``work'' of the function is
15047 carried out. In the loop, Emacs determines the length of each
15048 definition and constructs a lengths' list containing the information.
15050 Emacs kills the buffer after working through it. This is to save
15051 space inside of Emacs. My version of GNU Emacs 19 contained over 300
15052 source files of interest; GNU Emacs 22 contains over a thousand source
15053 files. Another function will apply @code{lengths-list-file} to each
15056 Finally, the last expression within the @code{let} expression is the
15057 @code{lengths-list} variable; its value is returned as the value of
15058 the whole function.
15060 You can try this function by installing it in the usual fashion. Then
15061 place your cursor after the following expression and type @kbd{C-x
15062 C-e} (@code{eval-last-sexp}).
15064 @c !!! 22.1.1 lisp sources location here
15067 "/usr/local/share/emacs/22.1/lisp/emacs-lisp/debug.el")
15071 You may need to change the pathname of the file; the one here is for
15072 GNU Emacs version 22.1. To change the expression, copy it to
15073 the @file{*scratch*} buffer and edit it.
15077 Also, to see the full length of the list, rather than a truncated
15078 version, you may have to evaluate the following:
15079 @c We do not want to insert, so do not mention the zero prefix argument.
15082 (custom-set-variables '(eval-expression-print-length nil))
15086 (@xref{defcustom, , Specifying Variables using @code{defcustom}}.
15087 Then evaluate the @code{lengths-list-file} expression.)
15090 The lengths' list for @file{debug.el} takes less than a second to
15091 produce and looks like this in GNU Emacs 22:
15094 (83 113 105 144 289 22 30 97 48 89 25 52 52 88 28 29 77 49 43 290 232 587)
15098 (Using my old machine, the version 19 lengths' list for @file{debug.el}
15099 took seven seconds to produce and looked like this:
15102 (75 41 80 62 20 45 44 68 45 12 34 235)
15106 The newer version of @file{debug.el} contains more defuns than the
15107 earlier one; and my new machine is much faster than the old one.)
15109 Note that the length of the last definition in the file is first in
15112 @node Several files
15113 @section Count Words in @code{defuns} in Different Files
15115 In the previous section, we created a function that returns a list of
15116 the lengths of each definition in a file. Now, we want to define a
15117 function to return a master list of the lengths of the definitions in
15120 Working on each of a list of files is a repetitious act, so we can use
15121 either a @code{while} loop or recursion.
15124 * lengths-list-many-files:: Return a list of the lengths of defuns.
15125 * append:: Attach one list to another.
15129 @node lengths-list-many-files
15130 @unnumberedsubsec Determine the lengths of @code{defuns}
15133 The design using a @code{while} loop is routine. The argument passed
15134 the function is a list of files. As we saw earlier (@pxref{Loop
15135 Example}), you can write a @code{while} loop so that the body of the
15136 loop is evaluated if such a list contains elements, but to exit the
15137 loop if the list is empty. For this design to work, the body of the
15138 loop must contain an expression that shortens the list each time the
15139 body is evaluated, so that eventually the list is empty. The usual
15140 technique is to set the value of the list to the value of the @sc{cdr}
15141 of the list each time the body is evaluated.
15144 The template looks like this:
15148 (while @var{test-whether-list-is-empty}
15150 @var{set-list-to-cdr-of-list})
15154 Also, we remember that a @code{while} loop returns @code{nil} (the
15155 result of evaluating the true-or-false-test), not the result of any
15156 evaluation within its body. (The evaluations within the body of the
15157 loop are done for their side effects.) However, the expression that
15158 sets the lengths' list is part of the body---and that is the value
15159 that we want returned by the function as a whole. To do this, we
15160 enclose the @code{while} loop within a @code{let} expression, and
15161 arrange that the last element of the @code{let} expression contains
15162 the value of the lengths' list. (@xref{Incrementing Example, , Loop
15163 Example with an Incrementing Counter}.)
15165 @findex lengths-list-many-files
15167 These considerations lead us directly to the function itself:
15171 ;;; @r{Use @code{while} loop.}
15172 (defun lengths-list-many-files (list-of-files)
15173 "Return list of lengths of defuns in LIST-OF-FILES."
15176 (let (lengths-list)
15178 ;;; @r{true-or-false-test}
15179 (while list-of-files
15184 ;;; @r{Generate a lengths' list.}
15186 (expand-file-name (car list-of-files)))))
15190 ;;; @r{Make files' list shorter.}
15191 (setq list-of-files (cdr list-of-files)))
15193 ;;; @r{Return final value of lengths' list.}
15198 @code{expand-file-name} is a built-in function that converts a file
15199 name to the absolute, long, path name form. The function employs the
15200 name of the directory in which the function is called.
15202 @c !!! 22.1.1 lisp sources location here
15204 Thus, if @code{expand-file-name} is called on @code{debug.el} when
15205 Emacs is visiting the
15206 @file{/usr/local/share/emacs/22.1.1/lisp/emacs-lisp/} directory,
15216 @c !!! 22.1.1 lisp sources location here
15218 /usr/local/share/emacs/22.1.1/lisp/emacs-lisp/debug.el
15221 The only other new element of this function definition is the as yet
15222 unstudied function @code{append}, which merits a short section for
15226 @subsection The @code{append} Function
15229 The @code{append} function attaches one list to another. Thus,
15232 (append '(1 2 3 4) '(5 6 7 8))
15243 This is exactly how we want to attach two lengths' lists produced by
15244 @code{lengths-list-file} to each other. The results contrast with
15248 (cons '(1 2 3 4) '(5 6 7 8))
15253 which constructs a new list in which the first argument to @code{cons}
15254 becomes the first element of the new list:
15257 ((1 2 3 4) 5 6 7 8)
15260 @node Several files recursively
15261 @section Recursively Count Words in Different Files
15263 Besides a @code{while} loop, you can work on each of a list of files
15264 with recursion. A recursive version of @code{lengths-list-many-files}
15265 is short and simple.
15267 The recursive function has the usual parts: the ``do-again-test'', the
15268 ``next-step-expression'', and the recursive call. The ``do-again-test''
15269 determines whether the function should call itself again, which it
15270 will do if the @code{list-of-files} contains any remaining elements;
15271 the ``next-step-expression'' resets the @code{list-of-files} to the
15272 @sc{cdr} of itself, so eventually the list will be empty; and the
15273 recursive call calls itself on the shorter list. The complete
15274 function is shorter than this description!
15275 @findex recursive-lengths-list-many-files
15279 (defun recursive-lengths-list-many-files (list-of-files)
15280 "Return list of lengths of each defun in LIST-OF-FILES."
15281 (if list-of-files ; @r{do-again-test}
15284 (expand-file-name (car list-of-files)))
15285 (recursive-lengths-list-many-files
15286 (cdr list-of-files)))))
15291 In a sentence, the function returns the lengths' list for the first of
15292 the @code{list-of-files} appended to the result of calling itself on
15293 the rest of the @code{list-of-files}.
15295 Here is a test of @code{recursive-lengths-list-many-files}, along with
15296 the results of running @code{lengths-list-file} on each of the files
15299 Install @code{recursive-lengths-list-many-files} and
15300 @code{lengths-list-file}, if necessary, and then evaluate the
15301 following expressions. You may need to change the files' pathnames;
15302 those here work when this Info file and the Emacs sources are located
15303 in their customary places. To change the expressions, copy them to
15304 the @file{*scratch*} buffer, edit them, and then evaluate them.
15306 The results are shown after the @samp{@result{}}. (These results are
15307 for files from Emacs version 22.1.1; files from other versions of
15308 Emacs may produce different results.)
15310 @c !!! 22.1.1 lisp sources location here
15313 (cd "/usr/local/share/emacs/22.1.1/")
15315 (lengths-list-file "./lisp/macros.el")
15316 @result{} (283 263 480 90)
15320 (lengths-list-file "./lisp/mail/mailalias.el")
15321 @result{} (38 32 29 95 178 180 321 218 324)
15325 (lengths-list-file "./lisp/makesum.el")
15330 (recursive-lengths-list-many-files
15331 '("./lisp/macros.el"
15332 "./lisp/mail/mailalias.el"
15333 "./lisp/makesum.el"))
15334 @result{} (283 263 480 90 38 32 29 95 178 180 321 218 324 85 181)
15338 The @code{recursive-lengths-list-many-files} function produces the
15341 The next step is to prepare the data in the list for display in a graph.
15343 @node Prepare the data
15344 @section Prepare the Data for Display in a Graph
15346 The @code{recursive-lengths-list-many-files} function returns a list
15347 of numbers. Each number records the length of a function definition.
15348 What we need to do now is transform this data into a list of numbers
15349 suitable for generating a graph. The new list will tell how many
15350 functions definitions contain less than 10 words and
15351 symbols, how many contain between 10 and 19 words and symbols, how
15352 many contain between 20 and 29 words and symbols, and so on.
15354 In brief, we need to go through the lengths' list produced by the
15355 @code{recursive-lengths-list-many-files} function and count the number
15356 of defuns within each range of lengths, and produce a list of those
15360 * Data for Display in Detail::
15361 * Sorting:: Sorting lists.
15362 * Files List:: Making a list of files.
15363 * Counting function definitions::
15367 @node Data for Display in Detail
15368 @unnumberedsubsec The Data for Display in Detail
15371 Based on what we have done before, we can readily foresee that it
15372 should not be too hard to write a function that ``@sc{cdr}s'' down the
15373 lengths' list, looks at each element, determines which length range it
15374 is in, and increments a counter for that range.
15376 However, before beginning to write such a function, we should consider
15377 the advantages of sorting the lengths' list first, so the numbers are
15378 ordered from smallest to largest. First, sorting will make it easier
15379 to count the numbers in each range, since two adjacent numbers will
15380 either be in the same length range or in adjacent ranges. Second, by
15381 inspecting a sorted list, we can discover the highest and lowest
15382 number, and thereby determine the largest and smallest length range
15386 @subsection Sorting Lists
15389 Emacs contains a function to sort lists, called (as you might guess)
15390 @code{sort}. The @code{sort} function takes two arguments, the list
15391 to be sorted, and a predicate that determines whether the first of
15392 two list elements is ``less'' than the second.
15394 As we saw earlier (@pxref{Wrong Type of Argument, , Using the Wrong
15395 Type Object as an Argument}), a predicate is a function that
15396 determines whether some property is true or false. The @code{sort}
15397 function will reorder a list according to whatever property the
15398 predicate uses; this means that @code{sort} can be used to sort
15399 non-numeric lists by non-numeric criteria---it can, for example,
15400 alphabetize a list.
15403 The @code{<} function is used when sorting a numeric list. For example,
15406 (sort '(4 8 21 17 33 7 21 7) '<)
15414 (4 7 7 8 17 21 21 33)
15418 (Note that in this example, both the arguments are quoted so that the
15419 symbols are not evaluated before being passed to @code{sort} as
15422 Sorting the list returned by the
15423 @code{recursive-lengths-list-many-files} function is straightforward;
15424 it uses the @code{<} function:
15428 In GNU Emacs 22, eval
15430 (cd "/usr/local/share/emacs/22.0.50/")
15432 (recursive-lengths-list-many-files
15433 '("./lisp/macros.el"
15434 "./lisp/mail/mailalias.el"
15435 "./lisp/makesum.el"))
15443 (recursive-lengths-list-many-files
15444 '("./lisp/macros.el"
15445 "./lisp/mailalias.el"
15446 "./lisp/makesum.el"))
15456 (29 32 38 85 90 95 178 180 181 218 263 283 321 324 480)
15460 (Note that in this example, the first argument to @code{sort} is not
15461 quoted, since the expression must be evaluated so as to produce the
15462 list that is passed to @code{sort}.)
15465 @subsection Making a List of Files
15467 The @code{recursive-lengths-list-many-files} function requires a list
15468 of files as its argument. For our test examples, we constructed such
15469 a list by hand; but the Emacs Lisp source directory is too large for
15470 us to do for that. Instead, we will write a function to do the job
15471 for us. In this function, we will use both a @code{while} loop and a
15474 @findex directory-files
15475 We did not have to write a function like this for older versions of
15476 GNU Emacs, since they placed all the @samp{.el} files in one
15477 directory. Instead, we were able to use the @code{directory-files}
15478 function, which lists the names of files that match a specified
15479 pattern within a single directory.
15481 However, recent versions of Emacs place Emacs Lisp files in
15482 sub-directories of the top level @file{lisp} directory. This
15483 re-arrangement eases navigation. For example, all the mail related
15484 files are in a @file{lisp} sub-directory called @file{mail}. But at
15485 the same time, this arrangement forces us to create a file listing
15486 function that descends into the sub-directories.
15488 @findex files-in-below-directory
15489 We can create this function, called @code{files-in-below-directory},
15490 using familiar functions such as @code{car}, @code{nthcdr}, and
15491 @code{substring} in conjunction with an existing function called
15492 @code{directory-files-and-attributes}. This latter function not only
15493 lists all the filenames in a directory, including the names
15494 of sub-directories, but also their attributes.
15496 To restate our goal: to create a function that will enable us
15497 to feed filenames to @code{recursive-lengths-list-many-files}
15498 as a list that looks like this (but with more elements):
15502 ("./lisp/macros.el"
15503 "./lisp/mail/rmail.el"
15504 "./lisp/makesum.el")
15508 The @code{directory-files-and-attributes} function returns a list of
15509 lists. Each of the lists within the main list consists of 13
15510 elements. The first element is a string that contains the name of the
15511 file---which, in GNU/Linux, may be a ``directory file'', that is to
15512 say, a file with the special attributes of a directory. The second
15513 element of the list is @code{t} for a directory, a string
15514 for symbolic link (the string is the name linked to), or @code{nil}.
15516 For example, the first @samp{.el} file in the @file{lisp/} directory
15517 is @file{abbrev.el}. Its name is
15518 @file{/usr/local/share/emacs/22.1.1/lisp/abbrev.el} and it is not a
15519 directory or a symbolic link.
15522 This is how @code{directory-files-and-attributes} lists that file and
15534 (20615 27034 579989 697000)
15536 (20615 26327 734791 805000)
15548 On the other hand, @file{mail/} is a directory within the @file{lisp/}
15549 directory. The beginning of its listing looks like this:
15560 (To learn about the different attributes, look at the documentation of
15561 @code{file-attributes}. Bear in mind that the @code{file-attributes}
15562 function does not list the filename, so its first element is
15563 @code{directory-files-and-attributes}'s second element.)
15565 We will want our new function, @code{files-in-below-directory}, to
15566 list the @samp{.el} files in the directory it is told to check, and in
15567 any directories below that directory.
15569 This gives us a hint on how to construct
15570 @code{files-in-below-directory}: within a directory, the function
15571 should add @samp{.el} filenames to a list; and if, within a directory,
15572 the function comes upon a sub-directory, it should go into that
15573 sub-directory and repeat its actions.
15575 However, we should note that every directory contains a name that
15576 refers to itself, called @file{.}, (``dot'') and a name that refers to
15577 its parent directory, called @file{..} (``double dot''). (In
15578 @file{/}, the root directory, @file{..} refers to itself, since
15579 @file{/} has no parent.) Clearly, we do not want our
15580 @code{files-in-below-directory} function to enter those directories,
15581 since they always lead us, directly or indirectly, to the current
15584 Consequently, our @code{files-in-below-directory} function must do
15589 Check to see whether it is looking at a filename that ends in
15590 @samp{.el}; and if so, add its name to a list.
15593 Check to see whether it is looking at a filename that is the name of a
15594 directory; and if so,
15598 Check to see whether it is looking at @file{.} or @file{..}; and if
15602 Or else, go into that directory and repeat the process.
15606 Let's write a function definition to do these tasks. We will use a
15607 @code{while} loop to move from one filename to another within a
15608 directory, checking what needs to be done; and we will use a recursive
15609 call to repeat the actions on each sub-directory. The recursive
15610 pattern is ``accumulate''
15611 (@pxref{Accumulate}),
15612 using @code{append} as the combiner.
15615 (directory-files "/usr/local/src/emacs/lisp/" t "\\.el$")
15616 (shell-command "find /usr/local/src/emacs/lisp/ -name '*.el'")
15618 (directory-files "/usr/local/share/emacs/22.1.1/lisp/" t "\\.el$")
15619 (shell-command "find /usr/local/share/emacs/22.1.1/lisp/ -name '*.el'")
15622 @c /usr/local/share/emacs/22.1.1/lisp/
15625 Here is the function:
15629 (defun files-in-below-directory (directory)
15630 "List the .el files in DIRECTORY and in its sub-directories."
15631 ;; Although the function will be used non-interactively,
15632 ;; it will be easier to test if we make it interactive.
15633 ;; The directory will have a name such as
15634 ;; "/usr/local/share/emacs/22.1.1/lisp/"
15635 (interactive "DDirectory name: ")
15638 (let (el-files-list
15639 (current-directory-list
15640 (directory-files-and-attributes directory t)))
15641 ;; while we are in the current directory
15642 (while current-directory-list
15646 ;; check to see whether filename ends in '.el'
15647 ;; and if so, append its name to a list.
15648 ((equal ".el" (substring (car (car current-directory-list)) -3))
15649 (setq el-files-list
15650 (cons (car (car current-directory-list)) el-files-list)))
15653 ;; check whether filename is that of a directory
15654 ((eq t (car (cdr (car current-directory-list))))
15655 ;; decide whether to skip or recurse
15658 (substring (car (car current-directory-list)) -1))
15659 ;; then do nothing since filename is that of
15660 ;; current directory or parent, "." or ".."
15664 ;; else descend into the directory and repeat the process
15665 (setq el-files-list
15667 (files-in-below-directory
15668 (car (car current-directory-list)))
15670 ;; move to the next filename in the list; this also
15671 ;; shortens the list so the while loop eventually comes to an end
15672 (setq current-directory-list (cdr current-directory-list)))
15673 ;; return the filenames
15678 @c (files-in-below-directory "/usr/local/src/emacs/lisp/")
15679 @c (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/")
15681 The @code{files-in-below-directory} @code{directory-files} function
15682 takes one argument, the name of a directory.
15685 Thus, on my system,
15687 @c (length (files-in-below-directory "/usr/local/src/emacs/lisp/"))
15689 @c !!! 22.1.1 lisp sources location here
15693 (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/"))
15698 tells me that in and below my Lisp sources directory are 1031
15701 @code{files-in-below-directory} returns a list in reverse alphabetical
15702 order. An expression to sort the list in alphabetical order looks
15708 (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/")
15715 "Test how long it takes to find lengths of all sorted elisp defuns."
15716 (insert "\n" (current-time-string) "\n")
15719 (recursive-lengths-list-many-files
15720 (files-in-below-directory "/usr/local/src/emacs/lisp/"))
15722 (insert (format "%s" (current-time-string))))
15725 @node Counting function definitions
15726 @subsection Counting function definitions
15728 Our immediate goal is to generate a list that tells us how many
15729 function definitions contain fewer than 10 words and symbols, how many
15730 contain between 10 and 19 words and symbols, how many contain between
15731 20 and 29 words and symbols, and so on.
15733 With a sorted list of numbers, this is easy: count how many elements
15734 of the list are smaller than 10, then, after moving past the numbers
15735 just counted, count how many are smaller than 20, then, after moving
15736 past the numbers just counted, count how many are smaller than 30, and
15737 so on. Each of the numbers, 10, 20, 30, 40, and the like, is one
15738 larger than the top of that range. We can call the list of such
15739 numbers the @code{top-of-ranges} list.
15742 If we wished, we could generate this list automatically, but it is
15743 simpler to write a list manually. Here it is:
15744 @vindex top-of-ranges
15748 (defvar top-of-ranges
15751 110 120 130 140 150
15752 160 170 180 190 200
15753 210 220 230 240 250
15754 260 270 280 290 300)
15755 "List specifying ranges for `defuns-per-range'.")
15759 To change the ranges, we edit this list.
15761 Next, we need to write the function that creates the list of the
15762 number of definitions within each range. Clearly, this function must
15763 take the @code{sorted-lengths} and the @code{top-of-ranges} lists
15766 The @code{defuns-per-range} function must do two things again and
15767 again: it must count the number of definitions within a range
15768 specified by the current top-of-range value; and it must shift to the
15769 next higher value in the @code{top-of-ranges} list after counting the
15770 number of definitions in the current range. Since each of these
15771 actions is repetitive, we can use @code{while} loops for the job.
15772 One loop counts the number of definitions in the range defined by the
15773 current top-of-range value, and the other loop selects each of the
15774 top-of-range values in turn.
15776 Several entries of the @code{sorted-lengths} list are counted for each
15777 range; this means that the loop for the @code{sorted-lengths} list
15778 will be inside the loop for the @code{top-of-ranges} list, like a
15779 small gear inside a big gear.
15781 The inner loop counts the number of definitions within the range. It
15782 is a simple counting loop of the type we have seen before.
15783 (@xref{Incrementing Loop, , A loop with an incrementing counter}.)
15784 The true-or-false test of the loop tests whether the value from the
15785 @code{sorted-lengths} list is smaller than the current value of the
15786 top of the range. If it is, the function increments the counter and
15787 tests the next value from the @code{sorted-lengths} list.
15790 The inner loop looks like this:
15794 (while @var{length-element-smaller-than-top-of-range}
15795 (setq number-within-range (1+ number-within-range))
15796 (setq sorted-lengths (cdr sorted-lengths)))
15800 The outer loop must start with the lowest value of the
15801 @code{top-of-ranges} list, and then be set to each of the succeeding
15802 higher values in turn. This can be done with a loop like this:
15806 (while top-of-ranges
15807 @var{body-of-loop}@dots{}
15808 (setq top-of-ranges (cdr top-of-ranges)))
15813 Put together, the two loops look like this:
15817 (while top-of-ranges
15819 ;; @r{Count the number of elements within the current range.}
15820 (while @var{length-element-smaller-than-top-of-range}
15821 (setq number-within-range (1+ number-within-range))
15822 (setq sorted-lengths (cdr sorted-lengths)))
15824 ;; @r{Move to next range.}
15825 (setq top-of-ranges (cdr top-of-ranges)))
15829 In addition, in each circuit of the outer loop, Emacs should record
15830 the number of definitions within that range (the value of
15831 @code{number-within-range}) in a list. We can use @code{cons} for
15832 this purpose. (@xref{cons, , @code{cons}}.)
15834 The @code{cons} function works fine, except that the list it
15835 constructs will contain the number of definitions for the highest
15836 range at its beginning and the number of definitions for the lowest
15837 range at its end. This is because @code{cons} attaches new elements
15838 of the list to the beginning of the list, and since the two loops are
15839 working their way through the lengths' list from the lower end first,
15840 the @code{defuns-per-range-list} will end up largest number first.
15841 But we will want to print our graph with smallest values first and the
15842 larger later. The solution is to reverse the order of the
15843 @code{defuns-per-range-list}. We can do this using the
15844 @code{nreverse} function, which reverses the order of a list.
15851 (nreverse '(1 2 3 4))
15862 Note that the @code{nreverse} function is ``destructive''---that is,
15863 it changes the list to which it is applied; this contrasts with the
15864 @code{car} and @code{cdr} functions, which are non-destructive. In
15865 this case, we do not want the original @code{defuns-per-range-list},
15866 so it does not matter that it is destroyed. (The @code{reverse}
15867 function provides a reversed copy of a list, leaving the original list
15872 Put all together, the @code{defuns-per-range} looks like this:
15876 (defun defuns-per-range (sorted-lengths top-of-ranges)
15877 "SORTED-LENGTHS defuns in each TOP-OF-RANGES range."
