1 \input texinfo @c -*-texinfo-*-
2 @comment %**start of header
3 @setfilename ../../info/eintr
4 @c setfilename emacs-lisp-intro.info
5 @c sethtmlfilename emacs-lisp-intro.html
6 @settitle Programming in Emacs Lisp
11 @include emacsver.texi
13 @c ================ How to Print a Book in Various Sizes ================
15 @c This book can be printed in any of three different sizes.
16 @c Set the following @-commands appropriately.
26 @c European A4 size paper:
31 @c (Note: if you edit the book so as to change the length of the
32 @c table of contents, you may have to change the value of `pageno' below.)
34 @c <<<< For hard copy printing, this file is now
35 @c set for smallbook, which works for all sizes
36 @c of paper, and with PostScript figures >>>>
44 @c ================ Included Figures ================
46 @c If you clear this, the figures will be printed as ASCII diagrams
47 @c rather than PostScript/PDF.
48 @c (This is not relevant to Info, since Info only handles ASCII.)
49 @set print-postscript-figures
50 @c clear print-postscript-figures
52 @comment %**end of header
54 @c per rms and peterb, use 10pt fonts for the main text, mostly to
55 @c save on paper cost.
56 @c Do this inside @tex for now, so current makeinfo does not complain.
62 \global\hbadness=6666 % don't worry about not-too-underfull boxes
65 @c These refer to the printed book sold by the FSF.
66 @set edition-number 3.10
67 @set update-date 28 October 2009
69 @c For next or subsequent edition:
70 @c create function using with-output-to-temp-buffer
71 @c create a major mode, with keymaps
72 @c run an asynchronous process, like grep or diff
74 @c For 8.5 by 11 inch format: do not use such a small amount of
75 @c whitespace between paragraphs as smallbook format
78 \global\parskip 6pt plus 1pt
82 @c For all sized formats: print within-book cross
83 @c reference with ``...'' rather than [...]
85 @c This works with the texinfo.tex file, version 2003-05-04.08,
86 @c in the Texinfo version 4.6 of the 2003 Jun 13 distribution.
89 \if \xrefprintnodename
90 \global\def\xrefprintnodename#1{\unskip, ``#1''}
92 \global\def\xrefprintnodename#1{ ``#1''}
94 % \global\def\xrefprintnodename#1{, ``#1''}
97 @c ----------------------------------------------------
99 @dircategory GNU Emacs Lisp
101 * Emacs Lisp Intro: (eintr).
102 A simple introduction to Emacs Lisp programming.
106 This is an @cite{Introduction to Programming in Emacs Lisp}, for
107 people who are not programmers.
110 Edition @value{edition-number}, @value{update-date}
113 Distributed with Emacs version @value{EMACSVER}.
116 Copyright @copyright{} 1990--1995, 1997, 2001--2013 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
145 Permission is granted to copy, distribute and/or modify this document
146 under the terms of the GNU Free Documentation License, Version 1.3 or
147 any later version published by the Free Software Foundation; there
148 being no Invariant Section, with the Front-Cover Texts being ``A GNU
149 Manual'', and with the Back-Cover Texts as in (a) below. A copy of
150 the license is included in the section entitled ``GNU Free
151 Documentation License''.
153 (a) The FSF's Back-Cover Text is: ``You have the freedom to
154 copy and modify this GNU manual. Buying copies from the FSF
155 supports it in developing GNU and promoting software freedom.''
158 @c half title; two lines here, so do not use `shorttitlepage'
161 \hbox{}\vskip 1.5in \chaprm \centerline{An Introduction to}%
163 {\begingroup\hbox{}\vskip 0.25in \chaprm%
164 \centerline{Programming in Emacs Lisp}%
165 \endgroup\page\hbox{}\page}
170 @center @titlefont{An Introduction to}
172 @center @titlefont{Programming in Emacs Lisp}
174 @center Revised Third Edition
176 @center by Robert J. Chassell
179 @vskip 0pt plus 1filll
185 @evenheading @thispage @| @| @thischapter
186 @oddheading @thissection @| @| @thispage
190 @c Keep T.O.C. short by tightening up for largebook
193 \global\parskip 2pt plus 1pt
194 \global\advance\baselineskip by -1pt
204 @top An Introduction to Programming in Emacs Lisp
208 <p>The homepage for GNU Emacs is at
209 <a href="/software/emacs/">http://www.gnu.org/software/emacs/</a>.<br>
210 To view this manual in other formats, click
211 <a href="/software/emacs/manual/eintr.html">here</a>.
217 This master menu first lists each chapter and index; then it lists
218 every node in every chapter.
221 @c >>>> Set pageno appropriately <<<<
223 @c The first page of the Preface is a roman numeral; it is the first
224 @c right handed page after the Table of Contents; hence the following
225 @c setting must be for an odd negative number.
228 @c global@pageno = -11
231 @set COUNT-WORDS count-words-example
232 @c Length of variable name chosen so that things still line up when expanded.
235 * Preface:: What to look for.
236 * List Processing:: What is Lisp?
237 * Practicing Evaluation:: Running several programs.
238 * Writing Defuns:: How to write function definitions.
239 * Buffer Walk Through:: Exploring a few buffer-related functions.
240 * More Complex:: A few, even more complex functions.
241 * Narrowing & Widening:: Restricting your and Emacs attention to
243 * car cdr & cons:: Fundamental functions in Lisp.
244 * Cutting & Storing Text:: Removing text and saving it.
245 * List Implementation:: How lists are implemented in the computer.
246 * Yanking:: Pasting stored text.
247 * Loops & Recursion:: How to repeat a process.
248 * Regexp Search:: Regular expression searches.
249 * Counting Words:: A review of repetition and regexps.
250 * Words in a defun:: Counting words in a @code{defun}.
251 * Readying a Graph:: A prototype graph printing function.
252 * Emacs Initialization:: How to write a @file{.emacs} file.
253 * Debugging:: How to run the Emacs Lisp debuggers.
254 * Conclusion:: Now you have the basics.
255 * the-the:: An appendix: how to find reduplicated words.
256 * Kill Ring:: An appendix: how the kill ring works.
257 * Full Graph:: How to create a graph with labeled axes.
258 * Free Software and Free Manuals::
259 * GNU Free Documentation License::
264 --- The Detailed Node Listing ---
268 * Why:: Why learn Emacs Lisp?
269 * On Reading this Text:: Read, gain familiarity, pick up habits....
270 * Who You Are:: For whom this is written.
272 * Note for Novices:: You can read this as a novice.
277 * Lisp Lists:: What are lists?
278 * Run a Program:: Any list in Lisp is a program ready to run.
279 * Making Errors:: Generating an error message.
280 * Names & Definitions:: Names of symbols and function definitions.
281 * Lisp Interpreter:: What the Lisp interpreter does.
282 * Evaluation:: Running a program.
283 * Variables:: Returning a value from a variable.
284 * Arguments:: Passing information to a function.
285 * set & setq:: Setting the value of a variable.
286 * Summary:: The major points.
287 * Error Message Exercises::
291 * Numbers Lists:: List have numbers, other lists, in them.
292 * Lisp Atoms:: Elemental entities.
293 * Whitespace in Lists:: Formatting lists to be readable.
294 * Typing Lists:: How GNU Emacs helps you type lists.
298 * Complications:: Variables, Special forms, Lists within.
299 * Byte Compiling:: Specially processing code for speed.
303 * How the Interpreter Acts:: Returns and Side Effects...
304 * Evaluating Inner Lists:: Lists within lists...
308 * fill-column Example::
309 * Void Function:: The error message for a symbol
311 * Void Variable:: The error message for a symbol without a value.
315 * Data types:: Types of data passed to a function.
316 * Args as Variable or List:: An argument can be the value
317 of a variable or list.
318 * Variable Number of Arguments:: Some functions may take a
319 variable number of arguments.
320 * Wrong Type of Argument:: Passing an argument of the wrong type
322 * message:: A useful function for sending messages.
324 Setting the Value of a Variable
326 * Using set:: Setting values.
327 * Using setq:: Setting a quoted value.
328 * Counting:: Using @code{setq} to count.
330 Practicing Evaluation
332 * How to Evaluate:: Typing editing commands or @kbd{C-x C-e}
334 * Buffer Names:: Buffers and files are different.
335 * Getting Buffers:: Getting a buffer itself, not merely its name.
336 * Switching Buffers:: How to change to another buffer.
337 * Buffer Size & Locations:: Where point is located and the size of
339 * Evaluation Exercise::
341 How To Write Function Definitions
343 * Primitive Functions::
344 * defun:: The @code{defun} macro.
345 * Install:: Install a function definition.
346 * Interactive:: Making a function interactive.
347 * Interactive Options:: Different options for @code{interactive}.
348 * Permanent Installation:: Installing code permanently.
349 * let:: Creating and initializing local variables.
351 * else:: If--then--else expressions.
352 * Truth & Falsehood:: What Lisp considers false and true.
353 * save-excursion:: Keeping track of point, mark, and buffer.
357 Install a Function Definition
359 * Effect of installation::
360 * Change a defun:: How to change a function definition.
362 Make a Function Interactive
364 * Interactive multiply-by-seven:: An overview.
365 * multiply-by-seven in detail:: The interactive version.
369 * Prevent confusion::
370 * Parts of let Expression::
371 * Sample let Expression::
372 * Uninitialized let Variables::
374 The @code{if} Special Form
376 * if in more detail::
377 * type-of-animal in detail:: An example of an @code{if} expression.
379 Truth and Falsehood in Emacs Lisp
381 * nil explained:: @code{nil} has two meanings.
383 @code{save-excursion}
385 * Point and mark:: A review of various locations.
386 * Template for save-excursion::
388 A Few Buffer--Related Functions
390 * Finding More:: How to find more information.
391 * simplified-beginning-of-buffer:: Shows @code{goto-char},
392 @code{point-min}, and @code{push-mark}.
393 * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}.
394 * append-to-buffer:: Uses @code{save-excursion} and
395 @code{insert-buffer-substring}.
396 * Buffer Related Review:: Review.
399 The Definition of @code{mark-whole-buffer}
401 * mark-whole-buffer overview::
402 * Body of mark-whole-buffer:: Only three lines of code.
404 The Definition of @code{append-to-buffer}
406 * append-to-buffer overview::
407 * append interactive:: A two part interactive expression.
408 * append-to-buffer body:: Incorporates a @code{let} expression.
409 * append save-excursion:: How the @code{save-excursion} works.
411 A Few More Complex Functions
413 * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}.
414 * insert-buffer:: Read-only, and with @code{or}.
415 * beginning-of-buffer:: Shows @code{goto-char},
416 @code{point-min}, and @code{push-mark}.
417 * Second Buffer Related Review::
418 * optional Exercise::
420 The Definition of @code{insert-buffer}
422 * insert-buffer code::
423 * insert-buffer interactive:: When you can read, but not write.
424 * insert-buffer body:: The body has an @code{or} and a @code{let}.
425 * if & or:: Using an @code{if} instead of an @code{or}.
426 * Insert or:: How the @code{or} expression works.
427 * Insert let:: Two @code{save-excursion} expressions.
428 * New insert-buffer::
430 The Interactive Expression in @code{insert-buffer}
432 * Read-only buffer:: When a buffer cannot be modified.
433 * b for interactive:: An existing buffer or else its name.
435 Complete Definition of @code{beginning-of-buffer}
437 * Optional Arguments::
438 * beginning-of-buffer opt arg:: Example with optional argument.
439 * beginning-of-buffer complete::
441 @code{beginning-of-buffer} with an Argument
443 * Disentangle beginning-of-buffer::
444 * Large buffer case::
445 * Small buffer case::
447 Narrowing and Widening
449 * Narrowing advantages:: The advantages of narrowing
450 * save-restriction:: The @code{save-restriction} special form.
451 * what-line:: The number of the line that point is on.
454 @code{car}, @code{cdr}, @code{cons}: Fundamental Functions
456 * Strange Names:: An historical aside: why the strange names?
457 * car & cdr:: Functions for extracting part of a list.
458 * cons:: Constructing a list.
459 * nthcdr:: Calling @code{cdr} repeatedly.
461 * setcar:: Changing the first element of a list.
462 * setcdr:: Changing the rest of a list.
468 * length:: How to find the length of a list.
470 Cutting and Storing Text
472 * Storing Text:: Text is stored in a list.
473 * zap-to-char:: Cutting out text up to a character.
474 * kill-region:: Cutting text out of a region.
475 * copy-region-as-kill:: A definition for copying text.
476 * Digression into C:: Minor note on C programming language macros.
477 * defvar:: How to give a variable an initial value.
478 * cons & search-fwd Review::
483 * Complete zap-to-char:: The complete implementation.
484 * zap-to-char interactive:: A three part interactive expression.
485 * zap-to-char body:: A short overview.
486 * search-forward:: How to search for a string.
487 * progn:: The @code{progn} special form.
488 * Summing up zap-to-char:: Using @code{point} and @code{search-forward}.
492 * Complete kill-region:: The function definition.
493 * condition-case:: Dealing with a problem.
496 @code{copy-region-as-kill}
498 * Complete copy-region-as-kill:: The complete function definition.
499 * copy-region-as-kill body:: The body of @code{copy-region-as-kill}.
501 The Body of @code{copy-region-as-kill}
503 * last-command & this-command::
504 * kill-append function::
505 * kill-new function::
507 Initializing a Variable with @code{defvar}
509 * See variable current value::
510 * defvar and asterisk::
512 How Lists are Implemented
515 * Symbols as Chest:: Exploring a powerful metaphor.
520 * Kill Ring Overview::
521 * kill-ring-yank-pointer:: The kill ring is a list.
522 * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable.
526 * while:: Causing a stretch of code to repeat.
528 * Recursion:: Causing a function to call itself.
533 * Looping with while:: Repeat so long as test returns true.
534 * Loop Example:: A @code{while} loop that uses a list.
535 * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}.
536 * Incrementing Loop:: A loop with an incrementing counter.
537 * Incrementing Loop Details::
538 * Decrementing Loop:: A loop with a decrementing counter.
540 Details of an Incrementing Loop
542 * Incrementing Example:: Counting pebbles in a triangle.
543 * Inc Example parts:: The parts of the function definition.
544 * Inc Example altogether:: Putting the function definition together.
546 Loop with a Decrementing Counter
548 * Decrementing Example:: More pebbles on the beach.
549 * Dec Example parts:: The parts of the function definition.
550 * Dec Example altogether:: Putting the function definition together.
552 Save your time: @code{dolist} and @code{dotimes}
559 * Building Robots:: Same model, different serial number ...
560 * Recursive Definition Parts:: Walk until you stop ...
561 * Recursion with list:: Using a list as the test whether to recurse.
562 * Recursive triangle function::
563 * Recursion with cond::
564 * Recursive Patterns:: Often used templates.
565 * No Deferment:: Don't store up work ...
566 * No deferment solution::
568 Recursion in Place of a Counter
570 * Recursive Example arg of 1 or 2::
571 * Recursive Example arg of 3 or 4::
579 Regular Expression Searches
581 * sentence-end:: The regular expression for @code{sentence-end}.
582 * re-search-forward:: Very similar to @code{search-forward}.
583 * forward-sentence:: A straightforward example of regexp search.
584 * forward-paragraph:: A somewhat complex example.
585 * etags:: How to create your own @file{TAGS} table.
587 * re-search Exercises::
589 @code{forward-sentence}
591 * Complete forward-sentence::
592 * fwd-sentence while loops:: Two @code{while} loops.
593 * fwd-sentence re-search:: A regular expression search.
595 @code{forward-paragraph}: a Goldmine of Functions
597 * forward-paragraph in brief:: Key parts of the function definition.
598 * fwd-para let:: The @code{let*} expression.
599 * fwd-para while:: The forward motion @code{while} loop.
601 Counting: Repetition and Regexps
604 * @value{COUNT-WORDS}:: Use a regexp, but find a problem.
605 * recursive-count-words:: Start with case of no words in region.
606 * Counting Exercise::
608 The @code{@value{COUNT-WORDS}} Function
610 * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop.
611 * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}.
613 Counting Words in a @code{defun}
615 * Divide and Conquer::
616 * Words and Symbols:: What to count?
617 * Syntax:: What constitutes a word or symbol?
618 * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}.
619 * Several defuns:: Counting several defuns in a file.
620 * Find a File:: Do you want to look at a file?
621 * lengths-list-file:: A list of the lengths of many definitions.
622 * Several files:: Counting in definitions in different files.
623 * Several files recursively:: Recursively counting in different files.
624 * Prepare the data:: Prepare the data for display in a graph.
626 Count Words in @code{defuns} in Different Files
628 * lengths-list-many-files:: Return a list of the lengths of defuns.
629 * append:: Attach one list to another.
631 Prepare the Data for Display in a Graph
633 * Data for Display in Detail::
634 * Sorting:: Sorting lists.
635 * Files List:: Making a list of files.
636 * Counting function definitions::
640 * Columns of a graph::
641 * graph-body-print:: How to print the body of a graph.
642 * recursive-graph-body-print::
644 * Line Graph Exercise::
646 Your @file{.emacs} File
648 * Default Configuration::
649 * Site-wide Init:: You can write site-wide init files.
650 * defcustom:: Emacs will write code for you.
651 * Beginning a .emacs File:: How to write a @code{.emacs file}.
652 * Text and Auto-fill:: Automatically wrap lines.
653 * Mail Aliases:: Use abbreviations for email addresses.
654 * Indent Tabs Mode:: Don't use tabs with @TeX{}
655 * Keybindings:: Create some personal keybindings.
656 * Keymaps:: More about key binding.
657 * Loading Files:: Load (i.e., evaluate) files automatically.
658 * Autoload:: Make functions available.
659 * Simple Extension:: Define a function; bind it to a key.
660 * X11 Colors:: Colors in X.
662 * Mode Line:: How to customize your mode line.
666 * debug:: How to use the built-in debugger.
667 * debug-on-entry:: Start debugging when you call a function.
668 * debug-on-quit:: Start debugging when you quit with @kbd{C-g}.
669 * edebug:: How to use Edebug, a source level debugger.
670 * Debugging Exercises::
672 Handling the Kill Ring
674 * What the Kill Ring Does::
676 * yank:: Paste a copy of a clipped element.
677 * yank-pop:: Insert element pointed to.
680 The @code{current-kill} Function
682 * Code for current-kill::
683 * Understanding current-kill::
685 @code{current-kill} in Outline
687 * Body of current-kill::
688 * Digression concerning error:: How to mislead humans, but not computers.
689 * Determining the Element::
691 A Graph with Labeled Axes
694 * print-graph Varlist:: @code{let} expression in @code{print-graph}.
695 * print-Y-axis:: Print a label for the vertical axis.
696 * print-X-axis:: Print a horizontal label.
697 * Print Whole Graph:: The function to print a complete graph.
699 The @code{print-Y-axis} Function
701 * print-Y-axis in Detail::
702 * Height of label:: What height for the Y axis?
703 * Compute a Remainder:: How to compute the remainder of a division.
704 * Y Axis Element:: Construct a line for the Y axis.
705 * Y-axis-column:: Generate a list of Y axis labels.
706 * print-Y-axis Penultimate:: A not quite final version.
708 The @code{print-X-axis} Function
710 * Similarities differences:: Much like @code{print-Y-axis}, but not exactly.
711 * X Axis Tic Marks:: Create tic marks for the horizontal axis.
713 Printing the Whole Graph
715 * The final version:: A few changes.
716 * Test print-graph:: Run a short test.
717 * Graphing words in defuns:: Executing the final code.
718 * lambda:: How to write an anonymous function.
719 * mapcar:: Apply a function to elements of a list.
720 * Another Bug:: Yet another bug @dots{} most insidious.
721 * Final printed graph:: The graph itself!
729 Most of the GNU Emacs integrated environment is written in the programming
730 language called Emacs Lisp. The code written in this programming
731 language is the software---the sets of instructions---that tell the
732 computer what to do when you give it commands. Emacs is designed so
733 that you can write new code in Emacs Lisp and easily install it as an
734 extension to the editor.
736 (GNU Emacs is sometimes called an ``extensible editor'', but it does
737 much more than provide editing capabilities. It is better to refer to
738 Emacs as an ``extensible computing environment''. However, that
739 phrase is quite a mouthful. It is easier to refer to Emacs simply as
740 an editor. Moreover, everything you do in Emacs---find the Mayan date
741 and phases of the moon, simplify polynomials, debug code, manage
742 files, read letters, write books---all these activities are kinds of
743 editing in the most general sense of the word.)
746 * Why:: Why learn Emacs Lisp?
747 * On Reading this Text:: Read, gain familiarity, pick up habits....
748 * Who You Are:: For whom this is written.
750 * Note for Novices:: You can read this as a novice.
756 @unnumberedsec Why Study Emacs Lisp?
759 Although Emacs Lisp is usually thought of in association only with Emacs,
760 it is a full computer programming language. You can use Emacs Lisp as
761 you would any other programming language.
763 Perhaps you want to understand programming; perhaps you want to extend
764 Emacs; or perhaps you want to become a programmer. This introduction to
765 Emacs Lisp is designed to get you started: to guide you in learning the
766 fundamentals of programming, and more importantly, to show you how you
767 can teach yourself to go further.
769 @node On Reading this Text
770 @unnumberedsec On Reading this Text
772 All through this document, you will see little sample programs you can
773 run inside of Emacs. If you read this document in Info inside of GNU
774 Emacs, you can run the programs as they appear. (This is easy to do and
775 is explained when the examples are presented.) Alternatively, you can
776 read this introduction as a printed book while sitting beside a computer
777 running Emacs. (This is what I like to do; I like printed books.) If
778 you don't have a running Emacs beside you, you can still read this book,
779 but in this case, it is best to treat it as a novel or as a travel guide
780 to a country not yet visited: interesting, but not the same as being
783 Much of this introduction is dedicated to walkthroughs or guided tours
784 of code used in GNU Emacs. These tours are designed for two purposes:
785 first, to give you familiarity with real, working code (code you use
786 every day); and, second, to give you familiarity with the way Emacs
787 works. It is interesting to see how a working environment is
790 hope that you will pick up the habit of browsing through source code.
791 You can learn from it and mine it for ideas. Having GNU Emacs is like
792 having a dragon's cave of treasures.
794 In addition to learning about Emacs as an editor and Emacs Lisp as a
795 programming language, the examples and guided tours will give you an
796 opportunity to get acquainted with Emacs as a Lisp programming
797 environment. GNU Emacs supports programming and provides tools that
798 you will want to become comfortable using, such as @kbd{M-.} (the key
799 which invokes the @code{find-tag} command). You will also learn about
800 buffers and other objects that are part of the environment.
801 Learning about these features of Emacs is like learning new routes
802 around your home town.
805 In addition, I have written several programs as extended examples.
806 Although these are examples, the programs are real. I use them.
807 Other people use them. You may use them. Beyond the fragments of
808 programs used for illustrations, there is very little in here that is
809 `just for teaching purposes'; what you see is used. This is a great
810 advantage of Emacs Lisp: it is easy to learn to use it for work.
813 Finally, I hope to convey some of the skills for using Emacs to
814 learn aspects of programming that you don't know. You can often use
815 Emacs to help you understand what puzzles you or to find out how to do
816 something new. This self-reliance is not only a pleasure, but an
820 @unnumberedsec For Whom This is Written
822 This text is written as an elementary introduction for people who are
823 not programmers. If you are a programmer, you may not be satisfied with
824 this primer. The reason is that you may have become expert at reading
825 reference manuals and be put off by the way this text is organized.
827 An expert programmer who reviewed this text said to me:
830 @i{I prefer to learn from reference manuals. I ``dive into'' each
831 paragraph, and ``come up for air'' between paragraphs.}
833 @i{When I get to the end of a paragraph, I assume that that subject is
834 done, finished, that I know everything I need (with the
835 possible exception of the case when the next paragraph starts talking
836 about it in more detail). I expect that a well written reference manual
837 will not have a lot of redundancy, and that it will have excellent
838 pointers to the (one) place where the information I want is.}
841 This introduction is not written for this person!
843 Firstly, I try to say everything at least three times: first, to
844 introduce it; second, to show it in context; and third, to show it in a
845 different context, or to review it.