15878 (let ((top-of-range (car top-of-ranges))
15879 (number-within-range 0)
15880 defuns-per-range-list)
15885 (while top-of-ranges
15891 ;; @r{Need number for numeric test.}
15892 (car sorted-lengths)
15893 (< (car sorted-lengths) top-of-range))
15897 ;; @r{Count number of definitions within current range.}
15898 (setq number-within-range (1+ number-within-range))
15899 (setq sorted-lengths (cdr sorted-lengths)))
15901 ;; @r{Exit inner loop but remain within outer loop.}
15905 (setq defuns-per-range-list
15906 (cons number-within-range defuns-per-range-list))
15907 (setq number-within-range 0) ; @r{Reset count to zero.}
15911 ;; @r{Move to next range.}
15912 (setq top-of-ranges (cdr top-of-ranges))
15913 ;; @r{Specify next top of range value.}
15914 (setq top-of-range (car top-of-ranges)))
15918 ;; @r{Exit outer loop and count the number of defuns larger than}
15919 ;; @r{ the largest top-of-range value.}
15920 (setq defuns-per-range-list
15922 (length sorted-lengths)
15923 defuns-per-range-list))
15927 ;; @r{Return a list of the number of definitions within each range,}
15928 ;; @r{ smallest to largest.}
15929 (nreverse defuns-per-range-list)))
15935 The function is straightforward except for one subtle feature. The
15936 true-or-false test of the inner loop looks like this:
15940 (and (car sorted-lengths)
15941 (< (car sorted-lengths) top-of-range))
15947 instead of like this:
15950 (< (car sorted-lengths) top-of-range)
15953 The purpose of the test is to determine whether the first item in the
15954 @code{sorted-lengths} list is less than the value of the top of the
15957 The simple version of the test works fine unless the
15958 @code{sorted-lengths} list has a @code{nil} value. In that case, the
15959 @code{(car sorted-lengths)} expression function returns
15960 @code{nil}. The @code{<} function cannot compare a number to
15961 @code{nil}, which is an empty list, so Emacs signals an error and
15962 stops the function from attempting to continue to execute.
15964 The @code{sorted-lengths} list always becomes @code{nil} when the
15965 counter reaches the end of the list. This means that any attempt to
15966 use the @code{defuns-per-range} function with the simple version of
15967 the test will fail.
15969 We solve the problem by using the @code{(car sorted-lengths)}
15970 expression in conjunction with the @code{and} expression. The
15971 @code{(car sorted-lengths)} expression returns a non-@code{nil}
15972 value so long as the list has at least one number within it, but
15973 returns @code{nil} if the list is empty. The @code{and} expression
15974 first evaluates the @code{(car sorted-lengths)} expression, and
15975 if it is @code{nil}, returns false @emph{without} evaluating the
15976 @code{<} expression. But if the @code{(car sorted-lengths)}
15977 expression returns a non-@code{nil} value, the @code{and} expression
15978 evaluates the @code{<} expression, and returns that value as the value
15979 of the @code{and} expression.
15981 @c colon in printed section title causes problem in Info cross reference
15982 This way, we avoid an error.
15985 (For information about @code{and}, see
15986 @ref{kill-new function, , The @code{kill-new} function}.)
15990 (@xref{kill-new function, , The @code{kill-new} function}, for
15991 information about @code{and}.)
15994 Here is a short test of the @code{defuns-per-range} function. First,
15995 evaluate the expression that binds (a shortened)
15996 @code{top-of-ranges} list to the list of values, then evaluate the
15997 expression for binding the @code{sorted-lengths} list, and then
15998 evaluate the @code{defuns-per-range} function.
16002 ;; @r{(Shorter list than we will use later.)}
16003 (setq top-of-ranges
16004 '(110 120 130 140 150
16005 160 170 180 190 200))
16007 (setq sorted-lengths
16008 '(85 86 110 116 122 129 154 176 179 200 265 300 300))
16010 (defuns-per-range sorted-lengths top-of-ranges)
16016 The list returned looks like this:
16019 (2 2 2 0 0 1 0 2 0 0 4)
16023 Indeed, there are two elements of the @code{sorted-lengths} list
16024 smaller than 110, two elements between 110 and 119, two elements
16025 between 120 and 129, and so on. There are four elements with a value
16028 @c The next step is to turn this numbers' list into a graph.
16029 @node Readying a Graph
16030 @chapter Readying a Graph
16031 @cindex Readying a graph
16032 @cindex Graph prototype
16033 @cindex Prototype graph
16034 @cindex Body of graph
16036 Our goal is to construct a graph showing the numbers of function
16037 definitions of various lengths in the Emacs lisp sources.
16039 As a practical matter, if you were creating a graph, you would
16040 probably use a program such as @code{gnuplot} to do the job.
16041 (@code{gnuplot} is nicely integrated into GNU Emacs.) In this case,
16042 however, we create one from scratch, and in the process we will
16043 re-acquaint ourselves with some of what we learned before and learn
16046 In this chapter, we will first write a simple graph printing function.
16047 This first definition will be a @dfn{prototype}, a rapidly written
16048 function that enables us to reconnoiter this unknown graph-making
16049 territory. We will discover dragons, or find that they are myth.
16050 After scouting the terrain, we will feel more confident and enhance
16051 the function to label the axes automatically.
16054 * Columns of a graph::
16055 * graph-body-print:: How to print the body of a graph.
16056 * recursive-graph-body-print::
16058 * Line Graph Exercise::
16062 @node Columns of a graph
16063 @unnumberedsec Printing the Columns of a Graph
16066 Since Emacs is designed to be flexible and work with all kinds of
16067 terminals, including character-only terminals, the graph will need to
16068 be made from one of the ``typewriter'' symbols. An asterisk will do; as
16069 we enhance the graph-printing function, we can make the choice of
16070 symbol a user option.
16072 We can call this function @code{graph-body-print}; it will take a
16073 @code{numbers-list} as its only argument. At this stage, we will not
16074 label the graph, but only print its body.
16076 The @code{graph-body-print} function inserts a vertical column of
16077 asterisks for each element in the @code{numbers-list}. The height of
16078 each line is determined by the value of that element of the
16079 @code{numbers-list}.
16081 Inserting columns is a repetitive act; that means that this function can
16082 be written either with a @code{while} loop or recursively.
16084 Our first challenge is to discover how to print a column of asterisks.
16085 Usually, in Emacs, we print characters onto a screen horizontally,
16086 line by line, by typing. We have two routes we can follow: write our
16087 own column-insertion function or discover whether one exists in Emacs.
16089 To see whether there is one in Emacs, we can use the @kbd{M-x apropos}
16090 command. This command is like the @kbd{C-h a} (@code{command-apropos})
16091 command, except that the latter finds only those functions that are
16092 commands. The @kbd{M-x apropos} command lists all symbols that match
16093 a regular expression, including functions that are not interactive.
16096 What we want to look for is some command that prints or inserts
16097 columns. Very likely, the name of the function will contain either
16098 the word ``print'' or the word ``insert'' or the word ``column''.
16099 Therefore, we can simply type @kbd{M-x apropos RET
16100 print\|insert\|column RET} and look at the result. On my system, this
16101 command once too takes quite some time, and then produced a list of 79
16102 functions and variables. Now it does not take much time at all and
16103 produces a list of 211 functions and variables. Scanning down the
16104 list, the only function that looks as if it might do the job is
16105 @code{insert-rectangle}.
16108 Indeed, this is the function we want; its documentation says:
16113 Insert text of RECTANGLE with upper left corner at point.
16114 RECTANGLE's first line is inserted at point,
16115 its second line is inserted at a point vertically under point, etc.
16116 RECTANGLE should be a list of strings.
16117 After this command, the mark is at the upper left corner
16118 and point is at the lower right corner.
16122 We can run a quick test, to make sure it does what we expect of it.
16124 Here is the result of placing the cursor after the
16125 @code{insert-rectangle} expression and typing @kbd{C-u C-x C-e}
16126 (@code{eval-last-sexp}). The function inserts the strings
16127 @samp{"first"}, @samp{"second"}, and @samp{"third"} at and below
16128 point. Also the function returns @code{nil}.
16132 (insert-rectangle '("first" "second" "third"))first
16139 Of course, we won't be inserting the text of the
16140 @code{insert-rectangle} expression itself into the buffer in which we
16141 are making the graph, but will call the function from our program. We
16142 shall, however, have to make sure that point is in the buffer at the
16143 place where the @code{insert-rectangle} function will insert its
16146 If you are reading this in Info, you can see how this works by
16147 switching to another buffer, such as the @file{*scratch*} buffer,
16148 placing point somewhere in the buffer, typing @kbd{M-:}, typing the
16149 @code{insert-rectangle} expression into the minibuffer at the prompt,
16150 and then typing @key{RET}. This causes Emacs to evaluate the
16151 expression in the minibuffer, but to use as the value of point the
16152 position of point in the @file{*scratch*} buffer. (@kbd{M-:} is the
16153 keybinding for @code{eval-expression}. Also, @code{nil} does not
16154 appear in the @file{*scratch*} buffer since the expression is
16155 evaluated in the minibuffer.)
16157 We find when we do this that point ends up at the end of the last
16158 inserted line---that is to say, this function moves point as a
16159 side-effect. If we were to repeat the command, with point at this
16160 position, the next insertion would be below and to the right of the
16161 previous insertion. We don't want this! If we are going to make a
16162 bar graph, the columns need to be beside each other.
16164 So we discover that each cycle of the column-inserting @code{while}
16165 loop must reposition point to the place we want it, and that place
16166 will be at the top, not the bottom, of the column. Moreover, we
16167 remember that when we print a graph, we do not expect all the columns
16168 to be the same height. This means that the top of each column may be
16169 at a different height from the previous one. We cannot simply
16170 reposition point to the same line each time, but moved over to the
16171 right---or perhaps we can@dots{}
16173 We are planning to make the columns of the bar graph out of asterisks.
16174 The number of asterisks in the column is the number specified by the
16175 current element of the @code{numbers-list}. We need to construct a
16176 list of asterisks of the right length for each call to
16177 @code{insert-rectangle}. If this list consists solely of the requisite
16178 number of asterisks, then we will have position point the right number
16179 of lines above the base for the graph to print correctly. This could
16182 Alternatively, if we can figure out some way to pass
16183 @code{insert-rectangle} a list of the same length each time, then we
16184 can place point on the same line each time, but move it over one
16185 column to the right for each new column. If we do this, however, some
16186 of the entries in the list passed to @code{insert-rectangle} must be
16187 blanks rather than asterisks. For example, if the maximum height of
16188 the graph is 5, but the height of the column is 3, then
16189 @code{insert-rectangle} requires an argument that looks like this:
16192 (" " " " "*" "*" "*")
16195 This last proposal is not so difficult, so long as we can determine
16196 the column height. There are two ways for us to specify the column
16197 height: we can arbitrarily state what it will be, which would work
16198 fine for graphs of that height; or we can search through the list of
16199 numbers and use the maximum height of the list as the maximum height
16200 of the graph. If the latter operation were difficult, then the former
16201 procedure would be easiest, but there is a function built into Emacs
16202 that determines the maximum of its arguments. We can use that
16203 function. The function is called @code{max} and it returns the
16204 largest of all its arguments, which must be numbers. Thus, for
16212 returns 7. (A corresponding function called @code{min} returns the
16213 smallest of all its arguments.)
16217 However, we cannot simply call @code{max} on the @code{numbers-list};
16218 the @code{max} function expects numbers as its argument, not a list of
16219 numbers. Thus, the following expression,
16222 (max '(3 4 6 5 7 3))
16227 produces the following error message;
16230 Wrong type of argument: number-or-marker-p, (3 4 6 5 7 3)
16234 We need a function that passes a list of arguments to a function.
16235 This function is @code{apply}. This function ``applies'' its first
16236 argument (a function) to its remaining arguments, the last of which
16243 (apply 'max 3 4 7 3 '(4 8 5))
16249 (Incidentally, I don't know how you would learn of this function
16250 without a book such as this. It is possible to discover other
16251 functions, like @code{search-forward} or @code{insert-rectangle}, by
16252 guessing at a part of their names and then using @code{apropos}. Even
16253 though its base in metaphor is clear---``apply'' its first argument to
16254 the rest---I doubt a novice would come up with that particular word
16255 when using @code{apropos} or other aid. Of course, I could be wrong;
16256 after all, the function was first named by someone who had to invent
16259 The second and subsequent arguments to @code{apply} are optional, so
16260 we can use @code{apply} to call a function and pass the elements of a
16261 list to it, like this, which also returns 8:
16264 (apply 'max '(4 8 5))
16267 This latter way is how we will use @code{apply}. The
16268 @code{recursive-lengths-list-many-files} function returns a numbers'
16269 list to which we can apply @code{max} (we could also apply @code{max} to
16270 the sorted numbers' list; it does not matter whether the list is
16274 Hence, the operation for finding the maximum height of the graph is this:
16277 (setq max-graph-height (apply 'max numbers-list))
16280 Now we can return to the question of how to create a list of strings
16281 for a column of the graph. Told the maximum height of the graph
16282 and the number of asterisks that should appear in the column, the
16283 function should return a list of strings for the
16284 @code{insert-rectangle} command to insert.
16286 Each column is made up of asterisks or blanks. Since the function is
16287 passed the value of the height of the column and the number of
16288 asterisks in the column, the number of blanks can be found by
16289 subtracting the number of asterisks from the height of the column.
16290 Given the number of blanks and the number of asterisks, two
16291 @code{while} loops can be used to construct the list:
16295 ;;; @r{First version.}
16296 (defun column-of-graph (max-graph-height actual-height)
16297 "Return list of strings that is one column of a graph."
16298 (let ((insert-list nil)
16299 (number-of-top-blanks
16300 (- max-graph-height actual-height)))
16304 ;; @r{Fill in asterisks.}
16305 (while (> actual-height 0)
16306 (setq insert-list (cons "*" insert-list))
16307 (setq actual-height (1- actual-height)))
16311 ;; @r{Fill in blanks.}
16312 (while (> number-of-top-blanks 0)
16313 (setq insert-list (cons " " insert-list))
16314 (setq number-of-top-blanks
16315 (1- number-of-top-blanks)))
16319 ;; @r{Return whole list.}
16324 If you install this function and then evaluate the following
16325 expression you will see that it returns the list as desired:
16328 (column-of-graph 5 3)
16336 (" " " " "*" "*" "*")
16339 As written, @code{column-of-graph} contains a major flaw: the symbols
16340 used for the blank and for the marked entries in the column are
16341 ``hard-coded'' as a space and asterisk. This is fine for a prototype,
16342 but you, or another user, may wish to use other symbols. For example,
16343 in testing the graph function, you many want to use a period in place
16344 of the space, to make sure the point is being repositioned properly
16345 each time the @code{insert-rectangle} function is called; or you might
16346 want to substitute a @samp{+} sign or other symbol for the asterisk.
16347 You might even want to make a graph-column that is more than one
16348 display column wide. The program should be more flexible. The way to
16349 do that is to replace the blank and the asterisk with two variables
16350 that we can call @code{graph-blank} and @code{graph-symbol} and define
16351 those variables separately.
16353 Also, the documentation is not well written. These considerations
16354 lead us to the second version of the function:
16358 (defvar graph-symbol "*"
16359 "String used as symbol in graph, usually an asterisk.")
16363 (defvar graph-blank " "
16364 "String used as blank in graph, usually a blank space.
16365 graph-blank must be the same number of columns wide
16371 (For an explanation of @code{defvar}, see
16372 @ref{defvar, , Initializing a Variable with @code{defvar}}.)
16376 ;;; @r{Second version.}
16377 (defun column-of-graph (max-graph-height actual-height)
16378 "Return MAX-GRAPH-HEIGHT strings; ACTUAL-HEIGHT are graph-symbols.
16382 The graph-symbols are contiguous entries at the end
16384 The list will be inserted as one column of a graph.
16385 The strings are either graph-blank or graph-symbol."
16389 (let ((insert-list nil)
16390 (number-of-top-blanks
16391 (- max-graph-height actual-height)))
16395 ;; @r{Fill in @code{graph-symbols}.}
16396 (while (> actual-height 0)
16397 (setq insert-list (cons graph-symbol insert-list))
16398 (setq actual-height (1- actual-height)))
16402 ;; @r{Fill in @code{graph-blanks}.}
16403 (while (> number-of-top-blanks 0)
16404 (setq insert-list (cons graph-blank insert-list))
16405 (setq number-of-top-blanks
16406 (1- number-of-top-blanks)))
16408 ;; @r{Return whole list.}
16413 If we wished, we could rewrite @code{column-of-graph} a third time to
16414 provide optionally for a line graph as well as for a bar graph. This
16415 would not be hard to do. One way to think of a line graph is that it
16416 is no more than a bar graph in which the part of each bar that is
16417 below the top is blank. To construct a column for a line graph, the
16418 function first constructs a list of blanks that is one shorter than
16419 the value, then it uses @code{cons} to attach a graph symbol to the
16420 list; then it uses @code{cons} again to attach the ``top blanks'' to
16423 It is easy to see how to write such a function, but since we don't
16424 need it, we will not do it. But the job could be done, and if it were
16425 done, it would be done with @code{column-of-graph}. Even more
16426 important, it is worth noting that few changes would have to be made
16427 anywhere else. The enhancement, if we ever wish to make it, is
16430 Now, finally, we come to our first actual graph printing function.
16431 This prints the body of a graph, not the labels for the vertical and
16432 horizontal axes, so we can call this @code{graph-body-print}.
16434 @node graph-body-print
16435 @section The @code{graph-body-print} Function
16436 @findex graph-body-print
16438 After our preparation in the preceding section, the
16439 @code{graph-body-print} function is straightforward. The function
16440 will print column after column of asterisks and blanks, using the
16441 elements of a numbers' list to specify the number of asterisks in each
16442 column. This is a repetitive act, which means we can use a
16443 decrementing @code{while} loop or recursive function for the job. In
16444 this section, we will write the definition using a @code{while} loop.
16446 The @code{column-of-graph} function requires the height of the graph
16447 as an argument, so we should determine and record that as a local variable.
16449 This leads us to the following template for the @code{while} loop
16450 version of this function:
16454 (defun graph-body-print (numbers-list)
16455 "@var{documentation}@dots{}"
16456 (let ((height @dots{}
16461 (while numbers-list
16462 @var{insert-columns-and-reposition-point}
16463 (setq numbers-list (cdr numbers-list)))))
16468 We need to fill in the slots of the template.
16470 Clearly, we can use the @code{(apply 'max numbers-list)} expression to
16471 determine the height of the graph.
16473 The @code{while} loop will cycle through the @code{numbers-list} one
16474 element at a time. As it is shortened by the @code{(setq numbers-list
16475 (cdr numbers-list))} expression, the @sc{car} of each instance of the
16476 list is the value of the argument for @code{column-of-graph}.
16478 At each cycle of the @code{while} loop, the @code{insert-rectangle}
16479 function inserts the list returned by @code{column-of-graph}. Since
16480 the @code{insert-rectangle} function moves point to the lower right of
16481 the inserted rectangle, we need to save the location of point at the
16482 time the rectangle is inserted, move back to that position after the
16483 rectangle is inserted, and then move horizontally to the next place
16484 from which @code{insert-rectangle} is called.
16486 If the inserted columns are one character wide, as they will be if
16487 single blanks and asterisks are used, the repositioning command is
16488 simply @code{(forward-char 1)}; however, the width of a column may be
16489 greater than one. This means that the repositioning command should be
16490 written @code{(forward-char symbol-width)}. The @code{symbol-width}
16491 itself is the length of a @code{graph-blank} and can be found using
16492 the expression @code{(length graph-blank)}. The best place to bind
16493 the @code{symbol-width} variable to the value of the width of graph
16494 column is in the varlist of the @code{let} expression.
16497 These considerations lead to the following function definition:
16501 (defun graph-body-print (numbers-list)
16502 "Print a bar graph of the NUMBERS-LIST.
16503 The numbers-list consists of the Y-axis values."
16505 (let ((height (apply 'max numbers-list))
16506 (symbol-width (length graph-blank))
16511 (while numbers-list
16512 (setq from-position (point))
16514 (column-of-graph height (car numbers-list)))
16515 (goto-char from-position)
16516 (forward-char symbol-width)
16519 ;; @r{Draw graph column by column.}
16521 (setq numbers-list (cdr numbers-list)))
16524 ;; @r{Place point for X axis labels.}
16525 (forward-line height)
16532 The one unexpected expression in this function is the
16533 @w{@code{(sit-for 0)}} expression in the @code{while} loop. This
16534 expression makes the graph printing operation more interesting to
16535 watch than it would be otherwise. The expression causes Emacs to
16536 ``sit'' or do nothing for a zero length of time and then redraw the
16537 screen. Placed here, it causes Emacs to redraw the screen column by
16538 column. Without it, Emacs would not redraw the screen until the
16541 We can test @code{graph-body-print} with a short list of numbers.
16545 Install @code{graph-symbol}, @code{graph-blank},
16546 @code{column-of-graph}, which are in
16548 @ref{Readying a Graph, , Readying a Graph},
16551 @ref{Columns of a graph},
16553 and @code{graph-body-print}.
16557 Copy the following expression:
16560 (graph-body-print '(1 2 3 4 6 4 3 5 7 6 5 2 3))
16564 Switch to the @file{*scratch*} buffer and place the cursor where you
16565 want the graph to start.
16568 Type @kbd{M-:} (@code{eval-expression}).
16571 Yank the @code{graph-body-print} expression into the minibuffer
16572 with @kbd{C-y} (@code{yank)}.
16575 Press @key{RET} to evaluate the @code{graph-body-print} expression.
16579 Emacs will print a graph like this:
16593 @node recursive-graph-body-print
16594 @section The @code{recursive-graph-body-print} Function
16595 @findex recursive-graph-body-print
16597 The @code{graph-body-print} function may also be written recursively.
16598 The recursive solution is divided into two parts: an outside ``wrapper''
16599 that uses a @code{let} expression to determine the values of several
16600 variables that need only be found once, such as the maximum height of
16601 the graph, and an inside function that is called recursively to print
16605 The ``wrapper'' is uncomplicated:
16609 (defun recursive-graph-body-print (numbers-list)
16610 "Print a bar graph of the NUMBERS-LIST.
16611 The numbers-list consists of the Y-axis values."
16612 (let ((height (apply 'max numbers-list))
16613 (symbol-width (length graph-blank))
16615 (recursive-graph-body-print-internal
16622 The recursive function is a little more difficult. It has four parts:
16623 the ``do-again-test'', the printing code, the recursive call, and the
16624 ``next-step-expression''. The ``do-again-test'' is a @code{when}
16625 expression that determines whether the @code{numbers-list} contains
16626 any remaining elements; if it does, the function prints one column of
16627 the graph using the printing code and calls itself again. The
16628 function calls itself again according to the value produced by the
16629 ``next-step-expression'' which causes the call to act on a shorter
16630 version of the @code{numbers-list}.
16634 (defun recursive-graph-body-print-internal
16635 (numbers-list height symbol-width)
16636 "Print a bar graph.
16637 Used within recursive-graph-body-print function."
16642 (setq from-position (point))
16644 (column-of-graph height (car numbers-list)))
16647 (goto-char from-position)
16648 (forward-char symbol-width)
16649 (sit-for 0) ; @r{Draw graph column by column.}
16650 (recursive-graph-body-print-internal
16651 (cdr numbers-list) height symbol-width)))
16656 After installation, this expression can be tested; here is a sample:
16659 (recursive-graph-body-print '(3 2 5 6 7 5 3 4 6 4 3 2 1))
16663 Here is what @code{recursive-graph-body-print} produces:
16677 Either of these two functions, @code{graph-body-print} or
16678 @code{recursive-graph-body-print}, create the body of a graph.
16681 @section Need for Printed Axes
16683 A graph needs printed axes, so you can orient yourself. For a do-once
16684 project, it may be reasonable to draw the axes by hand using Emacs's
16685 Picture mode; but a graph drawing function may be used more than once.
16687 For this reason, I have written enhancements to the basic
16688 @code{print-graph-body} function that automatically print labels for
16689 the horizontal and vertical axes. Since the label printing functions
16690 do not contain much new material, I have placed their description in
16691 an appendix. @xref{Full Graph, , A Graph with Labeled Axes}.
16693 @node Line Graph Exercise
16696 Write a line graph version of the graph printing functions.
16698 @node Emacs Initialization
16699 @chapter Your @file{.emacs} File
16700 @cindex @file{.emacs} file
16701 @cindex Customizing your @file{.emacs} file
16702 @cindex Initialization file
16704 ``You don't have to like Emacs to like it''---this seemingly
16705 paradoxical statement is the secret of GNU Emacs. The plain, ``out of
16706 the box'' Emacs is a generic tool. Most people who use it, customize
16707 it to suit themselves.