847 Secondly, I hardly ever put all the information about a subject in one
848 place, much less in one paragraph. To my way of thinking, that imposes
849 too heavy a burden on the reader. Instead I try to explain only what
850 you need to know at the time. (Sometimes I include a little extra
851 information so you won't be surprised later when the additional
852 information is formally introduced.)
854 When you read this text, you are not expected to learn everything the
855 first time. Frequently, you need only make, as it were, a `nodding
856 acquaintance' with some of the items mentioned. My hope is that I have
857 structured the text and given you enough hints that you will be alert to
858 what is important, and concentrate on it.
860 You will need to ``dive into'' some paragraphs; there is no other way
861 to read them. But I have tried to keep down the number of such
862 paragraphs. This book is intended as an approachable hill, rather than
863 as a daunting mountain.
865 This introduction to @cite{Programming in Emacs Lisp} has a companion
868 @cite{The GNU Emacs Lisp Reference Manual}.
871 @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU
872 Emacs Lisp Reference Manual}.
874 The reference manual has more detail than this introduction. In the
875 reference manual, all the information about one topic is concentrated
876 in one place. You should turn to it if you are like the programmer
877 quoted above. And, of course, after you have read this
878 @cite{Introduction}, you will find the @cite{Reference Manual} useful
879 when you are writing your own programs.
882 @unnumberedsec Lisp History
885 Lisp was first developed in the late 1950s at the Massachusetts
886 Institute of Technology for research in artificial intelligence. The
887 great power of the Lisp language makes it superior for other purposes as
888 well, such as writing editor commands and integrated environments.
892 GNU Emacs Lisp is largely inspired by Maclisp, which was written at MIT
893 in the 1960s. It is somewhat inspired by Common Lisp, which became a
894 standard in the 1980s. However, Emacs Lisp is much simpler than Common
895 Lisp. (The standard Emacs distribution contains an optional extensions
896 file, @file{cl.el}, that adds many Common Lisp features to Emacs Lisp.)
898 @node Note for Novices
899 @unnumberedsec A Note for Novices
901 If you don't know GNU Emacs, you can still read this document
902 profitably. However, I recommend you learn Emacs, if only to learn to
903 move around your computer screen. You can teach yourself how to use
904 Emacs with the on-line tutorial. To use it, type @kbd{C-h t}. (This
905 means you press and release the @key{CTRL} key and the @kbd{h} at the
906 same time, and then press and release @kbd{t}.)
908 Also, I often refer to one of Emacs's standard commands by listing the
909 keys which you press to invoke the command and then giving the name of
910 the command in parentheses, like this: @kbd{M-C-\}
911 (@code{indent-region}). What this means is that the
912 @code{indent-region} command is customarily invoked by typing
913 @kbd{M-C-\}. (You can, if you wish, change the keys that are typed to
914 invoke the command; this is called @dfn{rebinding}. @xref{Keymaps, ,
915 Keymaps}.) The abbreviation @kbd{M-C-\} means that you type your
916 @key{META} key, @key{CTRL} key and @key{\} key all at the same time.
917 (On many modern keyboards the @key{META} key is labeled
919 Sometimes a combination like this is called a keychord, since it is
920 similar to the way you play a chord on a piano. If your keyboard does
921 not have a @key{META} key, the @key{ESC} key prefix is used in place
922 of it. In this case, @kbd{M-C-\} means that you press and release your
923 @key{ESC} key and then type the @key{CTRL} key and the @key{\} key at
924 the same time. But usually @kbd{M-C-\} means press the @key{CTRL} key
925 along with the key that is labeled @key{ALT} and, at the same time,
926 press the @key{\} key.
928 In addition to typing a lone keychord, you can prefix what you type
929 with @kbd{C-u}, which is called the `universal argument'. The
930 @kbd{C-u} keychord passes an argument to the subsequent command.
931 Thus, to indent a region of plain text by 6 spaces, mark the region,
932 and then type @w{@kbd{C-u 6 M-C-\}}. (If you do not specify a number,
933 Emacs either passes the number 4 to the command or otherwise runs the
934 command differently than it would otherwise.) @xref{Arguments, ,
935 Numeric Arguments, emacs, The GNU Emacs Manual}.
937 If you are reading this in Info using GNU Emacs, you can read through
938 this whole document just by pressing the space bar, @key{SPC}.
939 (To learn about Info, type @kbd{C-h i} and then select Info.)
941 A note on terminology: when I use the word Lisp alone, I often am
942 referring to the various dialects of Lisp in general, but when I speak
943 of Emacs Lisp, I am referring to GNU Emacs Lisp in particular.
946 @unnumberedsec Thank You
948 My thanks to all who helped me with this book. My especial thanks to
949 @r{Jim Blandy}, @r{Noah Friedman}, @w{Jim Kingdon}, @r{Roland
950 McGrath}, @w{Frank Ritter}, @w{Randy Smith}, @w{Richard M.
951 Stallman}, and @w{Melissa Weisshaus}. My thanks also go to both
952 @w{Philip Johnson} and @w{David Stampe} for their patient
953 encouragement. My mistakes are my own.
960 @c ================ Beginning of main text ================
962 @c Start main text on right-hand (verso) page
965 \par\vfill\supereject
968 \par\vfill\supereject
970 \par\vfill\supereject
972 \par\vfill\supereject
976 @c Note: this resetting of the page number back to 1 causes TeX to gripe
977 @c about already having seen page numbers 1-4 before (in the preface):
978 @c pdfTeX warning (ext4): destination with the same identifier (name{1})
979 @c has been already used, duplicate ignored
980 @c I guess that is harmless (what happens if a later part of the text
981 @c makes a link to something in the first 4 pages though?).
982 @c E.g., note that the Emacs manual has a preface, but does not bother
983 @c resetting the page numbers back to 1 after that.
986 @evenheading @thispage @| @| @thischapter
987 @oddheading @thissection @| @| @thispage
991 @node List Processing
992 @chapter List Processing
994 To the untutored eye, Lisp is a strange programming language. In Lisp
995 code there are parentheses everywhere. Some people even claim that
996 the name stands for `Lots of Isolated Silly Parentheses'. But the
997 claim is unwarranted. Lisp stands for LISt Processing, and the
998 programming language handles @emph{lists} (and lists of lists) by
999 putting them between parentheses. The parentheses mark the boundaries
1000 of the list. Sometimes a list is preceded by a single apostrophe or
1001 quotation mark, @samp{'}@footnote{The single apostrophe or quotation
1002 mark is an abbreviation for the function @code{quote}; you need not
1003 think about functions now; functions are defined in @ref{Making
1004 Errors, , Generate an Error Message}.} Lists are the basis of Lisp.
1007 * Lisp Lists:: What are lists?
1008 * Run a Program:: Any list in Lisp is a program ready to run.
1009 * Making Errors:: Generating an error message.
1010 * Names & Definitions:: Names of symbols and function definitions.
1011 * Lisp Interpreter:: What the Lisp interpreter does.
1012 * Evaluation:: Running a program.
1013 * Variables:: Returning a value from a variable.
1014 * Arguments:: Passing information to a function.
1015 * set & setq:: Setting the value of a variable.
1016 * Summary:: The major points.
1017 * Error Message Exercises::
1024 In Lisp, a list looks like this: @code{'(rose violet daisy buttercup)}.
1025 This list is preceded by a single apostrophe. It could just as well be
1026 written as follows, which looks more like the kind of list you are likely
1027 to be familiar with:
1039 The elements of this list are the names of the four different flowers,
1040 separated from each other by whitespace and surrounded by parentheses,
1041 like flowers in a field with a stone wall around them.
1042 @cindex Flowers in a field
1045 * Numbers Lists:: List have numbers, other lists, in them.
1046 * Lisp Atoms:: Elemental entities.
1047 * Whitespace in Lists:: Formatting lists to be readable.
1048 * Typing Lists:: How GNU Emacs helps you type lists.
1053 @unnumberedsubsec Numbers, Lists inside of Lists
1056 Lists can also have numbers in them, as in this list: @code{(+ 2 2)}.
1057 This list has a plus-sign, @samp{+}, followed by two @samp{2}s, each
1058 separated by whitespace.
1060 In Lisp, both data and programs are represented the same way; that is,
1061 they are both lists of words, numbers, or other lists, separated by
1062 whitespace and surrounded by parentheses. (Since a program looks like
1063 data, one program may easily serve as data for another; this is a very
1064 powerful feature of Lisp.) (Incidentally, these two parenthetical
1065 remarks are @emph{not} Lisp lists, because they contain @samp{;} and
1066 @samp{.} as punctuation marks.)
1069 Here is another list, this time with a list inside of it:
1072 '(this list has (a list inside of it))
1075 The components of this list are the words @samp{this}, @samp{list},
1076 @samp{has}, and the list @samp{(a list inside of it)}. The interior
1077 list is made up of the words @samp{a}, @samp{list}, @samp{inside},
1078 @samp{of}, @samp{it}.
1081 @subsection Lisp Atoms
1084 In Lisp, what we have been calling words are called @dfn{atoms}. This
1085 term comes from the historical meaning of the word atom, which means
1086 `indivisible'. As far as Lisp is concerned, the words we have been
1087 using in the lists cannot be divided into any smaller parts and still
1088 mean the same thing as part of a program; likewise with numbers and
1089 single character symbols like @samp{+}. On the other hand, unlike an
1090 ancient atom, a list can be split into parts. (@xref{car cdr & cons,
1091 , @code{car} @code{cdr} & @code{cons} Fundamental Functions}.)
1093 In a list, atoms are separated from each other by whitespace. They can be
1094 right next to a parenthesis.
1096 @cindex @samp{empty list} defined
1097 Technically speaking, a list in Lisp consists of parentheses surrounding
1098 atoms separated by whitespace or surrounding other lists or surrounding
1099 both atoms and other lists. A list can have just one atom in it or
1100 have nothing in it at all. A list with nothing in it looks like this:
1101 @code{()}, and is called the @dfn{empty list}. Unlike anything else, an
1102 empty list is considered both an atom and a list at the same time.
1104 @cindex Symbolic expressions, introduced
1105 @cindex @samp{expression} defined
1106 @cindex @samp{form} defined
1107 The printed representation of both atoms and lists are called
1108 @dfn{symbolic expressions} or, more concisely, @dfn{s-expressions}.
1109 The word @dfn{expression} by itself can refer to either the printed
1110 representation, or to the atom or list as it is held internally in the
1111 computer. Often, people use the term @dfn{expression}
1112 indiscriminately. (Also, in many texts, the word @dfn{form} is used
1113 as a synonym for expression.)
1115 Incidentally, the atoms that make up our universe were named such when
1116 they were thought to be indivisible; but it has been found that physical
1117 atoms are not indivisible. Parts can split off an atom or it can
1118 fission into two parts of roughly equal size. Physical atoms were named
1119 prematurely, before their truer nature was found. In Lisp, certain
1120 kinds of atom, such as an array, can be separated into parts; but the
1121 mechanism for doing this is different from the mechanism for splitting a
1122 list. As far as list operations are concerned, the atoms of a list are
1125 As in English, the meanings of the component letters of a Lisp atom
1126 are different from the meaning the letters make as a word. For
1127 example, the word for the South American sloth, the @samp{ai}, is
1128 completely different from the two words, @samp{a}, and @samp{i}.
1130 There are many kinds of atom in nature but only a few in Lisp: for
1131 example, @dfn{numbers}, such as 37, 511, or 1729, and @dfn{symbols}, such
1132 as @samp{+}, @samp{foo}, or @samp{forward-line}. The words we have
1133 listed in the examples above are all symbols. In everyday Lisp
1134 conversation, the word ``atom'' is not often used, because programmers
1135 usually try to be more specific about what kind of atom they are dealing
1136 with. Lisp programming is mostly about symbols (and sometimes numbers)
1137 within lists. (Incidentally, the preceding three word parenthetical
1138 remark is a proper list in Lisp, since it consists of atoms, which in
1139 this case are symbols, separated by whitespace and enclosed by
1140 parentheses, without any non-Lisp punctuation.)