16709 GNU Emacs is mostly written in Emacs Lisp; this means that by writing
16710 expressions in Emacs Lisp you can change or extend Emacs.
16713 * Default Configuration::
16714 * Site-wide Init:: You can write site-wide init files.
16715 * defcustom:: Emacs will write code for you.
16716 * Beginning init File:: How to write a @file{.emacs} init file.
16717 * Text and Auto-fill:: Automatically wrap lines.
16718 * Mail Aliases:: Use abbreviations for email addresses.
16719 * Indent Tabs Mode:: Don't use tabs with @TeX{}
16720 * Keybindings:: Create some personal keybindings.
16721 * Keymaps:: More about key binding.
16722 * Loading Files:: Load (i.e., evaluate) files automatically.
16723 * Autoload:: Make functions available.
16724 * Simple Extension:: Define a function; bind it to a key.
16725 * X11 Colors:: Colors in X.
16727 * Mode Line:: How to customize your mode line.
16731 @node Default Configuration
16732 @unnumberedsec Emacs's Default Configuration
16735 There are those who appreciate Emacs's default configuration. After
16736 all, Emacs starts you in C mode when you edit a C file, starts you in
16737 Fortran mode when you edit a Fortran file, and starts you in
16738 Fundamental mode when you edit an unadorned file. This all makes
16739 sense, if you do not know who is going to use Emacs. Who knows what a
16740 person hopes to do with an unadorned file? Fundamental mode is the
16741 right default for such a file, just as C mode is the right default for
16742 editing C code. (Enough programming languages have syntaxes
16743 that enable them to share or nearly share features, so C mode is
16744 now provided by CC mode, the ``C Collection''.)
16746 But when you do know who is going to use Emacs---you,
16747 yourself---then it makes sense to customize Emacs.
16749 For example, I seldom want Fundamental mode when I edit an
16750 otherwise undistinguished file; I want Text mode. This is why I
16751 customize Emacs: so it suits me.
16753 You can customize and extend Emacs by writing or adapting a
16754 @file{~/.emacs} file. This is your personal initialization file; its
16755 contents, written in Emacs Lisp, tell Emacs what to do.@footnote{You
16756 may also add @file{.el} to @file{~/.emacs} and call it a
16757 @file{~/.emacs.el} file. In the past, you were forbidden to type the
16758 extra keystrokes that the name @file{~/.emacs.el} requires, but now
16759 you may. The new format is consistent with the Emacs Lisp file
16760 naming conventions; the old format saves typing.}
16762 A @file{~/.emacs} file contains Emacs Lisp code. You can write this
16763 code yourself; or you can use Emacs's @code{customize} feature to write
16764 the code for you. You can combine your own expressions and
16765 auto-written Customize expressions in your @file{.emacs} file.
16767 (I myself prefer to write my own expressions, except for those,
16768 particularly fonts, that I find easier to manipulate using the
16769 @code{customize} command. I combine the two methods.)
16771 Most of this chapter is about writing expressions yourself. It
16772 describes a simple @file{.emacs} file; for more information, see
16773 @ref{Init File, , The Init File, emacs, The GNU Emacs Manual}, and
16774 @ref{Init File, , The Init File, elisp, The GNU Emacs Lisp Reference
16777 @node Site-wide Init
16778 @section Site-wide Initialization Files
16780 @cindex @file{default.el} init file
16781 @cindex @file{site-init.el} init file
16782 @cindex @file{site-load.el} init file
16783 In addition to your personal initialization file, Emacs automatically
16784 loads various site-wide initialization files, if they exist. These
16785 have the same form as your @file{.emacs} file, but are loaded by
16788 Two site-wide initialization files, @file{site-load.el} and
16789 @file{site-init.el}, are loaded into Emacs and then ``dumped'' if a
16790 ``dumped'' version of Emacs is created, as is most common. (Dumped
16791 copies of Emacs load more quickly. However, once a file is loaded and
16792 dumped, a change to it does not lead to a change in Emacs unless you
16793 load it yourself or re-dump Emacs. @xref{Building Emacs, , Building
16794 Emacs, elisp, The GNU Emacs Lisp Reference Manual}, and the
16795 @file{INSTALL} file.)
16797 Three other site-wide initialization files are loaded automatically
16798 each time you start Emacs, if they exist. These are
16799 @file{site-start.el}, which is loaded @emph{before} your @file{.emacs}
16800 file, and @file{default.el}, and the terminal type file, which are both
16801 loaded @emph{after} your @file{.emacs} file.
16803 Settings and definitions in your @file{.emacs} file will overwrite
16804 conflicting settings and definitions in a @file{site-start.el} file,
16805 if it exists; but the settings and definitions in a @file{default.el}
16806 or terminal type file will overwrite those in your @file{.emacs} file.
16807 (You can prevent interference from a terminal type file by setting
16808 @code{term-file-prefix} to @code{nil}. @xref{Simple Extension, , A
16809 Simple Extension}.)
16811 @c Rewritten to avoid overfull hbox.
16812 The @file{INSTALL} file that comes in the distribution contains
16813 descriptions of the @file{site-init.el} and @file{site-load.el} files.
16815 The @file{loadup.el}, @file{startup.el}, and @file{loaddefs.el} files
16816 control loading. These files are in the @file{lisp} directory of the
16817 Emacs distribution and are worth perusing.
16819 The @file{loaddefs.el} file contains a good many suggestions as to
16820 what to put into your own @file{.emacs} file, or into a site-wide
16821 initialization file.
16824 @section Specifying Variables using @code{defcustom}
16827 You can specify variables using @code{defcustom} so that you and
16828 others can then use Emacs's @code{customize} feature to set their
16829 values. (You cannot use @code{customize} to write function
16830 definitions; but you can write @code{defuns} in your @file{.emacs}
16831 file. Indeed, you can write any Lisp expression in your @file{.emacs}
16834 The @code{customize} feature depends on the @code{defcustom} macro.
16835 Although you can use @code{defvar} or @code{setq} for variables that
16836 users set, the @code{defcustom} macro is designed for the job.
16838 You can use your knowledge of @code{defvar} for writing the
16839 first three arguments for @code{defcustom}. The first argument to
16840 @code{defcustom} is the name of the variable. The second argument is
16841 the variable's initial value, if any; and this value is set only if
16842 the value has not already been set. The third argument is the
16845 The fourth and subsequent arguments to @code{defcustom} specify types
16846 and options; these are not featured in @code{defvar}. (These
16847 arguments are optional.)
16849 Each of these arguments consists of a keyword followed by a value.
16850 Each keyword starts with the colon character @samp{:}.
16853 For example, the customizable user option variable
16854 @code{text-mode-hook} looks like this:
16858 (defcustom text-mode-hook nil
16859 "Normal hook run when entering Text mode and many related modes."
16861 :options '(turn-on-auto-fill flyspell-mode)
16867 The name of the variable is @code{text-mode-hook}; it has no default
16868 value; and its documentation string tells you what it does.
16870 The @code{:type} keyword tells Emacs the kind of data to which
16871 @code{text-mode-hook} should be set and how to display the value in a
16872 Customization buffer.
16874 The @code{:options} keyword specifies a suggested list of values for
16875 the variable. Usually, @code{:options} applies to a hook.
16876 The list is only a suggestion; it is not exclusive; a person who sets
16877 the variable may set it to other values; the list shown following the
16878 @code{:options} keyword is intended to offer convenient choices to a
16881 Finally, the @code{:group} keyword tells the Emacs Customization
16882 command in which group the variable is located. This tells where to
16885 The @code{defcustom} macro recognizes more than a dozen keywords.
16886 For more information, see @ref{Customization, , Writing Customization
16887 Definitions, elisp, The GNU Emacs Lisp Reference Manual}.
16889 Consider @code{text-mode-hook} as an example.
16891 There are two ways to customize this variable. You can use the
16892 customization command or write the appropriate expressions yourself.
16895 Using the customization command, you can type:
16902 and find that the group for editing files of data is called ``data''.
16903 Enter that group. Text Mode Hook is the first member. You can click
16904 on its various options, such as @code{turn-on-auto-fill}, to set the
16905 values. After you click on the button to
16908 Save for Future Sessions
16912 Emacs will write an expression into your @file{.emacs} file.
16913 It will look like this:
16917 (custom-set-variables
16918 ;; custom-set-variables was added by Custom.
16919 ;; If you edit it by hand, you could mess it up, so be careful.
16920 ;; Your init file should contain only one such instance.
16921 ;; If there is more than one, they won't work right.
16922 '(text-mode-hook (quote (turn-on-auto-fill text-mode-hook-identify))))
16927 (The @code{text-mode-hook-identify} function tells
16928 @code{toggle-text-mode-auto-fill} which buffers are in Text mode.
16929 It comes on automatically.)
16931 The @code{custom-set-variables} function works somewhat differently
16932 than a @code{setq}. While I have never learned the differences, I
16933 modify the @code{custom-set-variables} expressions in my @file{.emacs}
16934 file by hand: I make the changes in what appears to me to be a
16935 reasonable manner and have not had any problems. Others prefer to use
16936 the Customization command and let Emacs do the work for them.
16938 Another @code{custom-set-@dots{}} function is @code{custom-set-faces}.
16939 This function sets the various font faces. Over time, I have set a
16940 considerable number of faces. Some of the time, I re-set them using
16941 @code{customize}; other times, I simply edit the
16942 @code{custom-set-faces} expression in my @file{.emacs} file itself.
16944 The second way to customize your @code{text-mode-hook} is to set it
16945 yourself in your @file{.emacs} file using code that has nothing to do
16946 with the @code{custom-set-@dots{}} functions.
16949 When you do this, and later use @code{customize}, you will see a
16953 CHANGED outside Customize; operating on it here may be unreliable.
16957 This message is only a warning. If you click on the button to
16960 Save for Future Sessions
16964 Emacs will write a @code{custom-set-@dots{}} expression near the end
16965 of your @file{.emacs} file that will be evaluated after your
16966 hand-written expression. It will, therefore, overrule your
16967 hand-written expression. No harm will be done. When you do this,
16968 however, be careful to remember which expression is active; if you
16969 forget, you may confuse yourself.
16971 So long as you remember where the values are set, you will have no
16972 trouble. In any event, the values are always set in your
16973 initialization file, which is usually called @file{.emacs}.
16975 I myself use @code{customize} for hardly anything. Mostly, I write
16976 expressions myself.
16980 Incidentally, to be more complete concerning defines: @code{defsubst}
16981 defines an inline function. The syntax is just like that of
16982 @code{defun}. @code{defconst} defines a symbol as a constant. The
16983 intent is that neither programs nor users should ever change a value
16984 set by @code{defconst}. (You can change it; the value set is a
16985 variable; but please do not.)
16987 @node Beginning init File
16988 @section Beginning a @file{.emacs} File
16989 @cindex @file{.emacs} file, beginning of
16991 When you start Emacs, it loads your @file{.emacs} file unless you tell
16992 it not to by specifying @samp{-q} on the command line. (The
16993 @code{emacs -q} command gives you a plain, out-of-the-box Emacs.)
16995 A @file{.emacs} file contains Lisp expressions. Often, these are no
16996 more than expressions to set values; sometimes they are function
16999 @xref{Init File, , The Init File @file{~/.emacs}, emacs, The GNU Emacs
17000 Manual}, for a short description of initialization files.
17002 This chapter goes over some of the same ground, but is a walk among
17003 extracts from a complete, long-used @file{.emacs} file---my own.
17005 The first part of the file consists of comments: reminders to myself.
17006 By now, of course, I remember these things, but when I started, I did
17012 ;;;; Bob's .emacs file
17013 ; Robert J. Chassell
17014 ; 26 September 1985
17019 Look at that date! I started this file a long time ago. I have been
17020 adding to it ever since.
17024 ; Each section in this file is introduced by a
17025 ; line beginning with four semicolons; and each
17026 ; entry is introduced by a line beginning with
17027 ; three semicolons.
17032 This describes the usual conventions for comments in Emacs Lisp.
17033 Everything on a line that follows a semicolon is a comment. Two,
17034 three, and four semicolons are used as subsection and section markers.
17035 (@xref{Comments, ,, elisp, The GNU Emacs Lisp Reference Manual}, for
17036 more about comments.)
17041 ; Control-h is the help key;
17042 ; after typing control-h, type a letter to
17043 ; indicate the subject about which you want help.
17044 ; For an explanation of the help facility,
17045 ; type control-h two times in a row.
17050 Just remember: type @kbd{C-h} two times for help.
17054 ; To find out about any mode, type control-h m
17055 ; while in that mode. For example, to find out
17056 ; about mail mode, enter mail mode and then type
17062 ``Mode help'', as I call this, is very helpful. Usually, it tells you
17063 all you need to know.
17065 Of course, you don't need to include comments like these in your
17066 @file{.emacs} file. I included them in mine because I kept forgetting
17067 about Mode help or the conventions for comments---but I was able to
17068 remember to look here to remind myself.
17070 @node Text and Auto-fill
17071 @section Text and Auto Fill Mode
17073 Now we come to the part that ``turns on'' Text mode and
17078 ;;; Text mode and Auto Fill mode
17079 ;; The next two lines put Emacs into Text mode
17080 ;; and Auto Fill mode, and are for writers who
17081 ;; want to start writing prose rather than code.
17082 (setq-default major-mode 'text-mode)
17083 (add-hook 'text-mode-hook 'turn-on-auto-fill)
17087 Here is the first part of this @file{.emacs} file that does something
17088 besides remind a forgetful human!
17090 The first of the two lines in parentheses tells Emacs to turn on Text
17091 mode when you find a file, @emph{unless} that file should go into some
17092 other mode, such as C mode.
17094 @cindex Per-buffer, local variables list
17095 @cindex Local variables list, per-buffer,
17096 @cindex Automatic mode selection
17097 @cindex Mode selection, automatic
17098 When Emacs reads a file, it looks at the extension to the file name,
17099 if any. (The extension is the part that comes after a @samp{.}.) If
17100 the file ends with a @samp{.c} or @samp{.h} extension then Emacs turns
17101 on C mode. Also, Emacs looks at first nonblank line of the file; if
17102 the line says @w{@samp{-*- C -*-}}, Emacs turns on C mode. Emacs
17103 possesses a list of extensions and specifications that it uses
17104 automatically. In addition, Emacs looks near the last page for a
17105 per-buffer, ``local variables list'', if any.
17108 @xref{Choosing Modes, , How Major Modes are Chosen, emacs, The GNU
17111 @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs
17115 See sections ``How Major Modes are Chosen'' and ``Local Variables in
17116 Files'' in @cite{The GNU Emacs Manual}.
17119 Now, back to the @file{.emacs} file.
17122 Here is the line again; how does it work?
17124 @cindex Text Mode turned on
17126 (setq major-mode 'text-mode)
17130 This line is a short, but complete Emacs Lisp expression.
17132 We are already familiar with @code{setq}. It sets the following variable,
17133 @code{major-mode}, to the subsequent value, which is @code{text-mode}.
17134 The single quote mark before @code{text-mode} tells Emacs to deal directly
17135 with the @code{text-mode} symbol, not with whatever it might stand for.
17136 @xref{set & setq, , Setting the Value of a Variable},
17137 for a reminder of how @code{setq} works.
17138 The main point is that there is no difference between the procedure you
17139 use to set a value in your @file{.emacs} file and the procedure you use
17140 anywhere else in Emacs.
17143 Here is the next line:
17145 @cindex Auto Fill mode turned on
17148 (add-hook 'text-mode-hook 'turn-on-auto-fill)
17152 In this line, the @code{add-hook} command adds
17153 @code{turn-on-auto-fill} to the variable.
17155 @code{turn-on-auto-fill} is the name of a program, that, you guessed
17156 it!, turns on Auto Fill mode.
17158 Every time Emacs turns on Text mode, Emacs runs the commands ``hooked''
17159 onto Text mode. So every time Emacs turns on Text mode, Emacs also
17160 turns on Auto Fill mode.
17162 In brief, the first line causes Emacs to enter Text mode when you edit a
17163 file, unless the file name extension, a first non-blank line, or local
17164 variables to tell Emacs otherwise.
17166 Text mode among other actions, sets the syntax table to work
17167 conveniently for writers. In Text mode, Emacs considers an apostrophe
17168 as part of a word like a letter; but Emacs does not consider a period
17169 or a space as part of a word. Thus, @kbd{M-f} moves you over
17170 @samp{it's}. On the other hand, in C mode, @kbd{M-f} stops just after
17171 the @samp{t} of @samp{it's}.
17173 The second line causes Emacs to turn on Auto Fill mode when it turns
17174 on Text mode. In Auto Fill mode, Emacs automatically breaks a line
17175 that is too wide and brings the excessively wide part of the line down
17176 to the next line. Emacs breaks lines between words, not within them.
17178 When Auto Fill mode is turned off, lines continue to the right as you
17179 type them. Depending on how you set the value of
17180 @code{truncate-lines}, the words you type either disappear off the
17181 right side of the screen, or else are shown, in a rather ugly and
17182 unreadable manner, as a continuation line on the screen.
17185 In addition, in this part of my @file{.emacs} file, I tell the Emacs
17186 fill commands to insert two spaces after a colon:
17189 (setq colon-double-space t)
17193 @section Mail Aliases
17195 Here is a @code{setq} that ``turns on'' mail aliases, along with more
17201 ; To enter mail mode, type 'C-x m'
17202 ; To enter RMAIL (for reading mail),
17204 (setq mail-aliases t)
17208 @cindex Mail aliases
17210 This @code{setq} command sets the value of the variable
17211 @code{mail-aliases} to @code{t}. Since @code{t} means true, the line
17212 says, in effect, ``Yes, use mail aliases.''
17214 Mail aliases are convenient short names for long email addresses or
17215 for lists of email addresses. The file where you keep your ``aliases''
17216 is @file{~/.mailrc}. You write an alias like this:
17219 alias geo george@@foobar.wiz.edu
17223 When you write a message to George, address it to @samp{geo}; the
17224 mailer will automatically expand @samp{geo} to the full address.
17226 @node Indent Tabs Mode
17227 @section Indent Tabs Mode
17228 @cindex Tabs, preventing
17229 @findex indent-tabs-mode
17231 By default, Emacs inserts tabs in place of multiple spaces when it
17232 formats a region. (For example, you might indent many lines of text
17233 all at once with the @code{indent-region} command.) Tabs look fine on
17234 a terminal or with ordinary printing, but they produce badly indented
17235 output when you use @TeX{} or Texinfo since @TeX{} ignores tabs.
17238 The following turns off Indent Tabs mode:
17242 ;;; Prevent Extraneous Tabs
17243 (setq-default indent-tabs-mode nil)
17247 Note that this line uses @code{setq-default} rather than the
17248 @code{setq} command that we have seen before. The @code{setq-default}
17249 command sets values only in buffers that do not have their own local
17250 values for the variable.
17253 @xref{Just Spaces, , Tabs vs.@: Spaces, emacs, The GNU Emacs Manual}.
17255 @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs
17259 See sections ``Tabs vs.@: Spaces'' and ``Local Variables in
17260 Files'' in @cite{The GNU Emacs Manual}.
17265 @section Some Keybindings
17267 Now for some personal keybindings:
17271 ;;; Compare windows
17272 (global-set-key "\C-cw" 'compare-windows)
17276 @findex compare-windows
17277 @code{compare-windows} is a nifty command that compares the text in
17278 your current window with text in the next window. It makes the
17279 comparison by starting at point in each window, moving over text in
17280 each window as far as they match. I use this command all the time.
17282 This also shows how to set a key globally, for all modes.
17284 @cindex Setting a key globally
17285 @cindex Global set key
17286 @cindex Key setting globally
17287 @findex global-set-key
17288 The command is @code{global-set-key}. It is followed by the
17289 keybinding. In a @file{.emacs} file, the keybinding is written as
17290 shown: @code{\C-c} stands for ``control-c'', which means ``press the
17291 control key and the @key{c} key at the same time''. The @code{w} means
17292 ``press the @key{w} key''. The keybinding is surrounded by double
17293 quotation marks. In documentation, you would write this as
17294 @w{@kbd{C-c w}}. (If you were binding a @key{META} key, such as
17295 @kbd{M-c}, rather than a @key{CTRL} key, you would write
17296 @w{@code{\M-c}} in your @file{.emacs} file. @xref{Init Rebinding, ,
17297 Rebinding Keys in Your Init File, emacs, The GNU Emacs Manual}, for
17300 The command invoked by the keys is @code{compare-windows}. Note that
17301 @code{compare-windows} is preceded by a single quote; otherwise, Emacs
17302 would first try to evaluate the symbol to determine its value.
17304 These three things, the double quotation marks, the backslash before
17305 the @samp{C}, and the single quote mark are necessary parts of
17306 keybinding that I tend to forget. Fortunately, I have come to
17307 remember that I should look at my existing @file{.emacs} file, and
17308 adapt what is there.
17310 As for the keybinding itself: @kbd{C-c w}. This combines the prefix
17311 key, @kbd{C-c}, with a single character, in this case, @kbd{w}. This
17312 set of keys, @kbd{C-c} followed by a single character, is strictly
17313 reserved for individuals' own use. (I call these ``own'' keys, since
17314 these are for my own use.) You should always be able to create such a
17315 keybinding for your own use without stomping on someone else's
17316 keybinding. If you ever write an extension to Emacs, please avoid
17317 taking any of these keys for public use. Create a key like @kbd{C-c
17318 C-w} instead. Otherwise, we will run out of ``own'' keys.
17321 Here is another keybinding, with a comment:
17325 ;;; Keybinding for 'occur'
17326 ; I use occur a lot, so let's bind it to a key:
17327 (global-set-key "\C-co" 'occur)
17332 The @code{occur} command shows all the lines in the current buffer
17333 that contain a match for a regular expression. Matching lines are
17334 shown in a buffer called @file{*Occur*}. That buffer serves as a menu
17335 to jump to occurrences.
17337 @findex global-unset-key
17338 @cindex Unbinding key
17339 @cindex Key unbinding
17341 Here is how to unbind a key, so it does not
17347 (global-unset-key "\C-xf")
17351 There is a reason for this unbinding: I found I inadvertently typed
17352 @w{@kbd{C-x f}} when I meant to type @kbd{C-x C-f}. Rather than find a
17353 file, as I intended, I accidentally set the width for filled text,
17354 almost always to a width I did not want. Since I hardly ever reset my
17355 default width, I simply unbound the key.
17357 @findex list-buffers, @r{rebound}
17358 @findex buffer-menu, @r{bound to key}
17360 The following rebinds an existing key:
17364 ;;; Rebind 'C-x C-b' for 'buffer-menu'
17365 (global-set-key "\C-x\C-b" 'buffer-menu)
17369 By default, @kbd{C-x C-b} runs the
17370 @code{list-buffers} command. This command lists
17371 your buffers in @emph{another} window. Since I
17372 almost always want to do something in that
17373 window, I prefer the @code{buffer-menu}
17374 command, which not only lists the buffers,
17375 but moves point into that window.
17380 @cindex Rebinding keys
17382 Emacs uses @dfn{keymaps} to record which keys call which commands.
17383 When you use @code{global-set-key} to set the keybinding for a single
17384 command in all parts of Emacs, you are specifying the keybinding in
17385 @code{current-global-map}.
17387 Specific modes, such as C mode or Text mode, have their own keymaps;
17388 the mode-specific keymaps override the global map that is shared by
17391 The @code{global-set-key} function binds, or rebinds, the global
17392 keymap. For example, the following binds the key @kbd{C-x C-b} to the
17393 function @code{buffer-menu}:
17396 (global-set-key "\C-x\C-b" 'buffer-menu)
17399 Mode-specific keymaps are bound using the @code{define-key} function,
17400 which takes a specific keymap as an argument, as well as the key and
17401 the command. For example, my @file{.emacs} file contains the
17402 following expression to bind the @code{texinfo-insert-@@group} command
17403 to @kbd{C-c C-c g}:
17407 (define-key texinfo-mode-map "\C-c\C-cg" 'texinfo-insert-@@group)
17412 The @code{texinfo-insert-@@group} function itself is a little extension
17413 to Texinfo mode that inserts @samp{@@group} into a Texinfo file. I
17414 use this command all the time and prefer to type the three strokes
17415 @kbd{C-c C-c g} rather than the six strokes @kbd{@@ g r o u p}.
17416 (@samp{@@group} and its matching @samp{@@end group} are commands that
17417 keep all enclosed text together on one page; many multi-line examples
17418 in this book are surrounded by @samp{@@group @dots{} @@end group}.)
17421 Here is the @code{texinfo-insert-@@group} function definition:
17425 (defun texinfo-insert-@@group ()
17426 "Insert the string @@group in a Texinfo buffer."
17428 (beginning-of-line)
17429 (insert "@@group\n"))
17433 (Of course, I could have used Abbrev mode to save typing, rather than
17434 write a function to insert a word; but I prefer key strokes consistent
17435 with other Texinfo mode key bindings.)
17437 You will see numerous @code{define-key} expressions in
17438 @file{loaddefs.el} as well as in the various mode libraries, such as
17439 @file{cc-mode.el} and @file{lisp-mode.el}.
17441 @xref{Key Bindings, , Customizing Key Bindings, emacs, The GNU Emacs
17442 Manual}, and @ref{Keymaps, , Keymaps, elisp, The GNU Emacs Lisp
17443 Reference Manual}, for more information about keymaps.
17445 @node Loading Files
17446 @section Loading Files
17447 @cindex Loading files
17450 Many people in the GNU Emacs community have written extensions to
17451 Emacs. As time goes by, these extensions are often included in new
17452 releases. For example, the Calendar and Diary packages are now part
17453 of the standard GNU Emacs, as is Calc.
17455 You can use a @code{load} command to evaluate a complete file and
17456 thereby install all the functions and variables in the file into Emacs.
17459 @c (auto-compression-mode t)
17462 (load "~/emacs/slowsplit")
17465 This evaluates, i.e., loads, the @file{slowsplit.el} file or if it
17466 exists, the faster, byte compiled @file{slowsplit.elc} file from the
17467 @file{emacs} sub-directory of your home directory. The file contains
17468 the function @code{split-window-quietly}, which John Robinson wrote in
17471 The @code{split-window-quietly} function splits a window with the
17472 minimum of redisplay. I installed it in 1989 because it worked well
17473 with the slow 1200 baud terminals I was then using. Nowadays, I only
17474 occasionally come across such a slow connection, but I continue to use
17475 the function because I like the way it leaves the bottom half of a
17476 buffer in the lower of the new windows and the top half in the upper
17480 To replace the key binding for the default
17481 @code{split-window-vertically}, you must also unset that key and bind
17482 the keys to @code{split-window-quietly}, like this:
17486 (global-unset-key "\C-x2")
17487 (global-set-key "\C-x2" 'split-window-quietly)
17492 If you load many extensions, as I do, then instead of specifying the
17493 exact location of the extension file, as shown above, you can specify
17494 that directory as part of Emacs's @code{load-path}. Then, when Emacs
17495 loads a file, it will search that directory as well as its default
17496 list of directories. (The default list is specified in @file{paths.h}
17497 when Emacs is built.)
17500 The following command adds your @file{~/emacs} directory to the
17501 existing load path:
17505 ;;; Emacs Load Path
17506 (setq load-path (cons "~/emacs" load-path))
17510 Incidentally, @code{load-library} is an interactive interface to the
17511 @code{load} function. The complete function looks like this:
17513 @findex load-library
17516 (defun load-library (library)
17517 "Load the library named LIBRARY.
17518 This is an interface to the function `load'."
17520 (list (completing-read "Load library: "
17521 (apply-partially 'locate-file-completion-table
17523 (get-load-suffixes)))))
17528 The name of the function, @code{load-library}, comes from the use of
17529 ``library'' as a conventional synonym for ``file''. The source for the
17530 @code{load-library} command is in the @file{files.el} library.
17532 Another interactive command that does a slightly different job is
17533 @code{load-file}. @xref{Lisp Libraries, , Libraries of Lisp Code for
17534 Emacs, emacs, The GNU Emacs Manual}, for information on the
17535 distinction between @code{load-library} and this command.
17538 @section Autoloading
17541 Instead of installing a function by loading the file that contains it,
17542 or by evaluating the function definition, you can make the function
17543 available but not actually install it until it is first called. This
17544 is called @dfn{autoloading}.
17546 When you execute an autoloaded function, Emacs automatically evaluates
17547 the file that contains the definition, and then calls the function.
17549 Emacs starts quicker with autoloaded functions, since their libraries
17550 are not loaded right away; but you need to wait a moment when you
17551 first use such a function, while its containing file is evaluated.
17553 Rarely used functions are frequently autoloaded. The
17554 @file{loaddefs.el} library contains thousands of autoloaded functions,
17555 from @code{5x5} to @code{zone}. Of course, you may
17556 come to use a ``rare'' function frequently. When you do, you should
17557 load that function's file with a @code{load} expression in your
17558 @file{.emacs} file.
17560 In my @file{.emacs} file, I load 14 libraries that contain functions
17561 that would otherwise be autoloaded. (Actually, it would have been
17562 better to include these files in my ``dumped'' Emacs, but I forgot.
17563 @xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp
17564 Reference Manual}, and the @file{INSTALL} file for more about
17567 You may also want to include autoloaded expressions in your @file{.emacs}
17568 file. @code{autoload} is a built-in function that takes up to five
17569 arguments, the final three of which are optional. The first argument
17570 is the name of the function to be autoloaded; the second is the name
17571 of the file to be loaded. The third argument is documentation for the
17572 function, and the fourth tells whether the function can be called
17573 interactively. The fifth argument tells what type of
17574 object---@code{autoload} can handle a keymap or macro as well as a
17575 function (the default is a function).
17578 Here is a typical example:
17582 (autoload 'html-helper-mode
17583 "html-helper-mode" "Edit HTML documents" t)
17588 (@code{html-helper-mode} is an older alternative to @code{html-mode},
17589 which is a standard part of the distribution.)
17592 This expression autoloads the @code{html-helper-mode} function. It
17593 takes it from the @file{html-helper-mode.el} file (or from the byte
17594 compiled version @file{html-helper-mode.elc}, if that exists.) The
17595 file must be located in a directory specified by @code{load-path}.
17596 The documentation says that this is a mode to help you edit documents
17597 written in the HyperText Markup Language. You can call this mode
17598 interactively by typing @kbd{M-x html-helper-mode}. (You need to
17599 duplicate the function's regular documentation in the autoload
17600 expression because the regular function is not yet loaded, so its
17601 documentation is not available.)
17603 @xref{Autoload, , Autoload, elisp, The GNU Emacs Lisp Reference
17604 Manual}, for more information.
17606 @node Simple Extension
17607 @section A Simple Extension: @code{line-to-top-of-window}
17608 @findex line-to-top-of-window
17609 @cindex Simple extension in @file{.emacs} file
17611 Here is a simple extension to Emacs that moves the line point is on to
17612 the top of the window. I use this all the time, to make text easier
17615 You can put the following code into a separate file and then load it
17616 from your @file{.emacs} file, or you can include it within your
17617 @file{.emacs} file.
17620 Here is the definition:
17624 ;;; Line to top of window;
17625 ;;; replace three keystroke sequence C-u 0 C-l
17626 (defun line-to-top-of-window ()
17627 "Move the line point is on to top of window."
17634 Now for the keybinding.
17636 Nowadays, function keys as well as mouse button events and
17637 non-@sc{ascii} characters are written within square brackets, without
17638 quotation marks. (In Emacs version 18 and before, you had to write
17639 different function key bindings for each different make of terminal.)
17641 I bind @code{line-to-top-of-window} to my @key{F6} function key like
17645 (global-set-key [f6] 'line-to-top-of-window)
17648 For more information, see @ref{Init Rebinding, , Rebinding Keys in
17649 Your Init File, emacs, The GNU Emacs Manual}.
17651 @cindex Conditional 'twixt two versions of Emacs
17652 @cindex Version of Emacs, choosing
17653 @cindex Emacs version, choosing
17654 If you run two versions of GNU Emacs, such as versions 22 and 23, and
17655 use one @file{.emacs} file, you can select which code to evaluate with
17656 the following conditional:
17661 ((= 22 emacs-major-version)
17662 ;; evaluate version 22 code
17664 ((= 23 emacs-major-version)
17665 ;; evaluate version 23 code
17670 For example, recent versions blink
17671 their cursors by default. I hate such blinking, as well as other
17672 features, so I placed the following in my @file{.emacs}
17673 file@footnote{When I start instances of Emacs that do not load my
17674 @file{.emacs} file or any site file, I also turn off blinking:
17677 emacs -q --no-site-file -eval '(blink-cursor-mode nil)'
17679 @exdent Or nowadays, using an even more sophisticated set of options,
17687 (when (>= emacs-major-version 21)
17688 (blink-cursor-mode 0)
17689 ;; Insert newline when you press 'C-n' (next-line)
17690 ;; at the end of the buffer
17691 (setq next-line-add-newlines t)
17694 ;; Turn on image viewing
17695 (auto-image-file-mode t)
17698 ;; Turn on menu bar (this bar has text)
17699 ;; (Use numeric argument to turn on)
17703 ;; Turn off tool bar (this bar has icons)
17704 ;; (Use numeric argument to turn on)
17705 (tool-bar-mode nil)
17708 ;; Turn off tooltip mode for tool bar
17709 ;; (This mode causes icon explanations to pop up)
17710 ;; (Use numeric argument to turn on)
17712 ;; If tooltips turned on, make tips appear promptly
17713 (setq tooltip-delay 0.1) ; default is 0.7 second
17719 @section X11 Colors
17721 You can specify colors when you use Emacs with the MIT X Windowing
17724 I dislike the default colors and specify my own.
17727 Here are the expressions in my @file{.emacs}
17728 file that set values:
17732 ;; Set cursor color
17733 (set-cursor-color "white")
17736 (set-mouse-color "white")
17738 ;; Set foreground and background
17739 (set-foreground-color "white")
17740 (set-background-color "darkblue")
17744 ;;; Set highlighting colors for isearch and drag
17745 (set-face-foreground 'highlight "white")
17746 (set-face-background 'highlight "blue")
17750 (set-face-foreground 'region "cyan")
17751 (set-face-background 'region "blue")
17755 (set-face-foreground 'secondary-selection "skyblue")
17756 (set-face-background 'secondary-selection "darkblue")
17760 ;; Set calendar highlighting colors
17761 (add-hook 'calendar-load-hook
17763 (set-face-foreground 'diary-face "skyblue")
17764 (set-face-background 'holiday-face "slate blue")
17765 (set-face-foreground 'holiday-face "white")))
17769 The various shades of blue soothe my eye and prevent me from seeing
17770 the screen flicker.
17772 Alternatively, I could have set my specifications in various X
17773 initialization files. For example, I could set the foreground,
17774 background, cursor, and pointer (i.e., mouse) colors in my
17775 @file{~/.Xresources} file like this:
17779 Emacs*foreground: white
17780 Emacs*background: darkblue
17781 Emacs*cursorColor: white
17782 Emacs*pointerColor: white
17786 In any event, since it is not part of Emacs, I set the root color of
17787 my X window in my @file{~/.xinitrc} file, like this@footnote{I also
17788 run more modern window managers, such as Enlightenment, Gnome, or KDE;
17789 in those cases, I often specify an image rather than a plain color.}:
17792 xsetroot -solid Navy -fg white &
17796 @node Miscellaneous
17797 @section Miscellaneous Settings for a @file{.emacs} File
17800 Here are a few miscellaneous settings:
17805 Set the shape and color of the mouse cursor:
17809 ; Cursor shapes are defined in
17810 ; '/usr/include/X11/cursorfont.h';
17811 ; for example, the 'target' cursor is number 128;
17812 ; the 'top_left_arrow' cursor is number 132.
17816 (let ((mpointer (x-get-resource "*mpointer"
17817 "*emacs*mpointer")))
17818 ;; If you have not set your mouse pointer
17819 ;; then set it, otherwise leave as is:
17820 (if (eq mpointer nil)
17821 (setq mpointer "132")) ; top_left_arrow
17824 (setq x-pointer-shape (string-to-int mpointer))
17825 (set-mouse-color "white"))
17830 Or you can set the values of a variety of features in an alist, like
17836 default-frame-alist
17837 '((cursor-color . "white")
17838 (mouse-color . "white")
17839 (foreground-color . "white")
17840 (background-color . "DodgerBlue4")
17841 ;; (cursor-type . bar)
17842 (cursor-type . box)
17845 (tool-bar-lines . 0)
17846 (menu-bar-lines . 1)
17850 "-Misc-Fixed-Medium-R-Normal--20-200-75-75-C-100-ISO8859-1")
17856 Convert @kbd{@key{CTRL}-h} into @key{DEL} and @key{DEL}
17857 into @kbd{@key{CTRL}-h}.@*
17858 (Some older keyboards needed this, although I have not seen the
17863 ;; Translate 'C-h' to <DEL>.
17864 ; (keyboard-translate ?\C-h ?\C-?)
17866 ;; Translate <DEL> to 'C-h'.
17867 (keyboard-translate ?\C-? ?\C-h)
17871 @item Turn off a blinking cursor!
17875 (if (fboundp 'blink-cursor-mode)
17876 (blink-cursor-mode -1))
17881 or start GNU Emacs with the command @code{emacs -nbc}.
17884 @item When using @command{grep}@*
17885 @samp{-i}@w{ } Ignore case distinctions@*
17886 @samp{-n}@w{ } Prefix each line of output with line number@*
17887 @samp{-H}@w{ } Print the filename for each match.@*
17888 @samp{-e}@w{ } Protect patterns beginning with a hyphen character, @samp{-}
17891 (setq grep-command "grep -i -nH -e ")
17895 @c Evidently, no longer needed in GNU Emacs 22
17897 item Automatically uncompress compressed files when visiting them
17900 (load "uncompress")
17905 @item Find an existing buffer, even if it has a different name@*
17906 This avoids problems with symbolic links.
17909 (setq find-file-existing-other-name t)
17912 @item Set your language environment and default input method
17916 (set-language-environment "latin-1")
17917 ;; Remember you can enable or disable multilingual text input
17918 ;; with the @code{toggle-input-method'} (@kbd{C-\}) command
17919 (setq default-input-method "latin-1-prefix")
17923 If you want to write with Chinese ``GB'' characters, set this instead:
17927 (set-language-environment "Chinese-GB")
17928 (setq default-input-method "chinese-tonepy")
17933 @subsubheading Fixing Unpleasant Key Bindings
17934 @cindex Key bindings, fixing
17935 @cindex Bindings, key, fixing unpleasant
17937 Some systems bind keys unpleasantly. Sometimes, for example, the
17938 @key{CTRL} key appears in an awkward spot rather than at the far left
17941 Usually, when people fix these sorts of keybindings, they do not
17942 change their @file{~/.emacs} file. Instead, they bind the proper keys
17943 on their consoles with the @code{loadkeys} or @code{install-keymap}
17944 commands in their boot script and then include @code{xmodmap} commands
17945 in their @file{.xinitrc} or @file{.Xsession} file for X Windows.
17953 loadkeys /usr/share/keymaps/i386/qwerty/emacs2.kmap.gz
17955 install-keymap emacs2
17961 For a @file{.xinitrc} or @file{.Xsession} file when the @key{Caps
17962 Lock} key is at the far left of the home row:
17966 # Bind the key labeled 'Caps Lock' to 'Control'
17967 # (Such a broken user interface suggests that keyboard manufacturers
17968 # think that computers are typewriters from 1885.)
17970 xmodmap -e "clear Lock"
17971 xmodmap -e "add Control = Caps_Lock"
17977 In a @file{.xinitrc} or @file{.Xsession} file, to convert an @key{ALT}
17978 key to a @key{META} key:
17982 # Some ill designed keyboards have a key labeled ALT and no Meta
17983 xmodmap -e "keysym Alt_L = Meta_L Alt_L"
17989 @section A Modified Mode Line
17990 @vindex mode-line-format
17991 @cindex Mode line format
17993 Finally, a feature I really like: a modified mode line.
17995 When I work over a network, I forget which machine I am using. Also,
17996 I tend to I lose track of where I am, and which line point is on.
17998 So I reset my mode line to look like this:
18001 -:-- foo.texi rattlesnake:/home/bob/ Line 1 (Texinfo Fill) Top
18004 I am visiting a file called @file{foo.texi}, on my machine
18005 @file{rattlesnake} in my @file{/home/bob} buffer. I am on line 1, in
18006 Texinfo mode, and am at the top of the buffer.
18009 My @file{.emacs} file has a section that looks like this:
18013 ;; Set a Mode Line that tells me which machine, which directory,
18014 ;; and which line I am on, plus the other customary information.
18015 (setq-default mode-line-format
18019 "mouse-1: select window, mouse-2: delete others ..."))
18020 mode-line-mule-info
18022 mode-line-frame-identification
18026 mode-line-buffer-identification
18029 (system-name) 0 (string-match "\\..+" (system-name))))
18034 "mouse-1: select window, mouse-2: delete others ..."))
18035 (line-number-mode " Line %l ")
18041 "mouse-1: select window, mouse-2: delete others ..."))
18042 (:eval (mode-line-mode-name))
18045 #("%n" 0 2 (help-echo "mouse-2: widen" local-map (keymap ...)))
18054 Here, I redefine the default mode line. Most of the parts are from
18055 the original; but I make a few changes. I set the @emph{default} mode
18056 line format so as to permit various modes, such as Info, to override
18059 Many elements in the list are self-explanatory:
18060 @code{mode-line-modified} is a variable that tells whether the buffer
18061 has been modified, @code{mode-name} tells the name of the mode, and so
18062 on. However, the format looks complicated because of two features we
18063 have not discussed.
18065 @cindex Properties, in mode line example
18066 The first string in the mode line is a dash or hyphen, @samp{-}. In
18067 the old days, it would have been specified simply as @code{"-"}. But
18068 nowadays, Emacs can add properties to a string, such as highlighting
18069 or, as in this case, a help feature. If you place your mouse cursor
18070 over the hyphen, some help information appears (By default, you must
18071 wait seven-tenths of a second before the information appears. You can
18072 change that timing by changing the value of @code{tooltip-delay}.)
18075 The new string format has a special syntax:
18078 #("-" 0 1 (help-echo "mouse-1: select window, ..."))
18082 The @code{#(} begins a list. The first element of the list is the
18083 string itself, just one @samp{-}. The second and third
18084 elements specify the range over which the fourth element applies. A
18085 range starts @emph{after} a character, so a zero means the range
18086 starts just before the first character; a 1 means that the range ends
18087 just after the first character. The third element is the property for
18088 the range. It consists of a property list, a
18089 property name, in this case, @samp{help-echo}, followed by a value, in this
18090 case, a string. The second, third, and fourth elements of this new
18091 string format can be repeated.
18093 @xref{Text Properties, , Text Properties, elisp, The GNU Emacs Lisp
18094 Reference Manual}, and see @ref{Mode Line Format, , Mode Line Format,
18095 elisp, The GNU Emacs Lisp Reference Manual}, for more information.
18097 @code{mode-line-buffer-identification}
18098 displays the current buffer name. It is a list
18099 beginning @code{(#("%12b" 0 4 @dots{}}.
18100 The @code{#(} begins the list.
18102 The @samp{"%12b"} displays the current buffer name, using the
18103 @code{buffer-name} function with which we are familiar; the @samp{12}
18104 specifies the maximum number of characters that will be displayed.
18105 When a name has fewer characters, whitespace is added to fill out to
18106 this number. (Buffer names can and often should be longer than 12
18107 characters; this length works well in a typical 80 column wide
18110 @code{:eval} says to evaluate the following form and use the result as
18111 a string to display. In this case, the expression displays the first
18112 component of the full system name. The end of the first component is
18113 a @samp{.} (``period''), so I use the @code{string-match} function to
18114 tell me the length of the first component. The substring from the
18115 zeroth character to that length is the name of the machine.
18118 This is the expression:
18123 (system-name) 0 (string-match "\\..+" (system-name))))
18127 @samp{%[} and @samp{%]} cause a pair of square brackets
18128 to appear for each recursive editing level. @samp{%n} says ``Narrow''
18129 when narrowing is in effect. @samp{%P} tells you the percentage of
18130 the buffer that is above the bottom of the window, or ``Top'', ``Bottom'',
18131 or ``All''. (A lower case @samp{p} tell you the percentage above the
18132 @emph{top} of the window.) @samp{%-} inserts enough dashes to fill
18135 Remember, ``You don't have to like Emacs to like it''---your own
18136 Emacs can have different colors, different commands, and different
18137 keys than a default Emacs.
18139 On the other hand, if you want to bring up a plain ``out of the box''
18140 Emacs, with no customization, type:
18147 This will start an Emacs that does @emph{not} load your
18148 @file{~/.emacs} initialization file. A plain, default Emacs. Nothing
18155 GNU Emacs has two debuggers, @code{debug} and @code{edebug}. The
18156 first is built into the internals of Emacs and is always with you;
18157 the second requires that you instrument a function before you can use it.
18159 Both debuggers are described extensively in @ref{Debugging, ,
18160 Debugging Lisp Programs, elisp, The GNU Emacs Lisp Reference Manual}.
18161 In this chapter, I will walk through a short example of each.
18164 * debug:: How to use the built-in debugger.
18165 * debug-on-entry:: Start debugging when you call a function.
18166 * debug-on-quit:: Start debugging when you quit with @kbd{C-g}.
18167 * edebug:: How to use Edebug, a source level debugger.
18168 * Debugging Exercises::
18172 @section @code{debug}
18175 Suppose you have written a function definition that is intended to
18176 return the sum of the numbers 1 through a given number. (This is the
18177 @code{triangle} function discussed earlier. @xref{Decrementing
18178 Example, , Example with Decrementing Counter}, for a discussion.)
18179 @c xref{Decrementing Loop,, Loop with a Decrementing Counter}, for a discussion.)
18181 However, your function definition has a bug. You have mistyped
18182 @samp{1=} for @samp{1-}. Here is the broken definition:
18184 @findex triangle-bugged
18187 (defun triangle-bugged (number)
18188 "Return sum of numbers 1 through NUMBER inclusive."
18190 (while (> number 0)
18191 (setq total (+ total number))
18192 (setq number (1= number))) ; @r{Error here.}
18197 If you are reading this in Info, you can evaluate this definition in
18198 the normal fashion. You will see @code{triangle-bugged} appear in the
18202 Now evaluate the @code{triangle-bugged} function with an
18206 (triangle-bugged 4)
18210 In a recent GNU Emacs, you will create and enter a @file{*Backtrace*}
18216 ---------- Buffer: *Backtrace* ----------
18217 Debugger entered--Lisp error: (void-function 1=)
18219 (setq number (1= number))
18220 (while (> number 0) (setq total (+ total number))
18221 (setq number (1= number)))
18222 (let ((total 0)) (while (> number 0) (setq total ...)
18223 (setq number ...)) total)
18227 eval((triangle-bugged 4))
18228 eval-last-sexp-1(nil)
18229 eval-last-sexp(nil)
18230 call-interactively(eval-last-sexp)
18231 ---------- Buffer: *Backtrace* ----------
18236 (I have reformatted this example slightly; the debugger does not fold
18237 long lines. As usual, you can quit the debugger by typing @kbd{q} in
18238 the @file{*Backtrace*} buffer.)
18240 In practice, for a bug as simple as this, the ``Lisp error'' line will
18241 tell you what you need to know to correct the definition. The
18242 function @code{1=} is ``void''.
18246 In GNU Emacs 20 and before, you will see:
18249 Symbol's function definition is void:@: 1=
18253 which has the same meaning as the @file{*Backtrace*} buffer line in
18257 However, suppose you are not quite certain what is going on?
18258 You can read the complete backtrace.
18260 In this case, you need to run a recent GNU Emacs, which automatically
18261 starts the debugger that puts you in the @file{*Backtrace*} buffer; or
18262 else, you need to start the debugger manually as described below.