1143 Text between double quotation marks---even sentences or
1144 paragraphs---is also an atom. Here is an example:
1145 @cindex Text between double quotation marks
1148 '(this list includes "text between quotation marks.")
1151 @cindex @samp{string} defined
1153 In Lisp, all of the quoted text including the punctuation mark and the
1154 blank spaces is a single atom. This kind of atom is called a
1155 @dfn{string} (for `string of characters') and is the sort of thing that
1156 is used for messages that a computer can print for a human to read.
1157 Strings are a different kind of atom than numbers or symbols and are
1160 @node Whitespace in Lists
1161 @subsection Whitespace in Lists
1162 @cindex Whitespace in lists
1165 The amount of whitespace in a list does not matter. From the point of view
1166 of the Lisp language,
1177 is exactly the same as this:
1180 '(this list looks like this)
1183 Both examples show what to Lisp is the same list, the list made up of
1184 the symbols @samp{this}, @samp{list}, @samp{looks}, @samp{like}, and
1185 @samp{this} in that order.
1187 Extra whitespace and newlines are designed to make a list more readable
1188 by humans. When Lisp reads the expression, it gets rid of all the extra
1189 whitespace (but it needs to have at least one space between atoms in
1190 order to tell them apart.)
1192 Odd as it seems, the examples we have seen cover almost all of what Lisp
1193 lists look like! Every other list in Lisp looks more or less like one
1194 of these examples, except that the list may be longer and more complex.
1195 In brief, a list is between parentheses, a string is between quotation
1196 marks, a symbol looks like a word, and a number looks like a number.
1197 (For certain situations, square brackets, dots and a few other special
1198 characters may be used; however, we will go quite far without them.)
1201 @subsection GNU Emacs Helps You Type Lists
1202 @cindex Help typing lists
1203 @cindex Formatting help
1205 When you type a Lisp expression in GNU Emacs using either Lisp
1206 Interaction mode or Emacs Lisp mode, you have available to you several
1207 commands to format the Lisp expression so it is easy to read. For
1208 example, pressing the @key{TAB} key automatically indents the line the
1209 cursor is on by the right amount. A command to properly indent the
1210 code in a region is customarily bound to @kbd{M-C-\}. Indentation is
1211 designed so that you can see which elements of a list belong to which
1212 list---elements of a sub-list are indented more than the elements of
1215 In addition, when you type a closing parenthesis, Emacs momentarily
1216 jumps the cursor back to the matching opening parenthesis, so you can
1217 see which one it is. This is very useful, since every list you type
1218 in Lisp must have its closing parenthesis match its opening
1219 parenthesis. (@xref{Major Modes, , Major Modes, emacs, The GNU Emacs
1220 Manual}, for more information about Emacs's modes.)
1223 @section Run a Program
1224 @cindex Run a program
1225 @cindex Program, running one
1227 @cindex @samp{evaluate} defined
1228 A list in Lisp---any list---is a program ready to run. If you run it
1229 (for which the Lisp jargon is @dfn{evaluate}), the computer will do one
1230 of three things: do nothing except return to you the list itself; send
1231 you an error message; or, treat the first symbol in the list as a
1232 command to do something. (Usually, of course, it is the last of these
1233 three things that you really want!)
1235 @c use code for the single apostrophe, not samp.
1236 The single apostrophe, @code{'}, that I put in front of some of the
1237 example lists in preceding sections is called a @dfn{quote}; when it
1238 precedes a list, it tells Lisp to do nothing with the list, other than
1239 take it as it is written. But if there is no quote preceding a list,
1240 the first item of the list is special: it is a command for the computer
1241 to obey. (In Lisp, these commands are called @emph{functions}.) The list
1242 @code{(+ 2 2)} shown above did not have a quote in front of it, so Lisp
1243 understands that the @code{+} is an instruction to do something with the
1244 rest of the list: add the numbers that follow.
1247 If you are reading this inside of GNU Emacs in Info, here is how you can
1248 evaluate such a list: place your cursor immediately after the right
1249 hand parenthesis of the following list and then type @kbd{C-x C-e}:
1255 @c use code for the number four, not samp.
1257 You will see the number @code{4} appear in the echo area. (In the
1258 jargon, what you have just done is ``evaluate the list.'' The echo area
1259 is the line at the bottom of the screen that displays or ``echoes''
1260 text.) Now try the same thing with a quoted list: place the cursor
1261 right after the following list and type @kbd{C-x C-e}:
1264 '(this is a quoted list)
1268 You will see @code{(this is a quoted list)} appear in the echo area.
1270 @cindex Lisp interpreter, explained
1271 @cindex Interpreter, Lisp, explained
1272 In both cases, what you are doing is giving a command to the program
1273 inside of GNU Emacs called the @dfn{Lisp interpreter}---giving the
1274 interpreter a command to evaluate the expression. The name of the Lisp
1275 interpreter comes from the word for the task done by a human who comes
1276 up with the meaning of an expression---who ``interprets'' it.
1278 You can also evaluate an atom that is not part of a list---one that is
1279 not surrounded by parentheses; again, the Lisp interpreter translates
1280 from the humanly readable expression to the language of the computer.
1281 But before discussing this (@pxref{Variables}), we will discuss what the
1282 Lisp interpreter does when you make an error.
1285 @section Generate an Error Message
1286 @cindex Generate an error message
1287 @cindex Error message generation
1289 Partly so you won't worry if you do it accidentally, we will now give
1290 a command to the Lisp interpreter that generates an error message.
1291 This is a harmless activity; and indeed, we will often try to generate
1292 error messages intentionally. Once you understand the jargon, error
1293 messages can be informative. Instead of being called ``error''
1294 messages, they should be called ``help'' messages. They are like
1295 signposts to a traveler in a strange country; deciphering them can be
1296 hard, but once understood, they can point the way.
1298 The error message is generated by a built-in GNU Emacs debugger. We
1299 will `enter the debugger'. You get out of the debugger by typing @code{q}.
1301 What we will do is evaluate a list that is not quoted and does not
1302 have a meaningful command as its first element. Here is a list almost
1303 exactly the same as the one we just used, but without the single-quote
1304 in front of it. Position the cursor right after it and type @kbd{C-x
1308 (this is an unquoted list)
1313 What you see depends on which version of Emacs you are running. GNU
1314 Emacs version 22 provides more information than version 20 and before.
1315 First, the more recent result of generating an error; then the
1316 earlier, version 20 result.
1320 In GNU Emacs version 22, a @file{*Backtrace*} window will open up and
1321 you will see the following in it:
1324 A @file{*Backtrace*} window will open up and you should see the
1329 ---------- Buffer: *Backtrace* ----------
1330 Debugger entered--Lisp error: (void-function this)
1331 (this is an unquoted list)
1332 eval((this is an unquoted list))
1333 eval-last-sexp-1(nil)
1335 call-interactively(eval-last-sexp)
1336 ---------- Buffer: *Backtrace* ----------
1342 Your cursor will be in this window (you may have to wait a few seconds
1343 before it becomes visible). To quit the debugger and make the
1344 debugger window go away, type:
1351 Please type @kbd{q} right now, so you become confident that you can
1352 get out of the debugger. Then, type @kbd{C-x C-e} again to re-enter
1355 @cindex @samp{function} defined
1356 Based on what we already know, we can almost read this error message.
1358 You read the @file{*Backtrace*} buffer from the bottom up; it tells
1359 you what Emacs did. When you typed @kbd{C-x C-e}, you made an
1360 interactive call to the command @code{eval-last-sexp}. @code{eval} is
1361 an abbreviation for `evaluate' and @code{sexp} is an abbreviation for
1362 `symbolic expression'. The command means `evaluate last symbolic
1363 expression', which is the expression just before your cursor.
1365 Each line above tells you what the Lisp interpreter evaluated next.
1366 The most recent action is at the top. The buffer is called the
1367 @file{*Backtrace*} buffer because it enables you to track Emacs
1371 At the top of the @file{*Backtrace*} buffer, you see the line:
1374 Debugger entered--Lisp error: (void-function this)
1378 The Lisp interpreter tried to evaluate the first atom of the list, the
1379 word @samp{this}. It is this action that generated the error message
1380 @samp{void-function this}.
1382 The message contains the words @samp{void-function} and @samp{this}.
1384 @cindex @samp{function} defined
1385 The word @samp{function} was mentioned once before. It is a very
1386 important word. For our purposes, we can define it by saying that a
1387 @dfn{function} is a set of instructions to the computer that tell the
1388 computer to do something.
1390 Now we can begin to understand the error message: @samp{void-function
1391 this}. The function (that is, the word @samp{this}) does not have a
1392 definition of any set of instructions for the computer to carry out.
1394 The slightly odd word, @samp{void-function}, is designed to cover the
1395 way Emacs Lisp is implemented, which is that when a symbol does not
1396 have a function definition attached to it, the place that should
1397 contain the instructions is `void'.
1399 On the other hand, since we were able to add 2 plus 2 successfully, by
1400 evaluating @code{(+ 2 2)}, we can infer that the symbol @code{+} must
1401 have a set of instructions for the computer to obey and those
1402 instructions must be to add the numbers that follow the @code{+}.
1404 It is possible to prevent Emacs entering the debugger in cases like
1405 this. We do not explain how to do that here, but we will mention what
1406 the result looks like, because you may encounter a similar situation
1407 if there is a bug in some Emacs code that you are using. In such
1408 cases, you will see only one line of error message; it will appear in
1409 the echo area and look like this:
1412 Symbol's function definition is void:@: this
1417 (Also, your terminal may beep at you---some do, some don't; and others
1418 blink. This is just a device to get your attention.)
1420 The message goes away as soon as you type a key, even just to
1423 We know the meaning of the word @samp{Symbol}. It refers to the first
1424 atom of the list, the word @samp{this}. The word @samp{function}
1425 refers to the instructions that tell the computer what to do.
1426 (Technically, the symbol tells the computer where to find the
1427 instructions, but this is a complication we can ignore for the
1430 The error message can be understood: @samp{Symbol's function
1431 definition is void:@: this}. The symbol (that is, the word
1432 @samp{this}) lacks instructions for the computer to carry out.
1434 @node Names & Definitions
1435 @section Symbol Names and Function Definitions
1436 @cindex Symbol names
1438 We can articulate another characteristic of Lisp based on what we have
1439 discussed so far---an important characteristic: a symbol, like
1440 @code{+}, is not itself the set of instructions for the computer to
1441 carry out. Instead, the symbol is used, perhaps temporarily, as a way
1442 of locating the definition or set of instructions. What we see is the
1443 name through which the instructions can be found. Names of people
1444 work the same way. I can be referred to as @samp{Bob}; however, I am
1445 not the letters @samp{B}, @samp{o}, @samp{b} but am, or was, the
1446 consciousness consistently associated with a particular life-form.
1447 The name is not me, but it can be used to refer to me.
1449 In Lisp, one set of instructions can be attached to several names.
1450 For example, the computer instructions for adding numbers can be
1451 linked to the symbol @code{plus} as well as to the symbol @code{+}
1452 (and are in some dialects of Lisp). Among humans, I can be referred
1453 to as @samp{Robert} as well as @samp{Bob} and by other words as well.
1455 On the other hand, a symbol can have only one function definition
1456 attached to it at a time. Otherwise, the computer would be confused as
1457 to which definition to use. If this were the case among people, only
1458 one person in the world could be named @samp{Bob}. However, the function
1459 definition to which the name refers can be changed readily.
1460 (@xref{Install, , Install a Function Definition}.)
1462 Since Emacs Lisp is large, it is customary to name symbols in a way
1463 that identifies the part of Emacs to which the function belongs.
1464 Thus, all the names for functions that deal with Texinfo start with
1465 @samp{texinfo-} and those for functions that deal with reading mail
1466 start with @samp{rmail-}.
1468 @node Lisp Interpreter
1469 @section The Lisp Interpreter
1470 @cindex Lisp interpreter, what it does
1471 @cindex Interpreter, what it does
1473 Based on what we have seen, we can now start to figure out what the
1474 Lisp interpreter does when we command it to evaluate a list.
1475 First, it looks to see whether there is a quote before the list; if
1476 there is, the interpreter just gives us the list. On the other
1477 hand, if there is no quote, the interpreter looks at the first element
1478 in the list and sees whether it has a function definition. If it does,
1479 the interpreter carries out the instructions in the function definition.
1480 Otherwise, the interpreter prints an error message.
1482 This is how Lisp works. Simple. There are added complications which we
1483 will get to in a minute, but these are the fundamentals. Of course, to
1484 write Lisp programs, you need to know how to write function definitions
1485 and attach them to names, and how to do this without confusing either
1486 yourself or the computer.
1489 * Complications:: Variables, Special forms, Lists within.
1490 * Byte Compiling:: Specially processing code for speed.
1495 @unnumberedsubsec Complications
1498 Now, for the first complication. In addition to lists, the Lisp
1499 interpreter can evaluate a symbol that is not quoted and does not have
1500 parentheses around it. The Lisp interpreter will attempt to determine
1501 the symbol's value as a @dfn{variable}. This situation is described
1502 in the section on variables. (@xref{Variables}.)
1504 @cindex Special form
1505 The second complication occurs because some functions are unusual and
1506 do not work in the usual manner. Those that don't are called
1507 @dfn{special forms}. They are used for special jobs, like defining a
1508 function, and there are not many of them. In the next few chapters,
1509 you will be introduced to several of the more important special forms.
1511 As well as special forms, there are also @dfn{macros}. A macro
1512 is a construct defined in Lisp, which differs from a function in that it
1513 translates a Lisp expression into another expression that is to be
1514 evaluated in place of the original expression. (@xref{Lisp macro}.)
1516 For the purposes of this introduction, you do not need to worry too much
1517 about whether something is a special form, macro, or ordinary function.
1518 For example, @code{if} is a special form (@pxref{if}), but @code{when}
1519 is a macro (@pxref{Lisp macro}). In earlier versions of Emacs,
1520 @code{defun} was a special form, but now it is a macro (@pxref{defun}).
1521 It still behaves in the same way.
1523 The final complication is this: if the function that the
1524 Lisp interpreter is looking at is not a special form, and if it is part
1525 of a list, the Lisp interpreter looks to see whether the list has a list
1526 inside of it. If there is an inner list, the Lisp interpreter first
1527 figures out what it should do with the inside list, and then it works on
1528 the outside list. If there is yet another list embedded inside the
1529 inner list, it works on that one first, and so on. It always works on
1530 the innermost list first. The interpreter works on the innermost list
1531 first, to evaluate the result of that list. The result may be
1532 used by the enclosing expression.
1534 Otherwise, the interpreter works left to right, from one expression to
1537 @node Byte Compiling
1538 @subsection Byte Compiling
1539 @cindex Byte compiling
1541 One other aspect of interpreting: the Lisp interpreter is able to
1542 interpret two kinds of entity: humanly readable code, on which we will
1543 focus exclusively, and specially processed code, called @dfn{byte
1544 compiled} code, which is not humanly readable. Byte compiled code
1545 runs faster than humanly readable code.
1547 You can transform humanly readable code into byte compiled code by
1548 running one of the compile commands such as @code{byte-compile-file}.
1549 Byte compiled code is usually stored in a file that ends with a
1550 @file{.elc} extension rather than a @file{.el} extension. You will
1551 see both kinds of file in the @file{emacs/lisp} directory; the files
1552 to read are those with @file{.el} extensions.
1554 As a practical matter, for most things you might do to customize or
1555 extend Emacs, you do not need to byte compile; and I will not discuss
1556 the topic here. @xref{Byte Compilation, , Byte Compilation, elisp,
1557 The GNU Emacs Lisp Reference Manual}, for a full description of byte
1564 When the Lisp interpreter works on an expression, the term for the
1565 activity is called @dfn{evaluation}. We say that the interpreter
1566 `evaluates the expression'. I've used this term several times before.
1567 The word comes from its use in everyday language, `to ascertain the
1568 value or amount of; to appraise', according to @cite{Webster's New
1569 Collegiate Dictionary}.
1572 * How the Interpreter Acts:: Returns and Side Effects...
1573 * Evaluating Inner Lists:: Lists within lists...
1577 @node How the Interpreter Acts
1578 @unnumberedsubsec How the Lisp Interpreter Acts
1581 @cindex @samp{returned value} explained
1582 After evaluating an expression, the Lisp interpreter will most likely
1583 @dfn{return} the value that the computer produces by carrying out the
1584 instructions it found in the function definition, or perhaps it will
1585 give up on that function and produce an error message. (The interpreter
1586 may also find itself tossed, so to speak, to a different function or it
1587 may attempt to repeat continually what it is doing for ever and ever in
1588 what is called an `infinite loop'. These actions are less common; and
1589 we can ignore them.) Most frequently, the interpreter returns a value.
1591 @cindex @samp{side effect} defined
1592 At the same time the interpreter returns a value, it may do something
1593 else as well, such as move a cursor or copy a file; this other kind of
1594 action is called a @dfn{side effect}. Actions that we humans think are
1595 important, such as printing results, are often ``side effects'' to the
1596 Lisp interpreter. The jargon can sound peculiar, but it turns out that
1597 it is fairly easy to learn to use side effects.
1599 In summary, evaluating a symbolic expression most commonly causes the
1600 Lisp interpreter to return a value and perhaps carry out a side effect;
1601 or else produce an error.
1603 @node Evaluating Inner Lists
1604 @subsection Evaluating Inner Lists
1605 @cindex Inner list evaluation
1606 @cindex Evaluating inner lists
1608 If evaluation applies to a list that is inside another list, the outer
1609 list may use the value returned by the first evaluation as information
1610 when the outer list is evaluated. This explains why inner expressions
1611 are evaluated first: the values they return are used by the outer
1615 We can investigate this process by evaluating another addition example.
1616 Place your cursor after the following expression and type @kbd{C-x C-e}:
1623 The number 8 will appear in the echo area.
1625 What happens is that the Lisp interpreter first evaluates the inner
1626 expression, @code{(+ 3 3)}, for which the value 6 is returned; then it
1627 evaluates the outer expression as if it were written @code{(+ 2 6)}, which
1628 returns the value 8. Since there are no more enclosing expressions to
1629 evaluate, the interpreter prints that value in the echo area.
1631 Now it is easy to understand the name of the command invoked by the
1632 keystrokes @kbd{C-x C-e}: the name is @code{eval-last-sexp}. The
1633 letters @code{sexp} are an abbreviation for `symbolic expression', and
1634 @code{eval} is an abbreviation for `evaluate'. The command means
1635 `evaluate last symbolic expression'.
1637 As an experiment, you can try evaluating the expression by putting the
1638 cursor at the beginning of the next line immediately following the
1639 expression, or inside the expression.
1642 Here is another copy of the expression:
1649 If you place the cursor at the beginning of the blank line that
1650 immediately follows the expression and type @kbd{C-x C-e}, you will
1651 still get the value 8 printed in the echo area. Now try putting the
1652 cursor inside the expression. If you put it right after the next to
1653 last parenthesis (so it appears to sit on top of the last parenthesis),
1654 you will get a 6 printed in the echo area! This is because the command
1655 evaluates the expression @code{(+ 3 3)}.
1657 Now put the cursor immediately after a number. Type @kbd{C-x C-e} and
1658 you will get the number itself. In Lisp, if you evaluate a number, you
1659 get the number itself---this is how numbers differ from symbols. If you
1660 evaluate a list starting with a symbol like @code{+}, you will get a
1661 value returned that is the result of the computer carrying out the
1662 instructions in the function definition attached to that name. If a
1663 symbol by itself is evaluated, something different happens, as we will
1664 see in the next section.
1670 In Emacs Lisp, a symbol can have a value attached to it just as it can
1671 have a function definition attached to it. The two are different.
1672 The function definition is a set of instructions that a computer will
1673 obey. A value, on the other hand, is something, such as number or a
1674 name, that can vary (which is why such a symbol is called a variable).
1675 The value of a symbol can be any expression in Lisp, such as a symbol,
1676 number, list, or string. A symbol that has a value is often called a
1679 A symbol can have both a function definition and a value attached to
1680 it at the same time. Or it can have just one or the other.
1681 The two are separate. This is somewhat similar
1682 to the way the name Cambridge can refer to the city in Massachusetts
1683 and have some information attached to the name as well, such as
1684 ``great programming center''.
1687 (Incidentally, in Emacs Lisp, a symbol can have two
1688 other things attached to it, too: a property list and a documentation
1689 string; these are discussed later.)
1692 Another way to think about this is to imagine a symbol as being a chest
1693 of drawers. The function definition is put in one drawer, the value in
1694 another, and so on. What is put in the drawer holding the value can be
1695 changed without affecting the contents of the drawer holding the
1696 function definition, and vice-verse.
1699 * fill-column Example::
1700 * Void Function:: The error message for a symbol
1702 * Void Variable:: The error message for a symbol without a value.
1706 @node fill-column Example
1707 @unnumberedsubsec @code{fill-column}, an Example Variable
1710 @findex fill-column, @r{an example variable}
1711 @cindex Example variable, @code{fill-column}
1712 @cindex Variable, example of, @code{fill-column}
1713 The variable @code{fill-column} illustrates a symbol with a value
1714 attached to it: in every GNU Emacs buffer, this symbol is set to some
1715 value, usually 72 or 70, but sometimes to some other value. To find the
1716 value of this symbol, evaluate it by itself. If you are reading this in
1717 Info inside of GNU Emacs, you can do this by putting the cursor after
1718 the symbol and typing @kbd{C-x C-e}:
1725 After I typed @kbd{C-x C-e}, Emacs printed the number 72 in my echo
1726 area. This is the value for which @code{fill-column} is set for me as I
1727 write this. It may be different for you in your Info buffer. Notice
1728 that the value returned as a variable is printed in exactly the same way
1729 as the value returned by a function carrying out its instructions. From
1730 the point of view of the Lisp interpreter, a value returned is a value
1731 returned. What kind of expression it came from ceases to matter once
1734 A symbol can have any value attached to it or, to use the jargon, we can
1735 @dfn{bind} the variable to a value: to a number, such as 72; to a
1736 string, @code{"such as this"}; to a list, such as @code{(spruce pine
1737 oak)}; we can even bind a variable to a function definition.
1739 A symbol can be bound to a value in several ways. @xref{set & setq, ,
1740 Setting the Value of a Variable}, for information about one way to do
1744 @subsection Error Message for a Symbol Without a Function
1745 @cindex Symbol without function error
1746 @cindex Error for symbol without function
1748 When we evaluated @code{fill-column} to find its value as a variable,
1749 we did not place parentheses around the word. This is because we did
1750 not intend to use it as a function name.
1752 If @code{fill-column} were the first or only element of a list, the
1753 Lisp interpreter would attempt to find the function definition
1754 attached to it. But @code{fill-column} has no function definition.
1755 Try evaluating this:
1763 You will create a @file{*Backtrace*} buffer that says:
1767 ---------- Buffer: *Backtrace* ----------
1768 Debugger entered--Lisp error: (void-function fill-column)
1771 eval-last-sexp-1(nil)
1773 call-interactively(eval-last-sexp)
1774 ---------- Buffer: *Backtrace* ----------
1779 (Remember, to quit the debugger and make the debugger window go away,
1780 type @kbd{q} in the @file{*Backtrace*} buffer.)
1784 In GNU Emacs 20 and before, you will produce an error message that says:
1787 Symbol's function definition is void:@: fill-column
1791 (The message will go away as soon as you move the cursor or type
1796 @subsection Error Message for a Symbol Without a Value
1797 @cindex Symbol without value error
1798 @cindex Error for symbol without value
1800 If you attempt to evaluate a symbol that does not have a value bound to
1801 it, you will receive an error message. You can see this by
1802 experimenting with our 2 plus 2 addition. In the following expression,
1803 put your cursor right after the @code{+}, before the first number 2,
1812 In GNU Emacs 22, you will create a @file{*Backtrace*} buffer that
1817 ---------- Buffer: *Backtrace* ----------
1818 Debugger entered--Lisp error: (void-variable +)
1820 eval-last-sexp-1(nil)
1822 call-interactively(eval-last-sexp)
1823 ---------- Buffer: *Backtrace* ----------
1828 (Again, you can quit the debugger by
1829 typing @kbd{q} in the @file{*Backtrace*} buffer.)