18264 Read the @file{*Backtrace*} buffer from the bottom up; it tells you
18265 what Emacs did that led to the error. Emacs made an interactive call
18266 to @kbd{C-x C-e} (@code{eval-last-sexp}), which led to the evaluation
18267 of the @code{triangle-bugged} expression. Each line above tells you
18268 what the Lisp interpreter evaluated next.
18271 The third line from the top of the buffer is
18274 (setq number (1= number))
18278 Emacs tried to evaluate this expression; in order to do so, it tried
18279 to evaluate the inner expression shown on the second line from the
18288 This is where the error occurred; as the top line says:
18291 Debugger entered--Lisp error: (void-function 1=)
18295 You can correct the mistake, re-evaluate the function definition, and
18296 then run your test again.
18298 @node debug-on-entry
18299 @section @code{debug-on-entry}
18300 @findex debug-on-entry
18302 A recent GNU Emacs starts the debugger automatically when your
18303 function has an error.
18306 GNU Emacs version 20 and before did not; it simply
18307 presented you with an error message. You had to start the debugger
18311 Incidentally, you can start the debugger manually for all versions of
18312 Emacs; the advantage is that the debugger runs even if you do not have
18313 a bug in your code. Sometimes your code will be free of bugs!
18315 You can enter the debugger when you call the function by calling
18316 @code{debug-on-entry}.
18323 M-x debug-on-entry RET triangle-bugged RET
18328 Now, evaluate the following:
18331 (triangle-bugged 5)
18335 All versions of Emacs will create a @file{*Backtrace*} buffer and tell
18336 you that it is beginning to evaluate the @code{triangle-bugged}
18341 ---------- Buffer: *Backtrace* ----------
18342 Debugger entered--entering a function:
18343 * triangle-bugged(5)
18344 eval((triangle-bugged 5))
18347 eval-last-sexp-1(nil)
18348 eval-last-sexp(nil)
18349 call-interactively(eval-last-sexp)
18350 ---------- Buffer: *Backtrace* ----------
18354 In the @file{*Backtrace*} buffer, type @kbd{d}. Emacs will evaluate
18355 the first expression in @code{triangle-bugged}; the buffer will look
18360 ---------- Buffer: *Backtrace* ----------
18361 Debugger entered--beginning evaluation of function call form:
18362 * (let ((total 0)) (while (> number 0) (setq total ...)
18363 (setq number ...)) total)
18364 * triangle-bugged(5)
18365 eval((triangle-bugged 5))
18368 eval-last-sexp-1(nil)
18369 eval-last-sexp(nil)
18370 call-interactively(eval-last-sexp)
18371 ---------- Buffer: *Backtrace* ----------
18376 Now, type @kbd{d} again, eight times, slowly. Each time you type
18377 @kbd{d}, Emacs will evaluate another expression in the function
18381 Eventually, the buffer will look like this:
18385 ---------- Buffer: *Backtrace* ----------
18386 Debugger entered--beginning evaluation of function call form:
18387 * (setq number (1= number))
18388 * (while (> number 0) (setq total (+ total number))
18389 (setq number (1= number)))
18392 * (let ((total 0)) (while (> number 0) (setq total ...)
18393 (setq number ...)) total)
18394 * triangle-bugged(5)
18395 eval((triangle-bugged 5))
18398 eval-last-sexp-1(nil)
18399 eval-last-sexp(nil)
18400 call-interactively(eval-last-sexp)
18401 ---------- Buffer: *Backtrace* ----------
18407 Finally, after you type @kbd{d} two more times, Emacs will reach the
18408 error, and the top two lines of the @file{*Backtrace*} buffer will look
18413 ---------- Buffer: *Backtrace* ----------
18414 Debugger entered--Lisp error: (void-function 1=)
18417 ---------- Buffer: *Backtrace* ----------
18421 By typing @kbd{d}, you were able to step through the function.
18423 You can quit a @file{*Backtrace*} buffer by typing @kbd{q} in it; this
18424 quits the trace, but does not cancel @code{debug-on-entry}.
18426 @findex cancel-debug-on-entry
18427 To cancel the effect of @code{debug-on-entry}, call
18428 @code{cancel-debug-on-entry} and the name of the function, like this:
18431 M-x cancel-debug-on-entry RET triangle-bugged RET
18435 (If you are reading this in Info, cancel @code{debug-on-entry} now.)
18437 @node debug-on-quit
18438 @section @code{debug-on-quit} and @code{(debug)}
18440 In addition to setting @code{debug-on-error} or calling @code{debug-on-entry},
18441 there are two other ways to start @code{debug}.
18443 @findex debug-on-quit
18444 You can start @code{debug} whenever you type @kbd{C-g}
18445 (@code{keyboard-quit}) by setting the variable @code{debug-on-quit} to
18446 @code{t}. This is useful for debugging infinite loops.
18449 @cindex @code{(debug)} in code
18450 Or, you can insert a line that says @code{(debug)} into your code
18451 where you want the debugger to start, like this:
18455 (defun triangle-bugged (number)
18456 "Return sum of numbers 1 through NUMBER inclusive."
18458 (while (> number 0)
18459 (setq total (+ total number))
18460 (debug) ; @r{Start debugger.}
18461 (setq number (1= number))) ; @r{Error here.}
18466 The @code{debug} function is described in detail in @ref{Debugger, ,
18467 The Lisp Debugger, elisp, The GNU Emacs Lisp Reference Manual}.
18470 @section The @code{edebug} Source Level Debugger
18471 @cindex Source level debugger
18474 Edebug is a source level debugger. Edebug normally displays the
18475 source of the code you are debugging, with an arrow at the left that
18476 shows which line you are currently executing.
18478 You can walk through the execution of a function, line by line, or run
18479 quickly until reaching a @dfn{breakpoint} where execution stops.
18481 Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs
18482 Lisp Reference Manual}.
18485 Here is a bugged function definition for @code{triangle-recursively}.
18486 @xref{Recursive triangle function, , Recursion in place of a counter},
18487 for a review of it.
18491 (defun triangle-recursively-bugged (number)
18492 "Return sum of numbers 1 through NUMBER inclusive.
18497 (triangle-recursively-bugged
18498 (1= number))))) ; @r{Error here.}
18503 Normally, you would install this definition by positioning your cursor
18504 after the function's closing parenthesis and typing @kbd{C-x C-e}
18505 (@code{eval-last-sexp}) or else by positioning your cursor within the
18506 definition and typing @kbd{C-M-x} (@code{eval-defun}). (By default,
18507 the @code{eval-defun} command works only in Emacs Lisp mode or in Lisp
18511 However, to prepare this function definition for Edebug, you must
18512 first @dfn{instrument} the code using a different command. You can do
18513 this by positioning your cursor within or just after the definition
18517 M-x edebug-defun RET
18521 This will cause Emacs to load Edebug automatically if it is not
18522 already loaded, and properly instrument the function.
18524 After instrumenting the function, place your cursor after the
18525 following expression and type @kbd{C-x C-e} (@code{eval-last-sexp}):
18528 (triangle-recursively-bugged 3)
18532 You will be jumped back to the source for
18533 @code{triangle-recursively-bugged} and the cursor positioned at the
18534 beginning of the @code{if} line of the function. Also, you will see
18535 an arrowhead at the left hand side of that line. The arrowhead marks
18536 the line where the function is executing. (In the following examples,
18537 we show the arrowhead with @samp{=>}; in a windowing system, you may
18538 see the arrowhead as a solid triangle in the window ``fringe''.)
18541 =>@point{}(if (= number 1)
18546 In the example, the location of point is displayed with a star,
18547 @samp{@point{}} (in Info, it is displayed as @samp{-!-}).
18550 In the example, the location of point is displayed as @samp{@point{}}
18551 (in a printed book, it is displayed with a five pointed star).
18554 If you now press @key{SPC}, point will move to the next expression to
18555 be executed; the line will look like this:
18558 =>(if @point{}(= number 1)
18562 As you continue to press @key{SPC}, point will move from expression to
18563 expression. At the same time, whenever an expression returns a value,
18564 that value will be displayed in the echo area. For example, after you
18565 move point past @code{number}, you will see the following:
18568 Result: 3 (#o3, #x3, ?\C-c)
18572 This means the value of @code{number} is 3, which is octal three,
18573 hexadecimal three, and @sc{ascii} ``control-c'' (the third letter of the
18574 alphabet, in case you need to know this information).
18576 You can continue moving through the code until you reach the line with
18577 the error. Before evaluation, that line looks like this:
18580 => @point{}(1= number))))) ; @r{Error here.}
18585 When you press @key{SPC} once again, you will produce an error message
18589 Symbol's function definition is void:@: 1=
18595 Press @kbd{q} to quit Edebug.
18597 To remove instrumentation from a function definition, simply
18598 re-evaluate it with a command that does not instrument it.
18599 For example, you could place your cursor after the definition's
18600 closing parenthesis and type @kbd{C-x C-e}.
18602 Edebug does a great deal more than walk with you through a function.
18603 You can set it so it races through on its own, stopping only at an
18604 error or at specified stopping points; you can cause it to display the
18605 changing values of various expressions; you can find out how many
18606 times a function is called, and more.
18608 Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs
18609 Lisp Reference Manual}.
18612 @node Debugging Exercises
18613 @section Debugging Exercises
18617 Install the @code{@value{COUNT-WORDS}} function and then cause it to
18618 enter the built-in debugger when you call it. Run the command on a
18619 region containing two words. You will need to press @kbd{d} a
18620 remarkable number of times. On your system, is a ``hook'' called after
18621 the command finishes? (For information on hooks, see @ref{Command
18622 Overview, , Command Loop Overview, elisp, The GNU Emacs Lisp Reference
18626 Copy @code{@value{COUNT-WORDS}} into the @file{*scratch*} buffer,
18627 instrument the function for Edebug, and walk through its execution.
18628 The function does not need to have a bug, although you can introduce
18629 one if you wish. If the function lacks a bug, the walk-through
18630 completes without problems.
18633 While running Edebug, type @kbd{?} to see a list of all the Edebug commands.
18634 (The @code{global-edebug-prefix} is usually @kbd{C-x X}, i.e.,
18635 @kbd{@key{CTRL}-x} followed by an upper case @kbd{X}; use this prefix
18636 for commands made outside of the Edebug debugging buffer.)
18639 In the Edebug debugging buffer, use the @kbd{p}
18640 (@code{edebug-bounce-point}) command to see where in the region the
18641 @code{@value{COUNT-WORDS}} is working.
18644 Move point to some spot further down the function and then type the
18645 @kbd{h} (@code{edebug-goto-here}) command to jump to that location.
18648 Use the @kbd{t} (@code{edebug-trace-mode}) command to cause Edebug to
18649 walk through the function on its own; use an upper case @kbd{T} for
18650 @code{edebug-Trace-fast-mode}.
18653 Set a breakpoint, then run Edebug in Trace mode until it reaches the
18658 @chapter Conclusion
18660 We have now reached the end of this Introduction. You have now
18661 learned enough about programming in Emacs Lisp to set values, to write
18662 simple @file{.emacs} files for yourself and your friends, and write
18663 simple customizations and extensions to Emacs.
18665 This is a place to stop. Or, if you wish, you can now go onward, and
18668 You have learned some of the basic nuts and bolts of programming. But
18669 only some. There are a great many more brackets and hinges that are
18670 easy to use that we have not touched.
18672 A path you can follow right now lies among the sources to GNU Emacs
18675 @cite{The GNU Emacs Lisp Reference Manual}.
18678 @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU
18679 Emacs Lisp Reference Manual}.
18682 The Emacs Lisp sources are an adventure. When you read the sources and
18683 come across a function or expression that is unfamiliar, you need to
18684 figure out or find out what it does.
18686 Go to the Reference Manual. It is a thorough, complete, and fairly
18687 easy-to-read description of Emacs Lisp. It is written not only for
18688 experts, but for people who know what you know. (The @cite{Reference
18689 Manual} comes with the standard GNU Emacs distribution. Like this
18690 introduction, it comes as a Texinfo source file, so you can read it
18691 on your computer and as a typeset, printed book.)
18693 Go to the other built-in help that is part of GNU Emacs: the built-in
18694 documentation for all functions and variables, and @code{find-tag},
18695 the program that takes you to sources.
18697 Here is an example of how I explore the sources. Because of its name,
18698 @file{simple.el} is the file I looked at first, a long time ago. As
18699 it happens some of the functions in @file{simple.el} are complicated,
18700 or at least look complicated at first sight. The @code{open-line}
18701 function, for example, looks complicated.
18703 You may want to walk through this function slowly, as we did with the
18704 @code{forward-sentence} function. (@xref{forward-sentence, The
18705 @code{forward-sentence} function}.) Or you may want to skip that
18706 function and look at another, such as @code{split-line}. You don't
18707 need to read all the functions. According to
18708 @code{count-words-in-defun}, the @code{split-line} function contains
18709 102 words and symbols.
18711 Even though it is short, @code{split-line} contains expressions
18712 we have not studied: @code{skip-chars-forward}, @code{indent-to},
18713 @code{current-column} and @code{insert-and-inherit}.
18715 Consider the @code{skip-chars-forward} function. (It is part of the
18716 function definition for @code{back-to-indentation}, which is shown in
18717 @ref{Review, , Review}.)
18719 In GNU Emacs, you can find out more about @code{skip-chars-forward} by
18720 typing @kbd{C-h f} (@code{describe-function}) and the name of the
18721 function. This gives you the function documentation.
18723 You may be able to guess what is done by a well named function such as
18724 @code{indent-to}; or you can look it up, too. Incidentally, the
18725 @code{describe-function} function itself is in @file{help.el}; it is
18726 one of those long, but decipherable functions. You can look up
18727 @code{describe-function} using the @kbd{C-h f} command!
18729 In this instance, since the code is Lisp, the @file{*Help*} buffer
18730 contains the name of the library containing the function's source.
18731 You can put point over the name of the library and press the RET key,
18732 which in this situation is bound to @code{help-follow}, and be taken
18733 directly to the source, in the same way as @kbd{M-.}
18736 The definition for @code{describe-function} illustrates how to
18737 customize the @code{interactive} expression without using the standard
18738 character codes; and it shows how to create a temporary buffer.
18740 (The @code{indent-to} function is written in C rather than Emacs Lisp;
18741 it is a ``built-in'' function. @code{help-follow} takes you to its
18742 source as does @code{find-tag}, when properly set up.)
18744 You can look at a function's source using @code{find-tag}, which is
18745 bound to @kbd{M-.} Finally, you can find out what the Reference
18746 Manual has to say by visiting the manual in Info, and typing @kbd{i}
18747 (@code{Info-index}) and the name of the function, or by looking up the
18748 function in the index to a printed copy of the manual.
18750 Similarly, you can find out what is meant by
18751 @code{insert-and-inherit}.
18753 Other interesting source files include @file{paragraphs.el},
18754 @file{loaddefs.el}, and @file{loadup.el}. The @file{paragraphs.el}
18755 file includes short, easily understood functions as well as longer
18756 ones. The @file{loaddefs.el} file contains the many standard
18757 autoloads and many keymaps. I have never looked at it all; only at
18758 parts. @file{loadup.el} is the file that loads the standard parts of
18759 Emacs; it tells you a great deal about how Emacs is built.
18760 (@xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp
18761 Reference Manual}, for more about building.)
18763 As I said, you have learned some nuts and bolts; however, and very
18764 importantly, we have hardly touched major aspects of programming; I
18765 have said nothing about how to sort information, except to use the
18766 predefined @code{sort} function; I have said nothing about how to store
18767 information, except to use variables and lists; I have said nothing
18768 about how to write programs that write programs. These are topics for
18769 another, and different kind of book, a different kind of learning.
18771 What you have done is learn enough for much practical work with GNU
18772 Emacs. What you have done is get started. This is the end of a
18775 @c ================ Appendix ================
18778 @appendix The @code{the-the} Function
18780 @cindex Duplicated words function
18781 @cindex Words, duplicated
18783 Sometimes when you you write text, you duplicate words---as with ``you
18784 you'' near the beginning of this sentence. I find that most
18785 frequently, I duplicate ``the''; hence, I call the function for
18786 detecting duplicated words, @code{the-the}.
18789 As a first step, you could use the following regular expression to
18790 search for duplicates:
18793 \\(\\w+[ \t\n]+\\)\\1
18797 This regexp matches one or more word-constituent characters followed
18798 by one or more spaces, tabs, or newlines. However, it does not detect
18799 duplicated words on different lines, since the ending of the first
18800 word, the end of the line, is different from the ending of the second
18801 word, a space. (For more information about regular expressions, see
18802 @ref{Regexp Search, , Regular Expression Searches}, as well as
18803 @ref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs
18804 Manual}, and @ref{Regular Expressions, , Regular Expressions, elisp,
18805 The GNU Emacs Lisp Reference Manual}.)
18807 You might try searching just for duplicated word-constituent
18808 characters but that does not work since the pattern detects doubles
18809 such as the two occurrences of ``th'' in ``with the''.
18811 Another possible regexp searches for word-constituent characters
18812 followed by non-word-constituent characters, reduplicated. Here,
18813 @w{@samp{\\w+}} matches one or more word-constituent characters and
18814 @w{@samp{\\W*}} matches zero or more non-word-constituent characters.
18817 \\(\\(\\w+\\)\\W*\\)\\1
18823 Here is the pattern that I use. It is not perfect, but good enough.
18824 @w{@samp{\\b}} matches the empty string, provided it is at the beginning
18825 or end of a word; @w{@samp{[^@@ \n\t]+}} matches one or more occurrences of
18826 any characters that are @emph{not} an @@-sign, space, newline, or tab.
18829 \\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b
18832 One can write more complicated expressions, but I found that this
18833 expression is good enough, so I use it.
18835 Here is the @code{the-the} function, as I include it in my
18836 @file{.emacs} file, along with a handy global key binding:
18841 "Search forward for for a duplicated word."
18843 (message "Searching for for duplicated words ...")
18847 ;; This regexp is not perfect
18848 ;; but is fairly good over all:
18849 (if (re-search-forward
18850 "\\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b" nil 'move)
18851 (message "Found duplicated word.")
18852 (message "End of buffer")))
18856 ;; Bind 'the-the' to C-c \
18857 (global-set-key "\C-c\\" 'the-the)
18866 one two two three four five
18871 You can substitute the other regular expressions shown above in the
18872 function definition and try each of them on this list.
18875 @appendix Handling the Kill Ring
18876 @cindex Kill ring handling
18877 @cindex Handling the kill ring
18878 @cindex Ring, making a list like a
18880 The kill ring is a list that is transformed into a ring by the
18881 workings of the @code{current-kill} function. The @code{yank} and
18882 @code{yank-pop} commands use the @code{current-kill} function.
18884 This appendix describes the @code{current-kill} function as well as
18885 both the @code{yank} and the @code{yank-pop} commands, but first,
18886 consider the workings of the kill ring.
18889 * What the Kill Ring Does::
18891 * yank:: Paste a copy of a clipped element.
18892 * yank-pop:: Insert element pointed to.
18897 @node What the Kill Ring Does
18898 @unnumberedsec What the Kill Ring Does
18902 The kill ring has a default maximum length of sixty items; this number
18903 is too large for an explanation. Instead, set it to four. Please
18904 evaluate the following:
18908 (setq old-kill-ring-max kill-ring-max)
18909 (setq kill-ring-max 4)
18914 Then, please copy each line of the following indented example into the
18915 kill ring. You may kill each line with @kbd{C-k} or mark it and copy
18919 (In a read-only buffer, such as the @file{*info*} buffer, the kill
18920 command, @kbd{C-k} (@code{kill-line}), will not remove the text,
18921 merely copy it to the kill ring. However, your machine may beep at
18922 you. Alternatively, for silence, you may copy the region of each line
18923 with the @kbd{M-w} (@code{kill-ring-save}) command. You must mark
18924 each line for this command to succeed, but it does not matter at which
18925 end you put point or mark.)
18929 Please invoke the calls in order, so that five elements attempt to
18930 fill the kill ring:
18935 second piece of text
18937 fourth line of text
18944 Then find the value of @code{kill-ring} by evaluating
18956 ("fifth bit of text" "fourth line of text"
18957 "third line" "second piece of text")
18962 The first element, @samp{first some text}, was dropped.
18965 To return to the old value for the length of the kill ring, evaluate:
18968 (setq kill-ring-max old-kill-ring-max)
18972 @appendixsec The @code{current-kill} Function
18973 @findex current-kill
18975 The @code{current-kill} function changes the element in the kill ring
18976 to which @code{kill-ring-yank-pointer} points. (Also, the
18977 @code{kill-new} function sets @code{kill-ring-yank-pointer} to point
18978 to the latest element of the kill ring. The @code{kill-new}
18979 function is used directly or indirectly by @code{kill-append},
18980 @code{copy-region-as-kill}, @code{kill-ring-save}, @code{kill-line},
18981 and @code{kill-region}.)
18984 * Code for current-kill::
18985 * Understanding current-kill::
18989 @node Code for current-kill
18990 @unnumberedsubsec The code for @code{current-kill}
18995 The @code{current-kill} function is used by @code{yank} and by
18996 @code{yank-pop}. Here is the code for @code{current-kill}:
19000 (defun current-kill (n &optional do-not-move)
19001 "Rotate the yanking point by N places, and then return that kill.
19002 If N is zero, `interprogram-paste-function' is set, and calling it
19003 returns a string, then that string is added to the front of the
19004 kill ring and returned as the latest kill.
19007 If optional arg DO-NOT-MOVE is non-nil, then don't actually move the
19008 yanking point; just return the Nth kill forward."
19009 (let ((interprogram-paste (and (= n 0)
19010 interprogram-paste-function
19011 (funcall interprogram-paste-function))))
19014 (if interprogram-paste
19016 ;; Disable the interprogram cut function when we add the new
19017 ;; text to the kill ring, so Emacs doesn't try to own the
19018 ;; selection, with identical text.
19019 (let ((interprogram-cut-function nil))
19020 (kill-new interprogram-paste))
19021 interprogram-paste)
19024 (or kill-ring (error "Kill ring is empty"))
19025 (let ((ARGth-kill-element
19026 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19027 (length kill-ring))
19030 (setq kill-ring-yank-pointer ARGth-kill-element))
19031 (car ARGth-kill-element)))))
19035 Remember also that the @code{kill-new} function sets
19036 @code{kill-ring-yank-pointer} to the latest element of the kill
19037 ring, which means that all the functions that call it set the value
19038 indirectly: @code{kill-append}, @code{copy-region-as-kill},
19039 @code{kill-ring-save}, @code{kill-line}, and @code{kill-region}.
19042 Here is the line in @code{kill-new}, which is explained in
19043 @ref{kill-new function, , The @code{kill-new} function}.
19046 (setq kill-ring-yank-pointer kill-ring)
19050 @node Understanding current-kill
19051 @unnumberedsubsec @code{current-kill} in Outline
19054 The @code{current-kill} function looks complex, but as usual, it can
19055 be understood by taking it apart piece by piece. First look at it in
19060 (defun current-kill (n &optional do-not-move)
19061 "Rotate the yanking point by N places, and then return that kill."
19067 This function takes two arguments, one of which is optional. It has a
19068 documentation string. It is @emph{not} interactive.
19071 * Body of current-kill::
19072 * Digression concerning error:: How to mislead humans, but not computers.
19073 * Determining the Element::
19077 @node Body of current-kill
19078 @unnumberedsubsubsec The Body of @code{current-kill}
19081 The body of the function definition is a @code{let} expression, which
19082 itself has a body as well as a @var{varlist}.
19084 The @code{let} expression declares a variable that will be only usable
19085 within the bounds of this function. This variable is called
19086 @code{interprogram-paste} and is for copying to another program. It
19087 is not for copying within this instance of GNU Emacs. Most window
19088 systems provide a facility for interprogram pasting. Sadly, that
19089 facility usually provides only for the last element. Most windowing
19090 systems have not adopted a ring of many possibilities, even though
19091 Emacs has provided it for decades.
19093 The @code{if} expression has two parts, one if there exists
19094 @code{interprogram-paste} and one if not.