1831 This backtrace is different from the very first error message we saw,
1832 which said, @samp{Debugger entered--Lisp error: (void-function this)}.
1833 In this case, the function does not have a value as a variable; while
1834 in the other error message, the function (the word `this') did not
1837 In this experiment with the @code{+}, what we did was cause the Lisp
1838 interpreter to evaluate the @code{+} and look for the value of the
1839 variable instead of the function definition. We did this by placing the
1840 cursor right after the symbol rather than after the parenthesis of the
1841 enclosing list as we did before. As a consequence, the Lisp interpreter
1842 evaluated the preceding s-expression, which in this case was
1845 Since @code{+} does not have a value bound to it, just the function
1846 definition, the error message reported that the symbol's value as a
1851 In GNU Emacs version 20 and before, your error message will say:
1854 Symbol's value as variable is void:@: +
1858 The meaning is the same as in GNU Emacs 22.
1864 @cindex Passing information to functions
1866 To see how information is passed to functions, let's look again at
1867 our old standby, the addition of two plus two. In Lisp, this is written
1874 If you evaluate this expression, the number 4 will appear in your echo
1875 area. What the Lisp interpreter does is add the numbers that follow
1878 @cindex @samp{argument} defined
1879 The numbers added by @code{+} are called the @dfn{arguments} of the
1880 function @code{+}. These numbers are the information that is given to
1881 or @dfn{passed} to the function.
1883 The word `argument' comes from the way it is used in mathematics and
1884 does not refer to a disputation between two people; instead it refers to
1885 the information presented to the function, in this case, to the
1886 @code{+}. In Lisp, the arguments to a function are the atoms or lists
1887 that follow the function. The values returned by the evaluation of
1888 these atoms or lists are passed to the function. Different functions
1889 require different numbers of arguments; some functions require none at
1890 all.@footnote{It is curious to track the path by which the word `argument'
1891 came to have two different meanings, one in mathematics and the other in
1892 everyday English. According to the @cite{Oxford English Dictionary},
1893 the word derives from the Latin for @samp{to make clear, prove}; thus it
1894 came to mean, by one thread of derivation, `the evidence offered as
1895 proof', which is to say, `the information offered', which led to its
1896 meaning in Lisp. But in the other thread of derivation, it came to mean
1897 `to assert in a manner against which others may make counter
1898 assertions', which led to the meaning of the word as a disputation.
1899 (Note here that the English word has two different definitions attached
1900 to it at the same time. By contrast, in Emacs Lisp, a symbol cannot
1901 have two different function definitions at the same time.)}
1904 * Data types:: Types of data passed to a function.
1905 * Args as Variable or List:: An argument can be the value
1906 of a variable or list.
1907 * Variable Number of Arguments:: Some functions may take a
1908 variable number of arguments.
1909 * Wrong Type of Argument:: Passing an argument of the wrong type
1911 * message:: A useful function for sending messages.
1915 @subsection Arguments' Data Types
1917 @cindex Types of data
1918 @cindex Arguments' data types
1920 The type of data that should be passed to a function depends on what
1921 kind of information it uses. The arguments to a function such as
1922 @code{+} must have values that are numbers, since @code{+} adds numbers.
1923 Other functions use different kinds of data for their arguments.
1927 For example, the @code{concat} function links together or unites two or
1928 more strings of text to produce a string. The arguments are strings.
1929 Concatenating the two character strings @code{abc}, @code{def} produces
1930 the single string @code{abcdef}. This can be seen by evaluating the
1934 (concat "abc" "def")
1938 The value produced by evaluating this expression is @code{"abcdef"}.
1940 A function such as @code{substring} uses both a string and numbers as
1941 arguments. The function returns a part of the string, a substring of
1942 the first argument. This function takes three arguments. Its first
1943 argument is the string of characters, the second and third arguments are
1944 numbers that indicate the beginning and end of the substring. The
1945 numbers are a count of the number of characters (including spaces and
1946 punctuation) from the beginning of the string.
1949 For example, if you evaluate the following:
1952 (substring "The quick brown fox jumped." 16 19)
1956 you will see @code{"fox"} appear in the echo area. The arguments are the
1957 string and the two numbers.
1959 Note that the string passed to @code{substring} is a single atom even
1960 though it is made up of several words separated by spaces. Lisp counts
1961 everything between the two quotation marks as part of the string,
1962 including the spaces. You can think of the @code{substring} function as
1963 a kind of `atom smasher' since it takes an otherwise indivisible atom
1964 and extracts a part. However, @code{substring} is only able to extract
1965 a substring from an argument that is a string, not from another type of
1966 atom such as a number or symbol.
1968 @node Args as Variable or List
1969 @subsection An Argument as the Value of a Variable or List
1971 An argument can be a symbol that returns a value when it is evaluated.
1972 For example, when the symbol @code{fill-column} by itself is evaluated,
1973 it returns a number. This number can be used in an addition.
1976 Position the cursor after the following expression and type @kbd{C-x
1984 The value will be a number two more than what you get by evaluating
1985 @code{fill-column} alone. For me, this is 74, because my value of
1986 @code{fill-column} is 72.
1988 As we have just seen, an argument can be a symbol that returns a value
1989 when evaluated. In addition, an argument can be a list that returns a
1990 value when it is evaluated. For example, in the following expression,
1991 the arguments to the function @code{concat} are the strings
1992 @w{@code{"The "}} and @w{@code{" red foxes."}} and the list
1993 @code{(number-to-string (+ 2 fill-column))}.
1995 @c For GNU Emacs 22, need number-to-string
1997 (concat "The " (number-to-string (+ 2 fill-column)) " red foxes.")
2001 If you evaluate this expression---and if, as with my Emacs,
2002 @code{fill-column} evaluates to 72---@code{"The 74 red foxes."} will
2003 appear in the echo area. (Note that you must put spaces after the
2004 word @samp{The} and before the word @samp{red} so they will appear in
2005 the final string. The function @code{number-to-string} converts the
2006 integer that the addition function returns to a string.
2007 @code{number-to-string} is also known as @code{int-to-string}.)
2009 @node Variable Number of Arguments
2010 @subsection Variable Number of Arguments
2011 @cindex Variable number of arguments
2012 @cindex Arguments, variable number of
2014 Some functions, such as @code{concat}, @code{+} or @code{*}, take any
2015 number of arguments. (The @code{*} is the symbol for multiplication.)
2016 This can be seen by evaluating each of the following expressions in
2017 the usual way. What you will see in the echo area is printed in this
2018 text after @samp{@result{}}, which you may read as `evaluates to'.
2021 In the first set, the functions have no arguments:
2032 In this set, the functions have one argument each:
2043 In this set, the functions have three arguments each:
2047 (+ 3 4 5) @result{} 12
2049 (* 3 4 5) @result{} 60
2053 @node Wrong Type of Argument
2054 @subsection Using the Wrong Type Object as an Argument
2055 @cindex Wrong type of argument
2056 @cindex Argument, wrong type of
2058 When a function is passed an argument of the wrong type, the Lisp
2059 interpreter produces an error message. For example, the @code{+}
2060 function expects the values of its arguments to be numbers. As an
2061 experiment we can pass it the quoted symbol @code{hello} instead of a
2062 number. Position the cursor after the following expression and type
2070 When you do this you will generate an error message. What has happened
2071 is that @code{+} has tried to add the 2 to the value returned by
2072 @code{'hello}, but the value returned by @code{'hello} is the symbol
2073 @code{hello}, not a number. Only numbers can be added. So @code{+}
2074 could not carry out its addition.
2077 You will create and enter a @file{*Backtrace*} buffer that says:
2082 ---------- Buffer: *Backtrace* ----------
2083 Debugger entered--Lisp error:
2084 (wrong-type-argument number-or-marker-p hello)
2086 eval((+ 2 (quote hello)))
2087 eval-last-sexp-1(nil)
2089 call-interactively(eval-last-sexp)
2090 ---------- Buffer: *Backtrace* ----------
2095 As usual, the error message tries to be helpful and makes sense after you
2096 learn how to read it.@footnote{@code{(quote hello)} is an expansion of
2097 the abbreviation @code{'hello}.}
2099 The first part of the error message is straightforward; it says
2100 @samp{wrong type argument}. Next comes the mysterious jargon word
2101 @w{@samp{number-or-marker-p}}. This word is trying to tell you what
2102 kind of argument the @code{+} expected.
2104 The symbol @code{number-or-marker-p} says that the Lisp interpreter is
2105 trying to determine whether the information presented it (the value of
2106 the argument) is a number or a marker (a special object representing a
2107 buffer position). What it does is test to see whether the @code{+} is
2108 being given numbers to add. It also tests to see whether the
2109 argument is something called a marker, which is a specific feature of
2110 Emacs Lisp. (In Emacs, locations in a buffer are recorded as markers.
2111 When the mark is set with the @kbd{C-@@} or @kbd{C-@key{SPC}} command,
2112 its position is kept as a marker. The mark can be considered a
2113 number---the number of characters the location is from the beginning
2114 of the buffer.) In Emacs Lisp, @code{+} can be used to add the
2115 numeric value of marker positions as numbers.
2117 The @samp{p} of @code{number-or-marker-p} is the embodiment of a
2118 practice started in the early days of Lisp programming. The @samp{p}
2119 stands for `predicate'. In the jargon used by the early Lisp
2120 researchers, a predicate refers to a function to determine whether some
2121 property is true or false. So the @samp{p} tells us that
2122 @code{number-or-marker-p} is the name of a function that determines
2123 whether it is true or false that the argument supplied is a number or
2124 a marker. Other Lisp symbols that end in @samp{p} include @code{zerop},
2125 a function that tests whether its argument has the value of zero, and
2126 @code{listp}, a function that tests whether its argument is a list.
2128 Finally, the last part of the error message is the symbol @code{hello}.
2129 This is the value of the argument that was passed to @code{+}. If the
2130 addition had been passed the correct type of object, the value passed
2131 would have been a number, such as 37, rather than a symbol like
2132 @code{hello}. But then you would not have got the error message.
2136 In GNU Emacs version 20 and before, the echo area displays an error
2140 Wrong type argument:@: number-or-marker-p, hello
2143 This says, in different words, the same as the top line of the
2144 @file{*Backtrace*} buffer.
2148 @subsection The @code{message} Function
2151 Like @code{+}, the @code{message} function takes a variable number of
2152 arguments. It is used to send messages to the user and is so useful
2153 that we will describe it here.
2156 A message is printed in the echo area. For example, you can print a
2157 message in your echo area by evaluating the following list:
2160 (message "This message appears in the echo area!")
2163 The whole string between double quotation marks is a single argument
2164 and is printed @i{in toto}. (Note that in this example, the message
2165 itself will appear in the echo area within double quotes; that is
2166 because you see the value returned by the @code{message} function. In
2167 most uses of @code{message} in programs that you write, the text will
2168 be printed in the echo area as a side-effect, without the quotes.
2169 @xref{multiply-by-seven in detail, , @code{multiply-by-seven} in
2170 detail}, for an example of this.)
2172 However, if there is a @samp{%s} in the quoted string of characters, the
2173 @code{message} function does not print the @samp{%s} as such, but looks
2174 to the argument that follows the string. It evaluates the second
2175 argument and prints the value at the location in the string where the
2179 You can see this by positioning the cursor after the following
2180 expression and typing @kbd{C-x C-e}:
2183 (message "The name of this buffer is: %s." (buffer-name))
2187 In Info, @code{"The name of this buffer is: *info*."} will appear in the
2188 echo area. The function @code{buffer-name} returns the name of the
2189 buffer as a string, which the @code{message} function inserts in place
2192 To print a value as an integer, use @samp{%d} in the same way as
2193 @samp{%s}. For example, to print a message in the echo area that
2194 states the value of the @code{fill-column}, evaluate the following:
2197 (message "The value of fill-column is %d." fill-column)
2201 On my system, when I evaluate this list, @code{"The value of
2202 fill-column is 72."} appears in my echo area@footnote{Actually, you
2203 can use @code{%s} to print a number. It is non-specific. @code{%d}
2204 prints only the part of a number left of a decimal point, and not
2205 anything that is not a number.}.
2207 If there is more than one @samp{%s} in the quoted string, the value of
2208 the first argument following the quoted string is printed at the
2209 location of the first @samp{%s} and the value of the second argument is
2210 printed at the location of the second @samp{%s}, and so on.
2213 For example, if you evaluate the following,
2217 (message "There are %d %s in the office!"
2218 (- fill-column 14) "pink elephants")
2223 a rather whimsical message will appear in your echo area. On my system
2224 it says, @code{"There are 58 pink elephants in the office!"}.
2226 The expression @code{(- fill-column 14)} is evaluated and the resulting
2227 number is inserted in place of the @samp{%d}; and the string in double
2228 quotes, @code{"pink elephants"}, is treated as a single argument and
2229 inserted in place of the @samp{%s}. (That is to say, a string between
2230 double quotes evaluates to itself, like a number.)
2232 Finally, here is a somewhat complex example that not only illustrates
2233 the computation of a number, but also shows how you can use an
2234 expression within an expression to generate the text that is substituted
2239 (message "He saw %d %s"
2243 "The quick brown foxes jumped." 16 21)
2248 In this example, @code{message} has three arguments: the string,
2249 @code{"He saw %d %s"}, the expression, @code{(- fill-column 32)}, and
2250 the expression beginning with the function @code{concat}. The value
2251 resulting from the evaluation of @code{(- fill-column 32)} is inserted
2252 in place of the @samp{%d}; and the value returned by the expression
2253 beginning with @code{concat} is inserted in place of the @samp{%s}.
2255 When your fill column is 70 and you evaluate the expression, the
2256 message @code{"He saw 38 red foxes leaping."} appears in your echo
2260 @section Setting the Value of a Variable
2261 @cindex Variable, setting value
2262 @cindex Setting value of variable
2264 @cindex @samp{bind} defined
2265 There are several ways by which a variable can be given a value. One of
2266 the ways is to use either the function @code{set} or the function
2267 @code{setq}. Another way is to use @code{let} (@pxref{let}). (The
2268 jargon for this process is to @dfn{bind} a variable to a value.)
2270 The following sections not only describe how @code{set} and @code{setq}
2271 work but also illustrate how arguments are passed.
2274 * Using set:: Setting values.
2275 * Using setq:: Setting a quoted value.
2276 * Counting:: Using @code{setq} to count.
2280 @subsection Using @code{set}
2283 To set the value of the symbol @code{flowers} to the list @code{'(rose
2284 violet daisy buttercup)}, evaluate the following expression by
2285 positioning the cursor after the expression and typing @kbd{C-x C-e}.
2288 (set 'flowers '(rose violet daisy buttercup))
2292 The list @code{(rose violet daisy buttercup)} will appear in the echo
2293 area. This is what is @emph{returned} by the @code{set} function. As a
2294 side effect, the symbol @code{flowers} is bound to the list; that is,
2295 the symbol @code{flowers}, which can be viewed as a variable, is given
2296 the list as its value. (This process, by the way, illustrates how a
2297 side effect to the Lisp interpreter, setting the value, can be the
2298 primary effect that we humans are interested in. This is because every
2299 Lisp function must return a value if it does not get an error, but it
2300 will only have a side effect if it is designed to have one.)
2302 After evaluating the @code{set} expression, you can evaluate the symbol
2303 @code{flowers} and it will return the value you just set. Here is the
2304 symbol. Place your cursor after it and type @kbd{C-x C-e}.
2311 When you evaluate @code{flowers}, the list
2312 @code{(rose violet daisy buttercup)} appears in the echo area.
2314 Incidentally, if you evaluate @code{'flowers}, the variable with a quote
2315 in front of it, what you will see in the echo area is the symbol itself,
2316 @code{flowers}. Here is the quoted symbol, so you can try this:
2322 Note also, that when you use @code{set}, you need to quote both
2323 arguments to @code{set}, unless you want them evaluated. Since we do
2324 not want either argument evaluated, neither the variable
2325 @code{flowers} nor the list @code{(rose violet daisy buttercup)}, both
2326 are quoted. (When you use @code{set} without quoting its first
2327 argument, the first argument is evaluated before anything else is
2328 done. If you did this and @code{flowers} did not have a value
2329 already, you would get an error message that the @samp{Symbol's value
2330 as variable is void}; on the other hand, if @code{flowers} did return
2331 a value after it was evaluated, the @code{set} would attempt to set
2332 the value that was returned. There are situations where this is the
2333 right thing for the function to do; but such situations are rare.)
2336 @subsection Using @code{setq}
2339 As a practical matter, you almost always quote the first argument to
2340 @code{set}. The combination of @code{set} and a quoted first argument
2341 is so common that it has its own name: the special form @code{setq}.
2342 This special form is just like @code{set} except that the first argument
2343 is quoted automatically, so you don't need to type the quote mark
2344 yourself. Also, as an added convenience, @code{setq} permits you to set
2345 several different variables to different values, all in one expression.
2347 To set the value of the variable @code{carnivores} to the list
2348 @code{'(lion tiger leopard)} using @code{setq}, the following expression
2352 (setq carnivores '(lion tiger leopard))
2356 This is exactly the same as using @code{set} except the first argument
2357 is automatically quoted by @code{setq}. (The @samp{q} in @code{setq}
2358 means @code{quote}.)
2361 With @code{set}, the expression would look like this:
2364 (set 'carnivores '(lion tiger leopard))
2367 Also, @code{setq} can be used to assign different values to
2368 different variables. The first argument is bound to the value
2369 of the second argument, the third argument is bound to the value of the
2370 fourth argument, and so on. For example, you could use the following to
2371 assign a list of trees to the symbol @code{trees} and a list of herbivores
2372 to the symbol @code{herbivores}:
2376 (setq trees '(pine fir oak maple)
2377 herbivores '(gazelle antelope zebra))
2382 (The expression could just as well have been on one line, but it might
2383 not have fit on a page; and humans find it easier to read nicely
2386 Although I have been using the term `assign', there is another way of
2387 thinking about the workings of @code{set} and @code{setq}; and that is to
2388 say that @code{set} and @code{setq} make the symbol @emph{point} to the
2389 list. This latter way of thinking is very common and in forthcoming
2390 chapters we shall come upon at least one symbol that has `pointer' as
2391 part of its name. The name is chosen because the symbol has a value,
2392 specifically a list, attached to it; or, expressed another way,
2393 the symbol is set to ``point'' to the list.
2396 @subsection Counting
2399 Here is an example that shows how to use @code{setq} in a counter. You
2400 might use this to count how many times a part of your program repeats
2401 itself. First set a variable to zero; then add one to the number each
2402 time the program repeats itself. To do this, you need a variable that
2403 serves as a counter, and two expressions: an initial @code{setq}
2404 expression that sets the counter variable to zero; and a second
2405 @code{setq} expression that increments the counter each time it is
2410 (setq counter 0) ; @r{Let's call this the initializer.}
2412 (setq counter (+ counter 1)) ; @r{This is the incrementer.}
2414 counter ; @r{This is the counter.}
2419 (The text following the @samp{;} are comments. @xref{Change a
2420 defun, , Change a Function Definition}.)
2422 If you evaluate the first of these expressions, the initializer,
2423 @code{(setq counter 0)}, and then evaluate the third expression,
2424 @code{counter}, the number @code{0} will appear in the echo area. If
2425 you then evaluate the second expression, the incrementer, @code{(setq
2426 counter (+ counter 1))}, the counter will get the value 1. So if you
2427 again evaluate @code{counter}, the number @code{1} will appear in the
2428 echo area. Each time you evaluate the second expression, the value of
2429 the counter will be incremented.
2431 When you evaluate the incrementer, @code{(setq counter (+ counter 1))},
2432 the Lisp interpreter first evaluates the innermost list; this is the
2433 addition. In order to evaluate this list, it must evaluate the variable
2434 @code{counter} and the number @code{1}. When it evaluates the variable
2435 @code{counter}, it receives its current value. It passes this value and
2436 the number @code{1} to the @code{+} which adds them together. The sum
2437 is then returned as the value of the inner list and passed to the
2438 @code{setq} which sets the variable @code{counter} to this new value.
2439 Thus, the value of the variable, @code{counter}, is changed.
2444 Learning Lisp is like climbing a hill in which the first part is the
2445 steepest. You have now climbed the most difficult part; what remains
2446 becomes easier as you progress onwards.
2454 Lisp programs are made up of expressions, which are lists or single atoms.
2457 Lists are made up of zero or more atoms or inner lists, separated by whitespace and
2458 surrounded by parentheses. A list can be empty.
2461 Atoms are multi-character symbols, like @code{forward-paragraph}, single
2462 character symbols like @code{+}, strings of characters between double
2463 quotation marks, or numbers.
2466 A number evaluates to itself.
2469 A string between double quotes also evaluates to itself.
2472 When you evaluate a symbol by itself, its value is returned.
2475 When you evaluate a list, the Lisp interpreter looks at the first symbol
2476 in the list and then at the function definition bound to that symbol.
2477 Then the instructions in the function definition are carried out.
2480 A single quotation mark,
2487 , tells the Lisp interpreter that it should
2488 return the following expression as written, and not evaluate it as it
2489 would if the quote were not there.
2492 Arguments are the information passed to a function. The arguments to a
2493 function are computed by evaluating the rest of the elements of the list
2494 of which the function is the first element.
2497 A function always returns a value when it is evaluated (unless it gets
2498 an error); in addition, it may also carry out some action called a
2499 ``side effect''. In many cases, a function's primary purpose is to
2500 create a side effect.
2503 @node Error Message Exercises
2506 A few simple exercises:
2510 Generate an error message by evaluating an appropriate symbol that is
2511 not within parentheses.
2514 Generate an error message by evaluating an appropriate symbol that is
2515 between parentheses.
2518 Create a counter that increments by two rather than one.
2521 Write an expression that prints a message in the echo area when
2525 @node Practicing Evaluation
2526 @chapter Practicing Evaluation
2527 @cindex Practicing evaluation
2528 @cindex Evaluation practice
2530 Before learning how to write a function definition in Emacs Lisp, it is
2531 useful to spend a little time evaluating various expressions that have
2532 already been written. These expressions will be lists with the
2533 functions as their first (and often only) element. Since some of the
2534 functions associated with buffers are both simple and interesting, we
2535 will start with those. In this section, we will evaluate a few of
2536 these. In another section, we will study the code of several other
2537 buffer-related functions, to see how they were written.
2540 * How to Evaluate:: Typing editing commands or @kbd{C-x C-e}
2542 * Buffer Names:: Buffers and files are different.
2543 * Getting Buffers:: Getting a buffer itself, not merely its name.
2544 * Switching Buffers:: How to change to another buffer.
2545 * Buffer Size & Locations:: Where point is located and the size of
2547 * Evaluation Exercise::
2551 @node How to Evaluate
2552 @unnumberedsec How to Evaluate
2555 @i{Whenever you give an editing command} to Emacs Lisp, such as the
2556 command to move the cursor or to scroll the screen, @i{you are evaluating
2557 an expression,} the first element of which is a function. @i{This is
2560 @cindex @samp{interactive function} defined
2561 @cindex @samp{command} defined
2562 When you type keys, you cause the Lisp interpreter to evaluate an
2563 expression and that is how you get your results. Even typing plain text
2564 involves evaluating an Emacs Lisp function, in this case, one that uses
2565 @code{self-insert-command}, which simply inserts the character you
2566 typed. The functions you evaluate by typing keystrokes are called
2567 @dfn{interactive} functions, or @dfn{commands}; how you make a function
2568 interactive will be illustrated in the chapter on how to write function
2569 definitions. @xref{Interactive, , Making a Function Interactive}.
2571 In addition to typing keyboard commands, we have seen a second way to
2572 evaluate an expression: by positioning the cursor after a list and
2573 typing @kbd{C-x C-e}. This is what we will do in the rest of this
2574 section. There are other ways to evaluate an expression as well; these
2575 will be described as we come to them.