19097 Let us consider the ``if not'' or else-part of the @code{current-kill}
19098 function. (The then-part uses the @code{kill-new} function, which
19099 we have already described. @xref{kill-new function, , The
19100 @code{kill-new} function}.)
19104 (or kill-ring (error "Kill ring is empty"))
19105 (let ((ARGth-kill-element
19106 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19107 (length kill-ring))
19110 (setq kill-ring-yank-pointer ARGth-kill-element))
19111 (car ARGth-kill-element))
19116 The code first checks whether the kill ring has content; otherwise it
19120 Note that the @code{or} expression is very similar to testing length
19127 (if (zerop (length kill-ring)) ; @r{if-part}
19128 (error "Kill ring is empty")) ; @r{then-part}
19134 If there is not anything in the kill ring, its length must be zero and
19135 an error message sent to the user: @samp{Kill ring is empty}. The
19136 @code{current-kill} function uses an @code{or} expression which is
19137 simpler. But an @code{if} expression reminds us what goes on.
19139 This @code{if} expression uses the function @code{zerop} which returns
19140 true if the value it is testing is zero. When @code{zerop} tests
19141 true, the then-part of the @code{if} is evaluated. The then-part is a
19142 list starting with the function @code{error}, which is a function that
19143 is similar to the @code{message} function
19144 (@pxref{message, , The @code{message} Function}) in that
19145 it prints a one-line message in the echo area. However, in addition
19146 to printing a message, @code{error} also stops evaluation of the
19147 function within which it is embedded. This means that the rest of the
19148 function will not be evaluated if the length of the kill ring is zero.
19150 Then the @code{current-kill} function selects the element to return.
19151 The selection depends on the number of places that @code{current-kill}
19152 rotates and on where @code{kill-ring-yank-pointer} points.
19154 Next, either the optional @code{do-not-move} argument is true or the
19155 current value of @code{kill-ring-yank-pointer} is set to point to the
19156 list. Finally, another expression returns the first element of the
19157 list even if the @code{do-not-move} argument is true.
19160 @node Digression concerning error
19161 @unnumberedsubsubsec Digression about the word ``error''
19164 In my opinion, it is slightly misleading, at least to humans, to use
19165 the term ``error'' as the name of the @code{error} function. A better
19166 term would be ``cancel''. Strictly speaking, of course, you cannot
19167 point to, much less rotate a pointer to a list that has no length, so
19168 from the point of view of the computer, the word ``error'' is correct.
19169 But a human expects to attempt this sort of thing, if only to find out
19170 whether the kill ring is full or empty. This is an act of
19173 From the human point of view, the act of exploration and discovery is
19174 not necessarily an error, and therefore should not be labeled as one,
19175 even in the bowels of a computer. As it is, the code in Emacs implies
19176 that a human who is acting virtuously, by exploring his or her
19177 environment, is making an error. This is bad. Even though the computer
19178 takes the same steps as it does when there is an ``error'', a term such as
19179 ``cancel'' would have a clearer connotation.
19182 @node Determining the Element
19183 @unnumberedsubsubsec Determining the Element
19186 Among other actions, the else-part of the @code{if} expression sets
19187 the value of @code{kill-ring-yank-pointer} to
19188 @code{ARGth-kill-element} when the kill ring has something in it and
19189 the value of @code{do-not-move} is @code{nil}.
19192 The code looks like this:
19196 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19197 (length kill-ring))
19202 This needs some examination. Unless it is not supposed to move the
19203 pointer, the @code{current-kill} function changes where
19204 @code{kill-ring-yank-pointer} points.
19206 @w{@code{(setq kill-ring-yank-pointer ARGth-kill-element))}}
19207 expression does. Also, clearly, @code{ARGth-kill-element} is being
19208 set to be equal to some @sc{cdr} of the kill ring, using the
19209 @code{nthcdr} function that is described in an earlier section.
19210 (@xref{copy-region-as-kill}.) How does it do this?
19212 As we have seen before (@pxref{nthcdr}), the @code{nthcdr} function
19213 works by repeatedly taking the @sc{cdr} of a list---it takes the
19214 @sc{cdr} of the @sc{cdr} of the @sc{cdr} @dots{}
19217 The two following expressions produce the same result:
19221 (setq kill-ring-yank-pointer (cdr kill-ring))
19223 (setq kill-ring-yank-pointer (nthcdr 1 kill-ring))
19227 However, the @code{nthcdr} expression is more complicated. It uses
19228 the @code{mod} function to determine which @sc{cdr} to select.
19230 (You will remember to look at inner functions first; indeed, we will
19231 have to go inside the @code{mod}.)
19233 The @code{mod} function returns the value of its first argument modulo
19234 the second; that is to say, it returns the remainder after dividing
19235 the first argument by the second. The value returned has the same
19236 sign as the second argument.
19244 @result{} 0 ;; @r{because there is no remainder}
19251 In this case, the first argument is often smaller than the second.
19263 We can guess what the @code{-} function does. It is like @code{+} but
19264 subtracts instead of adds; the @code{-} function subtracts its second
19265 argument from its first. Also, we already know what the @code{length}
19266 function does (@pxref{length}). It returns the length of a list.
19268 And @code{n} is the name of the required argument to the
19269 @code{current-kill} function.
19272 So when the first argument to @code{nthcdr} is zero, the @code{nthcdr}
19273 expression returns the whole list, as you can see by evaluating the
19278 ;; kill-ring-yank-pointer @r{and} kill-ring @r{have a length of four}
19279 ;; @r{and} (mod (- 0 4) 4) @result{} 0
19280 (nthcdr (mod (- 0 4) 4)
19281 '("fourth line of text"
19283 "second piece of text"
19284 "first some text"))
19289 When the first argument to the @code{current-kill} function is one,
19290 the @code{nthcdr} expression returns the list without its first
19295 (nthcdr (mod (- 1 4) 4)
19296 '("fourth line of text"
19298 "second piece of text"
19299 "first some text"))
19303 @cindex @samp{global variable} defined
19304 @cindex @samp{variable, global}, defined
19305 Incidentally, both @code{kill-ring} and @code{kill-ring-yank-pointer}
19306 are @dfn{global variables}. That means that any expression in Emacs
19307 Lisp can access them. They are not like the local variables set by
19308 @code{let} or like the symbols in an argument list.
19309 Local variables can only be accessed
19310 within the @code{let} that defines them or the function that specifies
19311 them in an argument list (and within expressions called by them).
19314 @c texi2dvi fails when the name of the section is within ifnottex ...
19315 (@xref{Prevent confusion, , @code{let} Prevents Confusion}, and
19316 @ref{defun, , The @code{defun} Macro}.)
19320 @appendixsec @code{yank}
19323 After learning about @code{current-kill}, the code for the
19324 @code{yank} function is almost easy.
19326 The @code{yank} function does not use the
19327 @code{kill-ring-yank-pointer} variable directly. It calls
19328 @code{insert-for-yank} which calls @code{current-kill} which sets the
19329 @code{kill-ring-yank-pointer} variable.
19332 The code looks like this:
19337 (defun yank (&optional arg)
19338 "Reinsert (\"paste\") the last stretch of killed text.
19339 More precisely, reinsert the stretch of killed text most recently
19340 killed OR yanked. Put point at end, and set mark at beginning.
19341 With just \\[universal-argument] as argument, same but put point at
19342 beginning (and mark at end). With argument N, reinsert the Nth most
19343 recently killed stretch of killed text.
19345 When this command inserts killed text into the buffer, it honors
19346 `yank-excluded-properties' and `yank-handler' as described in the
19347 doc string for `insert-for-yank-1', which see.
19349 See also the command \\[yank-pop]."
19353 (setq yank-window-start (window-start))
19354 ;; If we don't get all the way thru, make last-command indicate that
19355 ;; for the following command.
19356 (setq this-command t)
19357 (push-mark (point))
19360 (insert-for-yank (current-kill (cond
19365 ;; This is like exchange-point-and-mark,
19366 ;; but doesn't activate the mark.
19367 ;; It is cleaner to avoid activation, even though the command
19368 ;; loop would deactivate the mark because we inserted text.
19369 (goto-char (prog1 (mark t)
19370 (set-marker (mark-marker) (point) (current-buffer)))))
19373 ;; If we do get all the way thru, make this-command indicate that.
19374 (if (eq this-command t)
19375 (setq this-command 'yank))
19380 The key expression is @code{insert-for-yank}, which inserts the string
19381 returned by @code{current-kill}, but removes some text properties from
19384 However, before getting to that expression, the function sets the value
19385 of @code{yank-window-start} to the position returned by the
19386 @code{(window-start)} expression, the position at which the display
19387 currently starts. The @code{yank} function also sets
19388 @code{this-command} and pushes the mark.
19390 After it yanks the appropriate element, if the optional argument is a
19391 @sc{cons} rather than a number or nothing, it puts point at beginning
19392 of the yanked text and mark at its end.
19394 (The @code{prog1} function is like @code{progn} but returns the value
19395 of its first argument rather than the value of its last argument. Its
19396 first argument is forced to return the buffer's mark as an integer.
19397 You can see the documentation for these functions by placing point
19398 over them in this buffer and then typing @kbd{C-h f}
19399 (@code{describe-function}) followed by a @kbd{RET}; the default is the
19402 The last part of the function tells what to do when it succeeds.
19405 @appendixsec @code{yank-pop}
19408 After understanding @code{yank} and @code{current-kill}, you know how
19409 to approach the @code{yank-pop} function. Leaving out the
19410 documentation to save space, it looks like this:
19415 (defun yank-pop (&optional arg)
19418 (if (not (eq last-command 'yank))
19419 (error "Previous command was not a yank"))
19422 (setq this-command 'yank)
19423 (unless arg (setq arg 1))
19424 (let ((inhibit-read-only t)
19425 (before (< (point) (mark t))))
19429 (funcall (or yank-undo-function 'delete-region) (point) (mark t))
19430 (funcall (or yank-undo-function 'delete-region) (mark t) (point)))
19431 (setq yank-undo-function nil)
19434 (set-marker (mark-marker) (point) (current-buffer))
19435 (insert-for-yank (current-kill arg))
19436 ;; Set the window start back where it was in the yank command,
19438 (set-window-start (selected-window) yank-window-start t)
19442 ;; This is like exchange-point-and-mark,
19443 ;; but doesn't activate the mark.
19444 ;; It is cleaner to avoid activation, even though the command
19445 ;; loop would deactivate the mark because we inserted text.
19446 (goto-char (prog1 (mark t)
19447 (set-marker (mark-marker)
19449 (current-buffer))))))
19454 The function is interactive with a small @samp{p} so the prefix
19455 argument is processed and passed to the function. The command can
19456 only be used after a previous yank; otherwise an error message is
19457 sent. This check uses the variable @code{last-command} which is set
19458 by @code{yank} and is discussed elsewhere.
19459 (@xref{copy-region-as-kill}.)
19461 The @code{let} clause sets the variable @code{before} to true or false
19462 depending whether point is before or after mark and then the region
19463 between point and mark is deleted. This is the region that was just
19464 inserted by the previous yank and it is this text that will be
19467 @code{funcall} calls its first argument as a function, passing
19468 remaining arguments to it. The first argument is whatever the
19469 @code{or} expression returns. The two remaining arguments are the
19470 positions of point and mark set by the preceding @code{yank} command.
19472 There is more, but that is the hardest part.
19475 @appendixsec The @file{ring.el} File
19476 @cindex @file{ring.el} file
19478 Interestingly, GNU Emacs posses a file called @file{ring.el} that
19479 provides many of the features we just discussed. But functions such
19480 as @code{kill-ring-yank-pointer} do not use this library, possibly
19481 because they were written earlier.
19484 @appendix A Graph with Labeled Axes
19486 Printed axes help you understand a graph. They convey scale. In an
19487 earlier chapter (@pxref{Readying a Graph, , Readying a Graph}), we
19488 wrote the code to print the body of a graph. Here we write the code
19489 for printing and labeling vertical and horizontal axes, along with the
19493 * Labeled Example::
19494 * print-graph Varlist:: @code{let} expression in @code{print-graph}.
19495 * print-Y-axis:: Print a label for the vertical axis.
19496 * print-X-axis:: Print a horizontal label.
19497 * Print Whole Graph:: The function to print a complete graph.
19501 @node Labeled Example
19502 @unnumberedsec Labeled Example Graph
19505 Since insertions fill a buffer to the right and below point, the new
19506 graph printing function should first print the Y or vertical axis,
19507 then the body of the graph, and finally the X or horizontal axis.
19508 This sequence lays out for us the contents of the function:
19518 Print body of graph.
19525 Here is an example of how a finished graph should look:
19538 1 - ****************
19545 In this graph, both the vertical and the horizontal axes are labeled
19546 with numbers. However, in some graphs, the horizontal axis is time
19547 and would be better labeled with months, like this:
19561 Indeed, with a little thought, we can easily come up with a variety of
19562 vertical and horizontal labeling schemes. Our task could become
19563 complicated. But complications breed confusion. Rather than permit
19564 this, it is better choose a simple labeling scheme for our first
19565 effort, and to modify or replace it later.
19568 These considerations suggest the following outline for the
19569 @code{print-graph} function:
19573 (defun print-graph (numbers-list)
19574 "@var{documentation}@dots{}"
19575 (let ((height @dots{}
19579 (print-Y-axis height @dots{} )
19580 (graph-body-print numbers-list)
19581 (print-X-axis @dots{} )))
19585 We can work on each part of the @code{print-graph} function definition
19588 @node print-graph Varlist
19589 @appendixsec The @code{print-graph} Varlist
19590 @cindex @code{print-graph} varlist
19592 In writing the @code{print-graph} function, the first task is to write
19593 the varlist in the @code{let} expression. (We will leave aside for the
19594 moment any thoughts about making the function interactive or about the
19595 contents of its documentation string.)
19597 The varlist should set several values. Clearly, the top of the label
19598 for the vertical axis must be at least the height of the graph, which
19599 means that we must obtain this information here. Note that the
19600 @code{print-graph-body} function also requires this information. There
19601 is no reason to calculate the height of the graph in two different
19602 places, so we should change @code{print-graph-body} from the way we
19603 defined it earlier to take advantage of the calculation.
19605 Similarly, both the function for printing the X axis labels and the
19606 @code{print-graph-body} function need to learn the value of the width of
19607 each symbol. We can perform the calculation here and change the
19608 definition for @code{print-graph-body} from the way we defined it in the
19611 The length of the label for the horizontal axis must be at least as long
19612 as the graph. However, this information is used only in the function
19613 that prints the horizontal axis, so it does not need to be calculated here.
19615 These thoughts lead us directly to the following form for the varlist
19616 in the @code{let} for @code{print-graph}:
19620 (let ((height (apply 'max numbers-list)) ; @r{First version.}
19621 (symbol-width (length graph-blank)))
19626 As we shall see, this expression is not quite right.
19630 @appendixsec The @code{print-Y-axis} Function
19631 @cindex Axis, print vertical
19632 @cindex Y axis printing
19633 @cindex Vertical axis printing
19634 @cindex Print vertical axis
19636 The job of the @code{print-Y-axis} function is to print a label for
19637 the vertical axis that looks like this:
19655 The function should be passed the height of the graph, and then should
19656 construct and insert the appropriate numbers and marks.
19659 * print-Y-axis in Detail::
19660 * Height of label:: What height for the Y axis?
19661 * Compute a Remainder:: How to compute the remainder of a division.
19662 * Y Axis Element:: Construct a line for the Y axis.
19663 * Y-axis-column:: Generate a list of Y axis labels.
19664 * print-Y-axis Penultimate:: A not quite final version.
19668 @node print-Y-axis in Detail
19669 @unnumberedsubsec The @code{print-Y-axis} Function in Detail
19672 It is easy enough to see in the figure what the Y axis label should
19673 look like; but to say in words, and then to write a function
19674 definition to do the job is another matter. It is not quite true to
19675 say that we want a number and a tic every five lines: there are only
19676 three lines between the @samp{1} and the @samp{5} (lines 2, 3, and 4),
19677 but four lines between the @samp{5} and the @samp{10} (lines 6, 7, 8,
19678 and 9). It is better to say that we want a number and a tic mark on
19679 the base line (number 1) and then that we want a number and a tic on
19680 the fifth line from the bottom and on every line that is a multiple of
19684 @node Height of label
19685 @unnumberedsubsec What height should the label be?
19688 The next issue is what height the label should be? Suppose the maximum
19689 height of tallest column of the graph is seven. Should the highest
19690 label on the Y axis be @samp{5 -}, and should the graph stick up above
19691 the label? Or should the highest label be @samp{7 -}, and mark the peak
19692 of the graph? Or should the highest label be @code{10 -}, which is a
19693 multiple of five, and be higher than the topmost value of the graph?
19695 The latter form is preferred. Most graphs are drawn within rectangles
19696 whose sides are an integral number of steps long---5, 10, 15, and so
19697 on for a step distance of five. But as soon as we decide to use a
19698 step height for the vertical axis, we discover that the simple
19699 expression in the varlist for computing the height is wrong. The
19700 expression is @code{(apply 'max numbers-list)}. This returns the
19701 precise height, not the maximum height plus whatever is necessary to
19702 round up to the nearest multiple of five. A more complex expression
19705 As usual in cases like this, a complex problem becomes simpler if it is
19706 divided into several smaller problems.
19708 First, consider the case when the highest value of the graph is an
19709 integral multiple of five---when it is 5, 10, 15, or some higher
19710 multiple of five. We can use this value as the Y axis height.
19712 A fairly simply way to determine whether a number is a multiple of
19713 five is to divide it by five and see if the division results in a
19714 remainder. If there is no remainder, the number is a multiple of
19715 five. Thus, seven divided by five has a remainder of two, and seven
19716 is not an integral multiple of five. Put in slightly different
19717 language, more reminiscent of the classroom, five goes into seven
19718 once, with a remainder of two. However, five goes into ten twice,
19719 with no remainder: ten is an integral multiple of five.
19721 @node Compute a Remainder
19722 @appendixsubsec Side Trip: Compute a Remainder
19724 @findex % @r{(remainder function)}
19725 @cindex Remainder function, @code{%}
19726 In Lisp, the function for computing a remainder is @code{%}. The
19727 function returns the remainder of its first argument divided by its
19728 second argument. As it happens, @code{%} is a function in Emacs Lisp
19729 that you cannot discover using @code{apropos}: you find nothing if you
19730 type @kbd{M-x apropos @key{RET} remainder @key{RET}}. The only way to
19731 learn of the existence of @code{%} is to read about it in a book such
19732 as this or in the Emacs Lisp sources.
19734 You can try the @code{%} function by evaluating the following two
19746 The first expression returns 2 and the second expression returns 0.
19748 To test whether the returned value is zero or some other number, we
19749 can use the @code{zerop} function. This function returns @code{t} if
19750 its argument, which must be a number, is zero.
19762 Thus, the following expression will return @code{t} if the height
19763 of the graph is evenly divisible by five:
19766 (zerop (% height 5))
19770 (The value of @code{height}, of course, can be found from @code{(apply
19771 'max numbers-list)}.)
19773 On the other hand, if the value of @code{height} is not a multiple of
19774 five, we want to reset the value to the next higher multiple of five.
19775 This is straightforward arithmetic using functions with which we are
19776 already familiar. First, we divide the value of @code{height} by five
19777 to determine how many times five goes into the number. Thus, five
19778 goes into twelve twice. If we add one to this quotient and multiply by
19779 five, we will obtain the value of the next multiple of five that is
19780 larger than the height. Five goes into twelve twice. Add one to two,
19781 and multiply by five; the result is fifteen, which is the next multiple
19782 of five that is higher than twelve. The Lisp expression for this is:
19785 (* (1+ (/ height 5)) 5)
19789 For example, if you evaluate the following, the result is 15:
19792 (* (1+ (/ 12 5)) 5)
19795 All through this discussion, we have been using ``five'' as the value
19796 for spacing labels on the Y axis; but we may want to use some other
19797 value. For generality, we should replace ``five'' with a variable to
19798 which we can assign a value. The best name I can think of for this
19799 variable is @code{Y-axis-label-spacing}.
19802 Using this term, and an @code{if} expression, we produce the
19807 (if (zerop (% height Y-axis-label-spacing))
19810 (* (1+ (/ height Y-axis-label-spacing))
19811 Y-axis-label-spacing))
19816 This expression returns the value of @code{height} itself if the height
19817 is an even multiple of the value of the @code{Y-axis-label-spacing} or
19818 else it computes and returns a value of @code{height} that is equal to
19819 the next higher multiple of the value of the @code{Y-axis-label-spacing}.
19821 We can now include this expression in the @code{let} expression of the
19822 @code{print-graph} function (after first setting the value of
19823 @code{Y-axis-label-spacing}):
19824 @vindex Y-axis-label-spacing
19828 (defvar Y-axis-label-spacing 5
19829 "Number of lines from one Y axis label to next.")
19834 (let* ((height (apply 'max numbers-list))
19835 (height-of-top-line
19836 (if (zerop (% height Y-axis-label-spacing))
19841 (* (1+ (/ height Y-axis-label-spacing))
19842 Y-axis-label-spacing)))
19843 (symbol-width (length graph-blank))))
19849 (Note use of the @code{let*} function: the initial value of height is
19850 computed once by the @code{(apply 'max numbers-list)} expression and
19851 then the resulting value of @code{height} is used to compute its
19852 final value. @xref{fwd-para let, , The @code{let*} expression}, for
19853 more about @code{let*}.)
19855 @node Y Axis Element
19856 @appendixsubsec Construct a Y Axis Element
19858 When we print the vertical axis, we want to insert strings such as
19859 @w{@samp{5 -}} and @w{@samp{10 - }} every five lines.
19860 Moreover, we want the numbers and dashes to line up, so shorter
19861 numbers must be padded with leading spaces. If some of the strings
19862 use two digit numbers, the strings with single digit numbers must
19863 include a leading blank space before the number.
19865 @findex number-to-string
19866 To figure out the length of the number, the @code{length} function is
19867 used. But the @code{length} function works only with a string, not with
19868 a number. So the number has to be converted from being a number to
19869 being a string. This is done with the @code{number-to-string} function.
19874 (length (number-to-string 35))
19877 (length (number-to-string 100))
19883 (@code{number-to-string} is also called @code{int-to-string}; you will
19884 see this alternative name in various sources.)
19886 In addition, in each label, each number is followed by a string such
19887 as @w{@samp{ - }}, which we will call the @code{Y-axis-tic} marker.
19888 This variable is defined with @code{defvar}:
19893 (defvar Y-axis-tic " - "
19894 "String that follows number in a Y axis label.")
19898 The length of the Y label is the sum of the length of the Y axis tic
19899 mark and the length of the number of the top of the graph.
19902 (length (concat (number-to-string height) Y-axis-tic)))
19905 This value will be calculated by the @code{print-graph} function in
19906 its varlist as @code{full-Y-label-width} and passed on. (Note that we
19907 did not think to include this in the varlist when we first proposed it.)
19909 To make a complete vertical axis label, a tic mark is concatenated
19910 with a number; and the two together may be preceded by one or more
19911 spaces depending on how long the number is. The label consists of
19912 three parts: the (optional) leading spaces, the number, and the tic
19913 mark. The function is passed the value of the number for the specific
19914 row, and the value of the width of the top line, which is calculated
19915 (just once) by @code{print-graph}.
19919 (defun Y-axis-element (number full-Y-label-width)
19920 "Construct a NUMBERed label element.
19921 A numbered element looks like this ' 5 - ',
19922 and is padded as needed so all line up with
19923 the element for the largest number."
19926 (let* ((leading-spaces
19927 (- full-Y-label-width
19929 (concat (number-to-string number)
19934 (make-string leading-spaces ? )
19935 (number-to-string number)
19940 The @code{Y-axis-element} function concatenates together the leading
19941 spaces, if any; the number, as a string; and the tic mark.
19943 To figure out how many leading spaces the label will need, the
19944 function subtracts the actual length of the label---the length of the
19945 number plus the length of the tic mark---from the desired label width.
19947 @findex make-string
19948 Blank spaces are inserted using the @code{make-string} function. This
19949 function takes two arguments: the first tells it how long the string
19950 will be and the second is a symbol for the character to insert, in a
19951 special format. The format is a question mark followed by a blank
19952 space, like this, @samp{? }. @xref{Character Type, , Character Type,
19953 elisp, The GNU Emacs Lisp Reference Manual}, for a description of the
19954 syntax for characters. (Of course, you might want to replace the
19955 blank space by some other character @dots{} You know what to do.)
19957 The @code{number-to-string} function is used in the concatenation
19958 expression, to convert the number to a string that is concatenated
19959 with the leading spaces and the tic mark.
19961 @node Y-axis-column
19962 @appendixsubsec Create a Y Axis Column
19964 The preceding functions provide all the tools needed to construct a
19965 function that generates a list of numbered and blank strings to insert
19966 as the label for the vertical axis:
19968 @findex Y-axis-column
19971 (defun Y-axis-column (height width-of-label)
19972 "Construct list of Y axis labels and blank strings.
19973 For HEIGHT of line above base and WIDTH-OF-LABEL."
19977 (while (> height 1)
19978 (if (zerop (% height Y-axis-label-spacing))
19979 ;; @r{Insert label.}
19982 (Y-axis-element height width-of-label)
19986 ;; @r{Else, insert blanks.}
19989 (make-string width-of-label ? )
19991 (setq height (1- height)))
19992 ;; @r{Insert base line.}
19994 (cons (Y-axis-element 1 width-of-label) Y-axis))
19995 (nreverse Y-axis)))
19999 In this function, we start with the value of @code{height} and
20000 repetitively subtract one from its value. After each subtraction, we
20001 test to see whether the value is an integral multiple of the
20002 @code{Y-axis-label-spacing}. If it is, we construct a numbered label
20003 using the @code{Y-axis-element} function; if not, we construct a
20004 blank label using the @code{make-string} function. The base line
20005 consists of the number one followed by a tic mark.
20008 @node print-Y-axis Penultimate
20009 @appendixsubsec The Not Quite Final Version of @code{print-Y-axis}
20011 The list constructed by the @code{Y-axis-column} function is passed to
20012 the @code{print-Y-axis} function, which inserts the list as a column.
20014 @findex print-Y-axis
20017 (defun print-Y-axis (height full-Y-label-width)
20018 "Insert Y axis using HEIGHT and FULL-Y-LABEL-WIDTH.
20019 Height must be the maximum height of the graph.
20020 Full width is the width of the highest label element."
20021 ;; Value of height and full-Y-label-width
20022 ;; are passed by 'print-graph'.
20025 (let ((start (point)))
20027 (Y-axis-column height full-Y-label-width))
20028 ;; @r{Place point ready for inserting graph.}
20030 ;; @r{Move point forward by value of} full-Y-label-width
20031 (forward-char full-Y-label-width)))
20035 The @code{print-Y-axis} uses the @code{insert-rectangle} function to
20036 insert the Y axis labels created by the @code{Y-axis-column} function.
20037 In addition, it places point at the correct position for printing the body of
20040 You can test @code{print-Y-axis}:
20048 Y-axis-label-spacing
20057 Copy the following expression:
20060 (print-Y-axis 12 5)
20064 Switch to the @file{*scratch*} buffer and place the cursor where you
20065 want the axis labels to start.
20068 Type @kbd{M-:} (@code{eval-expression}).
20071 Yank the @code{graph-body-print} expression into the minibuffer
20072 with @kbd{C-y} (@code{yank)}.
20075 Press @key{RET} to evaluate the expression.
20078 Emacs will print labels vertically, the top one being @w{@samp{10 -@w{
20079 }}}. (The @code{print-graph} function will pass the value of
20080 @code{height-of-top-line}, which in this case will end up as 15,
20081 thereby getting rid of what might appear as a bug.)
20085 @appendixsec The @code{print-X-axis} Function
20086 @cindex Axis, print horizontal
20087 @cindex X axis printing
20088 @cindex Print horizontal axis
20089 @cindex Horizontal axis printing
20091 X axis labels are much like Y axis labels, except that the ticks are on a
20092 line above the numbers. Labels should look like this:
20101 The first tic is under the first column of the graph and is preceded by
20102 several blank spaces. These spaces provide room in rows above for the Y
20103 axis labels. The second, third, fourth, and subsequent ticks are all
20104 spaced equally, according to the value of @code{X-axis-label-spacing}.
20106 The second row of the X axis consists of numbers, preceded by several
20107 blank spaces and also separated according to the value of the variable
20108 @code{X-axis-label-spacing}.
20110 The value of the variable @code{X-axis-label-spacing} should itself be
20111 measured in units of @code{symbol-width}, since you may want to change
20112 the width of the symbols that you are using to print the body of the
20113 graph without changing the ways the graph is labeled.
20116 * Similarities differences:: Much like @code{print-Y-axis}, but not exactly.
20117 * X Axis Tic Marks:: Create tic marks for the horizontal axis.
20121 @node Similarities differences
20122 @unnumberedsubsec Similarities and differences
20125 The @code{print-X-axis} function is constructed in more or less the
20126 same fashion as the @code{print-Y-axis} function except that it has
20127 two lines: the line of tic marks and the numbers. We will write a
20128 separate function to print each line and then combine them within the
20129 @code{print-X-axis} function.
20131 This is a three step process:
20135 Write a function to print the X axis tic marks, @code{print-X-axis-tic-line}.
20138 Write a function to print the X numbers, @code{print-X-axis-numbered-line}.
20141 Write a function to print both lines, the @code{print-X-axis} function,
20142 using @code{print-X-axis-tic-line} and
20143 @code{print-X-axis-numbered-line}.
20146 @node X Axis Tic Marks
20147 @appendixsubsec X Axis Tic Marks
20149 The first function should print the X axis tic marks. We must specify
20150 the tic marks themselves and their spacing:
20154 (defvar X-axis-label-spacing
20155 (if (boundp 'graph-blank)
20156 (* 5 (length graph-blank)) 5)
20157 "Number of units from one X axis label to next.")
20162 (Note that the value of @code{graph-blank} is set by another
20163 @code{defvar}. The @code{boundp} predicate checks whether it has
20164 already been set; @code{boundp} returns @code{nil} if it has not. If
20165 @code{graph-blank} were unbound and we did not use this conditional
20166 construction, in a recent GNU Emacs, we would enter the debugger and
20167 see an error message saying @samp{@w{Debugger entered--Lisp error:}
20168 @w{(void-variable graph-blank)}}.)
20171 Here is the @code{defvar} for @code{X-axis-tic-symbol}:
20175 (defvar X-axis-tic-symbol "|"
20176 "String to insert to point to a column in X axis.")
20181 The goal is to make a line that looks like this:
20187 The first tic is indented so that it is under the first column, which is
20188 indented to provide space for the Y axis labels.
20190 A tic element consists of the blank spaces that stretch from one tic to
20191 the next plus a tic symbol. The number of blanks is determined by the
20192 width of the tic symbol and the @code{X-axis-label-spacing}.
20195 The code looks like this:
20199 ;;; X-axis-tic-element
20203 ;; @r{Make a string of blanks.}
20204 (- (* symbol-width X-axis-label-spacing)
20205 (length X-axis-tic-symbol))
20207 ;; @r{Concatenate blanks with tic symbol.}
20213 Next, we determine how many blanks are needed to indent the first tic
20214 mark to the first column of the graph. This uses the value of
20215 @code{full-Y-label-width} passed it by the @code{print-graph} function.
20218 The code to make @code{X-axis-leading-spaces}
20223 ;; X-axis-leading-spaces
20225 (make-string full-Y-label-width ? )
20230 We also need to determine the length of the horizontal axis, which is
20231 the length of the numbers list, and the number of ticks in the horizontal
20238 (length numbers-list)
20244 (* symbol-width X-axis-label-spacing)
20248 ;; number-of-X-ticks
20249 (if (zerop (% (X-length tic-width)))
20250 (/ (X-length tic-width))
20251 (1+ (/ (X-length tic-width))))
20256 All this leads us directly to the function for printing the X axis tic line:
20258 @findex print-X-axis-tic-line
20261 (defun print-X-axis-tic-line
20262 (number-of-X-tics X-axis-leading-spaces X-axis-tic-element)
20263 "Print ticks for X axis."
20264 (insert X-axis-leading-spaces)
20265 (insert X-axis-tic-symbol) ; @r{Under first column.}
20268 ;; @r{Insert second tic in the right spot.}
20271 (- (* symbol-width X-axis-label-spacing)
20272 ;; @r{Insert white space up to second tic symbol.}
20273 (* 2 (length X-axis-tic-symbol)))
20275 X-axis-tic-symbol))
20278 ;; @r{Insert remaining ticks.}
20279 (while (> number-of-X-tics 1)
20280 (insert X-axis-tic-element)
20281 (setq number-of-X-tics (1- number-of-X-tics))))
20285 The line of numbers is equally straightforward:
20288 First, we create a numbered element with blank spaces before each number:
20290 @findex X-axis-element
20293 (defun X-axis-element (number)
20294 "Construct a numbered X axis element."
20295 (let ((leading-spaces
20296 (- (* symbol-width X-axis-label-spacing)
20297 (length (number-to-string number)))))
20298 (concat (make-string leading-spaces ? )
20299 (number-to-string number))))
20303 Next, we create the function to print the numbered line, starting with
20304 the number ``1'' under the first column:
20306 @findex print-X-axis-numbered-line
20309 (defun print-X-axis-numbered-line
20310 (number-of-X-tics X-axis-leading-spaces)
20311 "Print line of X-axis numbers"
20312 (let ((number X-axis-label-spacing))
20313 (insert X-axis-leading-spaces)
20319 ;; @r{Insert white space up to next number.}
20320 (- (* symbol-width X-axis-label-spacing) 2)
20322 (number-to-string number)))
20325 ;; @r{Insert remaining numbers.}
20326 (setq number (+ number X-axis-label-spacing))
20327 (while (> number-of-X-tics 1)
20328 (insert (X-axis-element number))
20329 (setq number (+ number X-axis-label-spacing))
20330 (setq number-of-X-tics (1- number-of-X-tics)))))
20334 Finally, we need to write the @code{print-X-axis} that uses
20335 @code{print-X-axis-tic-line} and
20336 @code{print-X-axis-numbered-line}.
20338 The function must determine the local values of the variables used by both
20339 @code{print-X-axis-tic-line} and @code{print-X-axis-numbered-line}, and
20340 then it must call them. Also, it must print the carriage return that
20341 separates the two lines.
20343 The function consists of a varlist that specifies five local variables,
20344 and calls to each of the two line printing functions:
20346 @findex print-X-axis
20349 (defun print-X-axis (numbers-list)
20350 "Print X axis labels to length of NUMBERS-LIST."
20351 (let* ((leading-spaces
20352 (make-string full-Y-label-width ? ))
20355 ;; symbol-width @r{is provided by} graph-body-print
20356 (tic-width (* symbol-width X-axis-label-spacing))
20357 (X-length (length numbers-list))
20365 ;; @r{Make a string of blanks.}
20366 (- (* symbol-width X-axis-label-spacing)
20367 (length X-axis-tic-symbol))
20371 ;; @r{Concatenate blanks with tic symbol.}
20372 X-axis-tic-symbol))
20376 (if (zerop (% X-length tic-width))
20377 (/ X-length tic-width)
20378 (1+ (/ X-length tic-width)))))
20381 (print-X-axis-tic-line tic-number leading-spaces X-tic)
20383 (print-X-axis-numbered-line tic-number leading-spaces)))
20388 You can test @code{print-X-axis}:
20392 Install @code{X-axis-tic-symbol}, @code{X-axis-label-spacing},
20393 @code{print-X-axis-tic-line}, as well as @code{X-axis-element},
20394 @code{print-X-axis-numbered-line}, and @code{print-X-axis}.
20397 Copy the following expression:
20402 (let ((full-Y-label-width 5)
20405 '(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16))))
20410 Switch to the @file{*scratch*} buffer and place the cursor where you
20411 want the axis labels to start.
20414 Type @kbd{M-:} (@code{eval-expression}).
20417 Yank the test expression into the minibuffer
20418 with @kbd{C-y} (@code{yank)}.
20421 Press @key{RET} to evaluate the expression.
20425 Emacs will print the horizontal axis like this:
20435 @node Print Whole Graph
20436 @appendixsec Printing the Whole Graph
20437 @cindex Printing the whole graph
20438 @cindex Whole graph printing
20439 @cindex Graph, printing all
20441 Now we are nearly ready to print the whole graph.
20443 The function to print the graph with the proper labels follows the
20444 outline we created earlier (@pxref{Full Graph, , A Graph with Labeled
20445 Axes}), but with additions.
20448 Here is the outline:
20452 (defun print-graph (numbers-list)
20453 "@var{documentation}@dots{}"
20454 (let ((height @dots{}
20458 (print-Y-axis height @dots{} )
20459 (graph-body-print numbers-list)
20460 (print-X-axis @dots{} )))
20465 * The final version:: A few changes.
20466 * Test print-graph:: Run a short test.
20467 * Graphing words in defuns:: Executing the final code.
20468 * lambda:: How to write an anonymous function.
20469 * mapcar:: Apply a function to elements of a list.
20470 * Another Bug:: Yet another bug @dots{} most insidious.
20471 * Final printed graph:: The graph itself!
20475 @node The final version
20476 @unnumberedsubsec Changes for the Final Version
20479 The final version is different from what we planned in two ways:
20480 first, it contains additional values calculated once in the varlist;
20481 second, it carries an option to specify the labels' increment per row.
20482 This latter feature turns out to be essential; otherwise, a graph may
20483 have more rows than fit on a display or on a sheet of paper.
20486 This new feature requires a change to the @code{Y-axis-column}
20487 function, to add @code{vertical-step} to it. The function looks like
20490 @findex Y-axis-column @r{Final version.}
20493 ;;; @r{Final version.}
20494 (defun Y-axis-column
20495 (height width-of-label &optional vertical-step)
20496 "Construct list of labels for Y axis.
20497 HEIGHT is maximum height of graph.
20498 WIDTH-OF-LABEL is maximum width of label.
20499 VERTICAL-STEP, an option, is a positive integer
20500 that specifies how much a Y axis label increments
20501 for each line. For example, a step of 5 means
20502 that each line is five units of the graph."
20506 (number-per-line (or vertical-step 1)))
20507 (while (> height 1)
20508 (if (zerop (% height Y-axis-label-spacing))
20511 ;; @r{Insert label.}
20515 (* height number-per-line)
20520 ;; @r{Else, insert blanks.}
20523 (make-string width-of-label ? )
20525 (setq height (1- height)))
20528 ;; @r{Insert base line.}
20529 (setq Y-axis (cons (Y-axis-element
20530 (or vertical-step 1)
20533 (nreverse Y-axis)))
20537 The values for the maximum height of graph and the width of a symbol
20538 are computed by @code{print-graph} in its @code{let} expression; so
20539 @code{graph-body-print} must be changed to accept them.
20541 @findex graph-body-print @r{Final version.}
20544 ;;; @r{Final version.}
20545 (defun graph-body-print (numbers-list height symbol-width)
20546 "Print a bar graph of the NUMBERS-LIST.
20547 The numbers-list consists of the Y-axis values.
20548 HEIGHT is maximum height of graph.
20549 SYMBOL-WIDTH is number of each column."
20552 (let (from-position)
20553 (while numbers-list
20554 (setq from-position (point))
20556 (column-of-graph height (car numbers-list)))
20557 (goto-char from-position)
20558 (forward-char symbol-width)
20561 ;; @r{Draw graph column by column.}
20563 (setq numbers-list (cdr numbers-list)))
20564 ;; @r{Place point for X axis labels.}
20565 (forward-line height)
20571 Finally, the code for the @code{print-graph} function:
20573 @findex print-graph @r{Final version.}
20576 ;;; @r{Final version.}
20578 (numbers-list &optional vertical-step)
20579 "Print labeled bar graph of the NUMBERS-LIST.
20580 The numbers-list consists of the Y-axis values.
20584 Optionally, VERTICAL-STEP, a positive integer,
20585 specifies how much a Y axis label increments for
20586 each line. For example, a step of 5 means that
20587 each row is five units."
20590 (let* ((symbol-width (length graph-blank))
20591 ;; @code{height} @r{is both the largest number}
20592 ;; @r{and the number with the most digits.}
20593 (height (apply 'max numbers-list))
20596 (height-of-top-line
20597 (if (zerop (% height Y-axis-label-spacing))
20600 (* (1+ (/ height Y-axis-label-spacing))
20601 Y-axis-label-spacing)))
20604 (vertical-step (or vertical-step 1))
20605 (full-Y-label-width
20611 (* height-of-top-line vertical-step))
20617 height-of-top-line full-Y-label-width vertical-step)
20621 numbers-list height-of-top-line symbol-width)
20622 (print-X-axis numbers-list)))
20626 @node Test print-graph
20627 @appendixsubsec Testing @code{print-graph}
20630 We can test the @code{print-graph} function with a short list of numbers:
20634 Install the final versions of @code{Y-axis-column},
20635 @code{graph-body-print}, and @code{print-graph} (in addition to the
20639 Copy the following expression:
20642 (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1))
20646 Switch to the @file{*scratch*} buffer and place the cursor where you
20647 want the axis labels to start.
20650 Type @kbd{M-:} (@code{eval-expression}).
20653 Yank the test expression into the minibuffer
20654 with @kbd{C-y} (@code{yank)}.
20657 Press @key{RET} to evaluate the expression.
20661 Emacs will print a graph that looks like this:
20682 On the other hand, if you pass @code{print-graph} a
20683 @code{vertical-step} value of 2, by evaluating this expression:
20686 (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1) 2)
20691 The graph looks like this:
20712 (A question: is the @samp{2} on the bottom of the vertical axis a bug or a
20713 feature? If you think it is a bug, and should be a @samp{1} instead, (or
20714 even a @samp{0}), you can modify the sources.)
20716 @node Graphing words in defuns
20717 @appendixsubsec Graphing Numbers of Words and Symbols
20719 Now for the graph for which all this code was written: a graph that
20720 shows how many function definitions contain fewer than 10 words and
20721 symbols, how many contain between 10 and 19 words and symbols, how
20722 many contain between 20 and 29 words and symbols, and so on.
20724 This is a multi-step process. First make sure you have loaded all the
20728 It is a good idea to reset the value of @code{top-of-ranges} in case
20729 you have set it to some different value. You can evaluate the
20734 (setq top-of-ranges
20737 110 120 130 140 150
20738 160 170 180 190 200
20739 210 220 230 240 250
20740 260 270 280 290 300)
20745 Next create a list of the number of words and symbols in each range.
20749 Evaluate the following:
20753 (setq list-for-graph
20756 (recursive-lengths-list-many-files
20757 (directory-files "/usr/local/emacs/lisp"
20765 On my old machine, this took about an hour. It looked though 303 Lisp
20766 files in my copy of Emacs version 19.23. After all that computing,
20767 the @code{list-for-graph} had this value:
20771 (537 1027 955 785 594 483 349 292 224 199 166 120 116 99
20772 90 80 67 48 52 45 41 33 28 26 25 20 12 28 11 13 220)
20777 This means that my copy of Emacs had 537 function definitions with
20778 fewer than 10 words or symbols in them, 1,027 function definitions
20779 with 10 to 19 words or symbols in them, 955 function definitions with
20780 20 to 29 words or symbols in them, and so on.
20782 Clearly, just by looking at this list we can see that most function
20783 definitions contain ten to thirty words and symbols.
20785 Now for printing. We do @emph{not} want to print a graph that is
20786 1,030 lines high @dots{} Instead, we should print a graph that is
20787 fewer than twenty-five lines high. A graph that height can be
20788 displayed on almost any monitor, and easily printed on a sheet of paper.
20790 This means that each value in @code{list-for-graph} must be reduced to
20791 one-fiftieth its present value.
20793 Here is a short function to do just that, using two functions we have
20794 not yet seen, @code{mapcar} and @code{lambda}.
20798 (defun one-fiftieth (full-range)
20799 "Return list, each number one-fiftieth of previous."
20800 (mapcar (lambda (arg) (/ arg 50)) full-range))
20805 @appendixsubsec A @code{lambda} Expression: Useful Anonymity
20806 @cindex Anonymous function
20809 @code{lambda} is the symbol for an anonymous function, a function
20810 without a name. Every time you use an anonymous function, you need to
20811 include its whole body.
20818 (lambda (arg) (/ arg 50))
20822 is a function definition that says ``return the value resulting from
20823 dividing whatever is passed to me as @code{arg} by 50''.
20826 Earlier, for example, we had a function @code{multiply-by-seven}; it
20827 multiplied its argument by 7. This function is similar, except it
20828 divides its argument by 50; and, it has no name. The anonymous
20829 equivalent of @code{multiply-by-seven} is:
20832 (lambda (number) (* 7 number))
20836 (@xref{defun, , The @code{defun} Macro}.)
20840 If we want to multiply 3 by 7, we can write:
20842 @c clear print-postscript-figures
20843 @c lambda example diagram #1
20847 (multiply-by-seven 3)
20848 \_______________/ ^
20854 @ifset print-postscript-figures
20857 @center @image{lambda-1}
20861 @ifclear print-postscript-figures
20865 (multiply-by-seven 3)
20866 \_______________/ ^
20875 This expression returns 21.
20879 Similarly, we can write:
20881 @c lambda example diagram #2
20885 ((lambda (number) (* 7 number)) 3)
20886 \____________________________/ ^
20888 anonymous function argument
20892 @ifset print-postscript-figures
20895 @center @image{lambda-2}
20899 @ifclear print-postscript-figures
20903 ((lambda (number) (* 7 number)) 3)
20904 \____________________________/ ^
20906 anonymous function argument
20914 If we want to divide 100 by 50, we can write:
20916 @c lambda example diagram #3
20920 ((lambda (arg) (/ arg 50)) 100)
20921 \______________________/ \_/
20923 anonymous function argument
20927 @ifset print-postscript-figures
20930 @center @image{lambda-3}
20934 @ifclear print-postscript-figures
20938 ((lambda (arg) (/ arg 50)) 100)
20939 \______________________/ \_/
20941 anonymous function argument
20948 This expression returns 2. The 100 is passed to the function, which
20949 divides that number by 50.
20951 @xref{Lambda Expressions, , Lambda Expressions, elisp, The GNU Emacs
20952 Lisp Reference Manual}, for more about @code{lambda}. Lisp and lambda
20953 expressions derive from the Lambda Calculus.
20956 @appendixsubsec The @code{mapcar} Function
20959 @code{mapcar} is a function that calls its first argument with each
20960 element of its second argument, in turn. The second argument must be
20963 The @samp{map} part of the name comes from the mathematical phrase,
20964 ``mapping over a domain'', meaning to apply a function to each of the
20965 elements in a domain. The mathematical phrase is based on the
20966 metaphor of a surveyor walking, one step at a time, over an area he is
20967 mapping. And @samp{car}, of course, comes from the Lisp notion of the
20976 (mapcar '1+ '(2 4 6))
20982 The function @code{1+} which adds one to its argument, is executed on
20983 @emph{each} element of the list, and a new list is returned.
20985 Contrast this with @code{apply}, which applies its first argument to
20987 (@xref{Readying a Graph, , Readying a Graph}, for a explanation of
20991 In the definition of @code{one-fiftieth}, the first argument is the
20992 anonymous function:
20995 (lambda (arg) (/ arg 50))
20999 and the second argument is @code{full-range}, which will be bound to
21000 @code{list-for-graph}.
21003 The whole expression looks like this:
21006 (mapcar (lambda (arg) (/ arg 50)) full-range))
21009 @xref{Mapping Functions, , Mapping Functions, elisp, The GNU Emacs
21010 Lisp Reference Manual}, for more about @code{mapcar}.
21012 Using the @code{one-fiftieth} function, we can generate a list in
21013 which each element is one-fiftieth the size of the corresponding
21014 element in @code{list-for-graph}.
21018 (setq fiftieth-list-for-graph
21019 (one-fiftieth list-for-graph))
21024 The resulting list looks like this:
21028 (10 20 19 15 11 9 6 5 4 3 3 2 2
21029 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 4)
21034 This, we are almost ready to print! (We also notice the loss of
21035 information: many of the higher ranges are 0, meaning that fewer than
21036 50 defuns had that many words or symbols---but not necessarily meaning
21037 that none had that many words or symbols.)
21040 @appendixsubsec Another Bug @dots{} Most Insidious
21041 @cindex Bug, most insidious type
21042 @cindex Insidious type of bug
21044 I said ``almost ready to print''! Of course, there is a bug in the
21045 @code{print-graph} function @dots{} It has a @code{vertical-step}
21046 option, but not a @code{horizontal-step} option. The
21047 @code{top-of-range} scale goes from 10 to 300 by tens. But the
21048 @code{print-graph} function will print only by ones.
21050 This is a classic example of what some consider the most insidious
21051 type of bug, the bug of omission. This is not the kind of bug you can
21052 find by studying the code, for it is not in the code; it is an omitted
21053 feature. Your best actions are to try your program early and often;
21054 and try to arrange, as much as you can, to write code that is easy to
21055 understand and easy to change. Try to be aware, whenever you can,
21056 that whatever you have written, @emph{will} be rewritten, if not soon,
21057 eventually. A hard maxim to follow.
21059 It is the @code{print-X-axis-numbered-line} function that needs the
21060 work; and then the @code{print-X-axis} and the @code{print-graph}
21061 functions need to be adapted. Not much needs to be done; there is one
21062 nicety: the numbers ought to line up under the tic marks. This takes
21066 Here is the corrected @code{print-X-axis-numbered-line}:
21070 (defun print-X-axis-numbered-line
21071 (number-of-X-tics X-axis-leading-spaces
21072 &optional horizontal-step)
21073 "Print line of X-axis numbers"
21074 (let ((number X-axis-label-spacing)
21075 (horizontal-step (or horizontal-step 1)))
21078 (insert X-axis-leading-spaces)
21079 ;; @r{Delete extra leading spaces.}
21082 (length (number-to-string horizontal-step)))))
21087 ;; @r{Insert white space.}
21089 X-axis-label-spacing)
21092 (number-to-string horizontal-step)))
21096 (* number horizontal-step))))
21099 ;; @r{Insert remaining numbers.}
21100 (setq number (+ number X-axis-label-spacing))
21101 (while (> number-of-X-tics 1)
21102 (insert (X-axis-element
21103 (* number horizontal-step)))
21104 (setq number (+ number X-axis-label-spacing))
21105 (setq number-of-X-tics (1- number-of-X-tics)))))
21110 If you are reading this in Info, you can see the new versions of
21111 @code{print-X-axis} @code{print-graph} and evaluate them. If you are
21112 reading this in a printed book, you can see the changed lines here
21113 (the full text is too much to print).
21118 (defun print-X-axis (numbers-list horizontal-step)
21120 (print-X-axis-numbered-line
21121 tic-number leading-spaces horizontal-step))
21129 &optional vertical-step horizontal-step)
21131 (print-X-axis numbers-list horizontal-step))
21139 (defun print-X-axis (numbers-list horizontal-step)
21140 "Print X axis labels to length of NUMBERS-LIST.
21141 Optionally, HORIZONTAL-STEP, a positive integer,
21142 specifies how much an X axis label increments for
21146 ;; Value of symbol-width and full-Y-label-width
21147 ;; are passed by 'print-graph'.
21148 (let* ((leading-spaces
21149 (make-string full-Y-label-width ? ))
21150 ;; symbol-width @r{is provided by} graph-body-print
21151 (tic-width (* symbol-width X-axis-label-spacing))
21152 (X-length (length numbers-list))
21158 ;; @r{Make a string of blanks.}
21159 (- (* symbol-width X-axis-label-spacing)
21160 (length X-axis-tic-symbol))
21164 ;; @r{Concatenate blanks with tic symbol.}
21165 X-axis-tic-symbol))
21167 (if (zerop (% X-length tic-width))
21168 (/ X-length tic-width)
21169 (1+ (/ X-length tic-width)))))
21173 (print-X-axis-tic-line
21174 tic-number leading-spaces X-tic)
21176 (print-X-axis-numbered-line
21177 tic-number leading-spaces horizontal-step)))
21184 (numbers-list &optional vertical-step horizontal-step)
21185 "Print labeled bar graph of the NUMBERS-LIST.
21186 The numbers-list consists of the Y-axis values.
21190 Optionally, VERTICAL-STEP, a positive integer,
21191 specifies how much a Y axis label increments for
21192 each line. For example, a step of 5 means that
21193 each row is five units.
21197 Optionally, HORIZONTAL-STEP, a positive integer,
21198 specifies how much an X axis label increments for
21200 (let* ((symbol-width (length graph-blank))
21201 ;; @code{height} @r{is both the largest number}
21202 ;; @r{and the number with the most digits.}
21203 (height (apply 'max numbers-list))
21206 (height-of-top-line
21207 (if (zerop (% height Y-axis-label-spacing))
21210 (* (1+ (/ height Y-axis-label-spacing))
21211 Y-axis-label-spacing)))
21214 (vertical-step (or vertical-step 1))
21215 (full-Y-label-width
21219 (* height-of-top-line vertical-step))
21224 height-of-top-line full-Y-label-width vertical-step)
21226 numbers-list height-of-top-line symbol-width)
21227 (print-X-axis numbers-list horizontal-step)))
21234 Graphing Definitions Re-listed
21237 Here are all the graphing definitions in their final form:
21241 (defvar top-of-ranges
21244 110 120 130 140 150
21245 160 170 180 190 200
21246 210 220 230 240 250)
21247 "List specifying ranges for `defuns-per-range'.")
21251 (defvar graph-symbol "*"
21252 "String used as symbol in graph, usually an asterisk.")
21256 (defvar graph-blank " "
21257 "String used as blank in graph, usually a blank space.
21258 graph-blank must be the same number of columns wide
21263 (defvar Y-axis-tic " - "
21264 "String that follows number in a Y axis label.")
21268 (defvar Y-axis-label-spacing 5
21269 "Number of lines from one Y axis label to next.")
21273 (defvar X-axis-tic-symbol "|"
21274 "String to insert to point to a column in X axis.")
21278 (defvar X-axis-label-spacing
21279 (if (boundp 'graph-blank)
21280 (* 5 (length graph-blank)) 5)
21281 "Number of units from one X axis label to next.")
21287 (defun count-words-in-defun ()
21288 "Return the number of words and symbols in a defun."
21289 (beginning-of-defun)
21291 (end (save-excursion (end-of-defun) (point))))
21296 (and (< (point) end)
21298 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
21300 (setq count (1+ count)))
21307 (defun lengths-list-file (filename)
21308 "Return list of definitions' lengths within FILE.
21309 The returned list is a list of numbers.
21310 Each number is the number of words or
21311 symbols in one function definition."
21315 (message "Working on '%s' ... " filename)
21317 (let ((buffer (find-file-noselect filename))
21319 (set-buffer buffer)
21320 (setq buffer-read-only t)
21322 (goto-char (point-min))
21326 (while (re-search-forward "^(defun" nil t)
21328 (cons (count-words-in-defun) lengths-list)))
21329 (kill-buffer buffer)
21336 (defun lengths-list-many-files (list-of-files)
21337 "Return list of lengths of defuns in LIST-OF-FILES."
21338 (let (lengths-list)
21339 ;;; @r{true-or-false-test}
21340 (while list-of-files
21346 ;;; @r{Generate a lengths' list.}
21348 (expand-file-name (car list-of-files)))))
21349 ;;; @r{Make files' list shorter.}
21350 (setq list-of-files (cdr list-of-files)))
21351 ;;; @r{Return final value of lengths' list.}
21358 (defun defuns-per-range (sorted-lengths top-of-ranges)
21359 "SORTED-LENGTHS defuns in each TOP-OF-RANGES range."
21360 (let ((top-of-range (car top-of-ranges))
21361 (number-within-range 0)
21362 defuns-per-range-list)
21367 (while top-of-ranges
21371 ;; @r{Need number for numeric test.}
21372 (car sorted-lengths)
21373 (< (car sorted-lengths) top-of-range))
21375 ;; @r{Count number of definitions within current range.}
21376 (setq number-within-range (1+ number-within-range))
21377 (setq sorted-lengths (cdr sorted-lengths)))
21381 ;; @r{Exit inner loop but remain within outer loop.}
21383 (setq defuns-per-range-list
21384 (cons number-within-range defuns-per-range-list))
21385 (setq number-within-range 0) ; @r{Reset count to zero.}
21387 ;; @r{Move to next range.}
21388 (setq top-of-ranges (cdr top-of-ranges))
21389 ;; @r{Specify next top of range value.}
21390 (setq top-of-range (car top-of-ranges)))
21394 ;; @r{Exit outer loop and count the number of defuns larger than}
21395 ;; @r{ the largest top-of-range value.}
21396 (setq defuns-per-range-list
21398 (length sorted-lengths)
21399 defuns-per-range-list))
21401 ;; @r{Return a list of the number of definitions within each range,}
21402 ;; @r{ smallest to largest.}
21403 (nreverse defuns-per-range-list)))
21409 (defun column-of-graph (max-graph-height actual-height)
21410 "Return list of MAX-GRAPH-HEIGHT strings;
21411 ACTUAL-HEIGHT are graph-symbols.
21412 The graph-symbols are contiguous entries at the end
21414 The list will be inserted as one column of a graph.
21415 The strings are either graph-blank or graph-symbol."
21419 (let ((insert-list nil)
21420 (number-of-top-blanks
21421 (- max-graph-height actual-height)))
21423 ;; @r{Fill in @code{graph-symbols}.}
21424 (while (> actual-height 0)
21425 (setq insert-list (cons graph-symbol insert-list))
21426 (setq actual-height (1- actual-height)))
21430 ;; @r{Fill in @code{graph-blanks}.}
21431 (while (> number-of-top-blanks 0)
21432 (setq insert-list (cons graph-blank insert-list))
21433 (setq number-of-top-blanks
21434 (1- number-of-top-blanks)))
21436 ;; @r{Return whole list.}
21443 (defun Y-axis-element (number full-Y-label-width)
21444 "Construct a NUMBERed label element.
21445 A numbered element looks like this ' 5 - ',
21446 and is padded as needed so all line up with
21447 the element for the largest number."
21450 (let* ((leading-spaces
21451 (- full-Y-label-width
21453 (concat (number-to-string number)
21458 (make-string leading-spaces ? )
21459 (number-to-string number)
21466 (defun print-Y-axis
21467 (height full-Y-label-width &optional vertical-step)
21468 "Insert Y axis by HEIGHT and FULL-Y-LABEL-WIDTH.
21469 Height must be the maximum height of the graph.
21470 Full width is the width of the highest label element.
21471 Optionally, print according to VERTICAL-STEP."
21474 ;; Value of height and full-Y-label-width
21475 ;; are passed by 'print-graph'.
21476 (let ((start (point)))
21478 (Y-axis-column height full-Y-label-width vertical-step))
21481 ;; @r{Place point ready for inserting graph.}
21483 ;; @r{Move point forward by value of} full-Y-label-width
21484 (forward-char full-Y-label-width)))
21490 (defun print-X-axis-tic-line
21491 (number-of-X-tics X-axis-leading-spaces X-axis-tic-element)
21492 "Print ticks for X axis."
21493 (insert X-axis-leading-spaces)
21494 (insert X-axis-tic-symbol) ; @r{Under first column.}
21497 ;; @r{Insert second tic in the right spot.}
21500 (- (* symbol-width X-axis-label-spacing)
21501 ;; @r{Insert white space up to second tic symbol.}
21502 (* 2 (length X-axis-tic-symbol)))
21504 X-axis-tic-symbol))
21507 ;; @r{Insert remaining ticks.}
21508 (while (> number-of-X-tics 1)
21509 (insert X-axis-tic-element)
21510 (setq number-of-X-tics (1- number-of-X-tics))))
21516 (defun X-axis-element (number)
21517 "Construct a numbered X axis element."
21518 (let ((leading-spaces
21519 (- (* symbol-width X-axis-label-spacing)
21520 (length (number-to-string number)))))
21521 (concat (make-string leading-spaces ? )
21522 (number-to-string number))))
21528 (defun graph-body-print (numbers-list height symbol-width)
21529 "Print a bar graph of the NUMBERS-LIST.
21530 The numbers-list consists of the Y-axis values.
21531 HEIGHT is maximum height of graph.
21532 SYMBOL-WIDTH is number of each column."
21535 (let (from-position)
21536 (while numbers-list
21537 (setq from-position (point))
21539 (column-of-graph height (car numbers-list)))
21540 (goto-char from-position)
21541 (forward-char symbol-width)
21544 ;; @r{Draw graph column by column.}
21546 (setq numbers-list (cdr numbers-list)))
21547 ;; @r{Place point for X axis labels.}
21548 (forward-line height)
21555 (defun Y-axis-column
21556 (height width-of-label &optional vertical-step)
21557 "Construct list of labels for Y axis.
21558 HEIGHT is maximum height of graph.
21559 WIDTH-OF-LABEL is maximum width of label.
21562 VERTICAL-STEP, an option, is a positive integer
21563 that specifies how much a Y axis label increments
21564 for each line. For example, a step of 5 means
21565 that each line is five units of the graph."
21567 (number-per-line (or vertical-step 1)))
21570 (while (> height 1)
21571 (if (zerop (% height Y-axis-label-spacing))
21572 ;; @r{Insert label.}
21576 (* height number-per-line)
21581 ;; @r{Else, insert blanks.}
21584 (make-string width-of-label ? )
21586 (setq height (1- height)))
21589 ;; @r{Insert base line.}
21590 (setq Y-axis (cons (Y-axis-element
21591 (or vertical-step 1)
21594 (nreverse Y-axis)))
21600 (defun print-X-axis-numbered-line
21601 (number-of-X-tics X-axis-leading-spaces
21602 &optional horizontal-step)
21603 "Print line of X-axis numbers"
21604 (let ((number X-axis-label-spacing)
21605 (horizontal-step (or horizontal-step 1)))
21608 (insert X-axis-leading-spaces)
21610 (delete-char (- (1- (length (number-to-string horizontal-step)))))
21613 ;; @r{Insert white space up to next number.}
21614 (- (* symbol-width X-axis-label-spacing)
21615 (1- (length (number-to-string horizontal-step)))
21618 (number-to-string (* number horizontal-step))))
21621 ;; @r{Insert remaining numbers.}
21622 (setq number (+ number X-axis-label-spacing))
21623 (while (> number-of-X-tics 1)
21624 (insert (X-axis-element (* number horizontal-step)))
21625 (setq number (+ number X-axis-label-spacing))
21626 (setq number-of-X-tics (1- number-of-X-tics)))))
21632 (defun print-X-axis (numbers-list horizontal-step)
21633 "Print X axis labels to length of NUMBERS-LIST.
21634 Optionally, HORIZONTAL-STEP, a positive integer,
21635 specifies how much an X axis label increments for
21639 ;; Value of symbol-width and full-Y-label-width
21640 ;; are passed by 'print-graph'.
21641 (let* ((leading-spaces
21642 (make-string full-Y-label-width ? ))
21643 ;; symbol-width @r{is provided by} graph-body-print
21644 (tic-width (* symbol-width X-axis-label-spacing))
21645 (X-length (length numbers-list))
21651 ;; @r{Make a string of blanks.}
21652 (- (* symbol-width X-axis-label-spacing)
21653 (length X-axis-tic-symbol))
21657 ;; @r{Concatenate blanks with tic symbol.}
21658 X-axis-tic-symbol))
21660 (if (zerop (% X-length tic-width))
21661 (/ X-length tic-width)
21662 (1+ (/ X-length tic-width)))))
21666 (print-X-axis-tic-line
21667 tic-number leading-spaces X-tic)
21669 (print-X-axis-numbered-line
21670 tic-number leading-spaces horizontal-step)))
21676 (defun one-fiftieth (full-range)
21677 "Return list, each number of which is 1/50th previous."
21678 (mapcar (lambda (arg) (/ arg 50)) full-range))
21685 (numbers-list &optional vertical-step horizontal-step)
21686 "Print labeled bar graph of the NUMBERS-LIST.
21687 The numbers-list consists of the Y-axis values.
21691 Optionally, VERTICAL-STEP, a positive integer,
21692 specifies how much a Y axis label increments for
21693 each line. For example, a step of 5 means that
21694 each row is five units.
21698 Optionally, HORIZONTAL-STEP, a positive integer,
21699 specifies how much an X axis label increments for
21701 (let* ((symbol-width (length graph-blank))
21702 ;; @code{height} @r{is both the largest number}
21703 ;; @r{and the number with the most digits.}
21704 (height (apply 'max numbers-list))
21707 (height-of-top-line
21708 (if (zerop (% height Y-axis-label-spacing))
21711 (* (1+ (/ height Y-axis-label-spacing))
21712 Y-axis-label-spacing)))
21715 (vertical-step (or vertical-step 1))
21716 (full-Y-label-width
21720 (* height-of-top-line vertical-step))
21726 height-of-top-line full-Y-label-width vertical-step)
21728 numbers-list height-of-top-line symbol-width)
21729 (print-X-axis numbers-list horizontal-step)))
21736 @node Final printed graph
21737 @appendixsubsec The Printed Graph
21739 When made and installed, you can call the @code{print-graph} command
21745 (print-graph fiftieth-list-for-graph 50 10)
21775 50 - ***************** * *
21777 10 50 100 150 200 250 300 350
21784 The largest group of functions contain 10--19 words and symbols each.
21786 @node Free Software and Free Manuals
21787 @appendix Free Software and Free Manuals
21789 @strong{by Richard M. Stallman}
21792 The biggest deficiency in free operating systems is not in the
21793 software---it is the lack of good free manuals that we can include in
21794 these systems. Many of our most important programs do not come with
21795 full manuals. Documentation is an essential part of any software
21796 package; when an important free software package does not come with a
21797 free manual, that is a major gap. We have many such gaps today.
21799 Once upon a time, many years ago, I thought I would learn Perl. I got
21800 a copy of a free manual, but I found it hard to read. When I asked
21801 Perl users about alternatives, they told me that there were better
21802 introductory manuals---but those were not free.
21804 Why was this? The authors of the good manuals had written them for
21805 O'Reilly Associates, which published them with restrictive terms---no
21806 copying, no modification, source files not available---which exclude
21807 them from the free software community.
21809 That wasn't the first time this sort of thing has happened, and (to
21810 our community's great loss) it was far from the last. Proprietary
21811 manual publishers have enticed a great many authors to restrict their
21812 manuals since then. Many times I have heard a GNU user eagerly tell me
21813 about a manual that he is writing, with which he expects to help the
21814 GNU project---and then had my hopes dashed, as he proceeded to explain
21815 that he had signed a contract with a publisher that would restrict it
21816 so that we cannot use it.
21818 Given that writing good English is a rare skill among programmers, we
21819 can ill afford to lose manuals this way.
21821 Free documentation, like free software, is a matter of freedom, not
21822 price. The problem with these manuals was not that O'Reilly Associates
21823 charged a price for printed copies---that in itself is fine. The Free
21824 Software Foundation @uref{http://shop.fsf.org, sells printed copies} of
21825 free @uref{http://www.gnu.org/doc/doc.html, GNU manuals}, too.
21826 But GNU manuals are available in source code form, while these manuals
21827 are available only on paper. GNU manuals come with permission to copy
21828 and modify; the Perl manuals do not. These restrictions are the
21831 The criterion for a free manual is pretty much the same as for free
21832 software: it is a matter of giving all users certain
21833 freedoms. Redistribution (including commercial redistribution) must be
21834 permitted, so that the manual can accompany every copy of the program,
21835 on-line or on paper. Permission for modification is crucial too.
21837 As a general rule, I don't believe that it is essential for people to
21838 have permission to modify all sorts of articles and books. The issues
21839 for writings are not necessarily the same as those for software. For
21840 example, I don't think you or I are obliged to give permission to
21841 modify articles like this one, which describe our actions and our
21844 But there is a particular reason why the freedom to modify is crucial
21845 for documentation for free software. When people exercise their right
21846 to modify the software, and add or change its features, if they are
21847 conscientious they will change the manual too---so they can provide
21848 accurate and usable documentation with the modified program. A manual
21849 which forbids programmers to be conscientious and finish the job, or
21850 more precisely requires them to write a new manual from scratch if
21851 they change the program, does not fill our community's needs.
21853 While a blanket prohibition on modification is unacceptable, some
21854 kinds of limits on the method of modification pose no problem. For
21855 example, requirements to preserve the original author's copyright
21856 notice, the distribution terms, or the list of authors, are ok. It is
21857 also no problem to require modified versions to include notice that
21858 they were modified, even to have entire sections that may not be
21859 deleted or changed, as long as these sections deal with nontechnical
21860 topics. (Some GNU manuals have them.)
21862 These kinds of restrictions are not a problem because, as a practical
21863 matter, they don't stop the conscientious programmer from adapting the
21864 manual to fit the modified program. In other words, they don't block
21865 the free software community from making full use of the manual.
21867 However, it must be possible to modify all the technical content of
21868 the manual, and then distribute the result in all the usual media,
21869 through all the usual channels; otherwise, the restrictions do block
21870 the community, the manual is not free, and so we need another manual.
21872 Unfortunately, it is often hard to find someone to write another
21873 manual when a proprietary manual exists. The obstacle is that many
21874 users think that a proprietary manual is good enough---so they don't
21875 see the need to write a free manual. They do not see that the free
21876 operating system has a gap that needs filling.
21878 Why do users think that proprietary manuals are good enough? Some have
21879 not considered the issue. I hope this article will do something to
21882 Other users consider proprietary manuals acceptable for the same
21883 reason so many people consider proprietary software acceptable: they
21884 judge in purely practical terms, not using freedom as a
21885 criterion. These people are entitled to their opinions, but since
21886 those opinions spring from values which do not include freedom, they
21887 are no guide for those of us who do value freedom.
21889 Please spread the word about this issue. We continue to lose manuals
21890 to proprietary publishing. If we spread the word that proprietary
21891 manuals are not sufficient, perhaps the next person who wants to help
21892 GNU by writing documentation will realize, before it is too late, that
21893 he must above all make it free.
21895 We can also encourage commercial publishers to sell free, copylefted
21896 manuals instead of proprietary ones. One way you can help this is to
21897 check the distribution terms of a manual before you buy it, and prefer
21898 copylefted manuals to non-copylefted ones.
21902 Note: The Free Software Foundation maintains a page on its Web site
21903 that lists free books available from other publishers:@*
21904 @uref{http://www.gnu.org/doc/other-free-books.html}
21906 @node GNU Free Documentation License
21907 @appendix GNU Free Documentation License
21909 @cindex FDL, GNU Free Documentation License
21910 @include doclicense.texi
21916 MENU ENTRY: NODE NAME.
21922 @c Place biographical information on right-hand (verso) page
21925 \par\vfill\supereject
21927 \global\evenheadline={\hfil} \global\evenfootline={\hfil}
21928 \global\oddheadline={\hfil} \global\oddfootline={\hfil}
21931 % \par\vfill\supereject
21932 \global\evenheadline={\hfil} \global\evenfootline={\hfil}
21933 \global\oddheadline={\hfil} \global\oddfootline={\hfil}
21934 %\page\hbox{}%\page
21935 %\page\hbox{}%\page
21942 @c ================ Biographical information ================
21946 @center About the Author
21951 @node About the Author
21952 @unnumbered About the Author
21956 Robert J. Chassell has worked with GNU Emacs since 1985. He writes
21957 and edits, teaches Emacs and Emacs Lisp, and speaks throughout the
21958 world on software freedom. Chassell was a founding Director and
21959 Treasurer of the Free Software Foundation, Inc. He is co-author of
21960 the @cite{Texinfo} manual, and has edited more than a dozen other
21961 books. He graduated from Cambridge University, in England. He has an
21962 abiding interest in social and economic history and flies his own
21969 @c @c Prevent page number on blank verso, so eject it first.
21971 @c \par\vfill\supereject
21976 @c @evenheading @thispage @| @| @thistitle
21977 @c @oddheading @| @| @thispage