2577 Besides being used for practicing evaluation, the functions shown in the
2578 next few sections are important in their own right. A study of these
2579 functions makes clear the distinction between buffers and files, how to
2580 switch to a buffer, and how to determine a location within it.
2583 @section Buffer Names
2585 @findex buffer-file-name
2587 The two functions, @code{buffer-name} and @code{buffer-file-name}, show
2588 the difference between a file and a buffer. When you evaluate the
2589 following expression, @code{(buffer-name)}, the name of the buffer
2590 appears in the echo area. When you evaluate @code{(buffer-file-name)},
2591 the name of the file to which the buffer refers appears in the echo
2592 area. Usually, the name returned by @code{(buffer-name)} is the same as
2593 the name of the file to which it refers, and the name returned by
2594 @code{(buffer-file-name)} is the full path-name of the file.
2596 A file and a buffer are two different entities. A file is information
2597 recorded permanently in the computer (unless you delete it). A buffer,
2598 on the other hand, is information inside of Emacs that will vanish at
2599 the end of the editing session (or when you kill the buffer). Usually,
2600 a buffer contains information that you have copied from a file; we say
2601 the buffer is @dfn{visiting} that file. This copy is what you work on
2602 and modify. Changes to the buffer do not change the file, until you
2603 save the buffer. When you save the buffer, the buffer is copied to the file
2604 and is thus saved permanently.
2607 If you are reading this in Info inside of GNU Emacs, you can evaluate
2608 each of the following expressions by positioning the cursor after it and
2609 typing @kbd{C-x C-e}.
2620 When I do this in Info, the value returned by evaluating
2621 @code{(buffer-name)} is @file{"*info*"}, and the value returned by
2622 evaluating @code{(buffer-file-name)} is @file{nil}.
2624 On the other hand, while I am writing this document, the value
2625 returned by evaluating @code{(buffer-name)} is
2626 @file{"introduction.texinfo"}, and the value returned by evaluating
2627 @code{(buffer-file-name)} is
2628 @file{"/gnu/work/intro/introduction.texinfo"}.
2630 @cindex @code{nil}, history of word
2631 The former is the name of the buffer and the latter is the name of the
2632 file. In Info, the buffer name is @file{"*info*"}. Info does not
2633 point to any file, so the result of evaluating
2634 @code{(buffer-file-name)} is @file{nil}. The symbol @code{nil} is
2635 from the Latin word for `nothing'; in this case, it means that the
2636 buffer is not associated with any file. (In Lisp, @code{nil} is also
2637 used to mean `false' and is a synonym for the empty list, @code{()}.)
2639 When I am writing, the name of my buffer is
2640 @file{"introduction.texinfo"}. The name of the file to which it
2641 points is @file{"/gnu/work/intro/introduction.texinfo"}.
2643 (In the expressions, the parentheses tell the Lisp interpreter to
2644 treat @w{@code{buffer-name}} and @w{@code{buffer-file-name}} as
2645 functions; without the parentheses, the interpreter would attempt to
2646 evaluate the symbols as variables. @xref{Variables}.)
2648 In spite of the distinction between files and buffers, you will often
2649 find that people refer to a file when they mean a buffer and vice-verse.
2650 Indeed, most people say, ``I am editing a file,'' rather than saying,
2651 ``I am editing a buffer which I will soon save to a file.'' It is
2652 almost always clear from context what people mean. When dealing with
2653 computer programs, however, it is important to keep the distinction in mind,
2654 since the computer is not as smart as a person.
2656 @cindex Buffer, history of word
2657 The word `buffer', by the way, comes from the meaning of the word as a
2658 cushion that deadens the force of a collision. In early computers, a
2659 buffer cushioned the interaction between files and the computer's
2660 central processing unit. The drums or tapes that held a file and the
2661 central processing unit were pieces of equipment that were very
2662 different from each other, working at their own speeds, in spurts. The
2663 buffer made it possible for them to work together effectively.
2664 Eventually, the buffer grew from being an intermediary, a temporary
2665 holding place, to being the place where work is done. This
2666 transformation is rather like that of a small seaport that grew into a
2667 great city: once it was merely the place where cargo was warehoused
2668 temporarily before being loaded onto ships; then it became a business
2669 and cultural center in its own right.
2671 Not all buffers are associated with files. For example, a
2672 @file{*scratch*} buffer does not visit any file. Similarly, a
2673 @file{*Help*} buffer is not associated with any file.
2675 In the old days, when you lacked a @file{~/.emacs} file and started an
2676 Emacs session by typing the command @code{emacs} alone, without naming
2677 any files, Emacs started with the @file{*scratch*} buffer visible.
2678 Nowadays, you will see a splash screen. You can follow one of the
2679 commands suggested on the splash screen, visit a file, or press the
2680 spacebar to reach the @file{*scratch*} buffer.
2682 If you switch to the @file{*scratch*} buffer, type
2683 @code{(buffer-name)}, position the cursor after it, and then type
2684 @kbd{C-x C-e} to evaluate the expression. The name @code{"*scratch*"}
2685 will be returned and will appear in the echo area. @code{"*scratch*"}
2686 is the name of the buffer. When you type @code{(buffer-file-name)} in
2687 the @file{*scratch*} buffer and evaluate that, @code{nil} will appear
2688 in the echo area, just as it does when you evaluate
2689 @code{(buffer-file-name)} in Info.
2691 Incidentally, if you are in the @file{*scratch*} buffer and want the
2692 value returned by an expression to appear in the @file{*scratch*}
2693 buffer itself rather than in the echo area, type @kbd{C-u C-x C-e}
2694 instead of @kbd{C-x C-e}. This causes the value returned to appear
2695 after the expression. The buffer will look like this:
2698 (buffer-name)"*scratch*"
2702 You cannot do this in Info since Info is read-only and it will not allow
2703 you to change the contents of the buffer. But you can do this in any
2704 buffer you can edit; and when you write code or documentation (such as
2705 this book), this feature is very useful.
2707 @node Getting Buffers
2708 @section Getting Buffers
2709 @findex current-buffer
2710 @findex other-buffer
2711 @cindex Getting a buffer
2713 The @code{buffer-name} function returns the @emph{name} of the buffer;
2714 to get the buffer @emph{itself}, a different function is needed: the
2715 @code{current-buffer} function. If you use this function in code, what
2716 you get is the buffer itself.
2718 A name and the object or entity to which the name refers are different
2719 from each other. You are not your name. You are a person to whom
2720 others refer by name. If you ask to speak to George and someone hands you
2721 a card with the letters @samp{G}, @samp{e}, @samp{o}, @samp{r},
2722 @samp{g}, and @samp{e} written on it, you might be amused, but you would
2723 not be satisfied. You do not want to speak to the name, but to the
2724 person to whom the name refers. A buffer is similar: the name of the
2725 scratch buffer is @file{*scratch*}, but the name is not the buffer. To
2726 get a buffer itself, you need to use a function such as
2727 @code{current-buffer}.
2729 However, there is a slight complication: if you evaluate
2730 @code{current-buffer} in an expression on its own, as we will do here,
2731 what you see is a printed representation of the name of the buffer
2732 without the contents of the buffer. Emacs works this way for two
2733 reasons: the buffer may be thousands of lines long---too long to be
2734 conveniently displayed; and, another buffer may have the same contents
2735 but a different name, and it is important to distinguish between them.
2738 Here is an expression containing the function:
2745 If you evaluate this expression in Info in Emacs in the usual way,
2746 @file{#<buffer *info*>} will appear in the echo area. The special
2747 format indicates that the buffer itself is being returned, rather than
2750 Incidentally, while you can type a number or symbol into a program, you
2751 cannot do that with the printed representation of a buffer: the only way
2752 to get a buffer itself is with a function such as @code{current-buffer}.
2754 A related function is @code{other-buffer}. This returns the most
2755 recently selected buffer other than the one you are in currently, not
2756 a printed representation of its name. If you have recently switched
2757 back and forth from the @file{*scratch*} buffer, @code{other-buffer}
2758 will return that buffer.
2761 You can see this by evaluating the expression:
2768 You should see @file{#<buffer *scratch*>} appear in the echo area, or
2769 the name of whatever other buffer you switched back from most
2770 recently@footnote{Actually, by default, if the buffer from which you
2771 just switched is visible to you in another window, @code{other-buffer}
2772 will choose the most recent buffer that you cannot see; this is a
2773 subtlety that I often forget.}.
2775 @node Switching Buffers
2776 @section Switching Buffers
2777 @findex switch-to-buffer
2779 @cindex Switching to a buffer
2781 The @code{other-buffer} function actually provides a buffer when it is
2782 used as an argument to a function that requires one. We can see this
2783 by using @code{other-buffer} and @code{switch-to-buffer} to switch to a
2786 But first, a brief introduction to the @code{switch-to-buffer}
2787 function. When you switched back and forth from Info to the
2788 @file{*scratch*} buffer to evaluate @code{(buffer-name)}, you most
2789 likely typed @kbd{C-x b} and then typed @file{*scratch*}@footnote{Or
2790 rather, to save typing, you probably only typed @kbd{RET} if the
2791 default buffer was @file{*scratch*}, or if it was different, then you
2792 typed just part of the name, such as @code{*sc}, pressed your
2793 @kbd{TAB} key to cause it to expand to the full name, and then typed
2794 @kbd{RET}.} when prompted in the minibuffer for the name of
2795 the buffer to which you wanted to switch. The keystrokes, @kbd{C-x
2796 b}, cause the Lisp interpreter to evaluate the interactive function
2797 @code{switch-to-buffer}. As we said before, this is how Emacs works:
2798 different keystrokes call or run different functions. For example,
2799 @kbd{C-f} calls @code{forward-char}, @kbd{M-e} calls
2800 @code{forward-sentence}, and so on.
2802 By writing @code{switch-to-buffer} in an expression, and giving it a
2803 buffer to switch to, we can switch buffers just the way @kbd{C-x b}
2807 (switch-to-buffer (other-buffer))
2811 The symbol @code{switch-to-buffer} is the first element of the list,
2812 so the Lisp interpreter will treat it as a function and carry out the
2813 instructions that are attached to it. But before doing that, the
2814 interpreter will note that @code{other-buffer} is inside parentheses
2815 and work on that symbol first. @code{other-buffer} is the first (and
2816 in this case, the only) element of this list, so the Lisp interpreter
2817 calls or runs the function. It returns another buffer. Next, the
2818 interpreter runs @code{switch-to-buffer}, passing to it, as an
2819 argument, the other buffer, which is what Emacs will switch to. If
2820 you are reading this in Info, try this now. Evaluate the expression.
2821 (To get back, type @kbd{C-x b @key{RET}}.)@footnote{Remember, this
2822 expression will move you to your most recent other buffer that you
2823 cannot see. If you really want to go to your most recently selected
2824 buffer, even if you can still see it, you need to evaluate the
2825 following more complex expression:
2828 (switch-to-buffer (other-buffer (current-buffer) t))
2832 In this case, the first argument to @code{other-buffer} tells it which
2833 buffer to skip---the current one---and the second argument tells
2834 @code{other-buffer} it is OK to switch to a visible buffer.
2835 In regular use, @code{switch-to-buffer} takes you to an invisible
2836 window since you would most likely use @kbd{C-x o} (@code{other-window})
2837 to go to another visible buffer.}
2839 In the programming examples in later sections of this document, you will
2840 see the function @code{set-buffer} more often than
2841 @code{switch-to-buffer}. This is because of a difference between
2842 computer programs and humans: humans have eyes and expect to see the
2843 buffer on which they are working on their computer terminals. This is
2844 so obvious, it almost goes without saying. However, programs do not
2845 have eyes. When a computer program works on a buffer, that buffer does
2846 not need to be visible on the screen.
2848 @code{switch-to-buffer} is designed for humans and does two different
2849 things: it switches the buffer to which Emacs's attention is directed; and
2850 it switches the buffer displayed in the window to the new buffer.
2851 @code{set-buffer}, on the other hand, does only one thing: it switches
2852 the attention of the computer program to a different buffer. The buffer
2853 on the screen remains unchanged (of course, normally nothing happens
2854 there until the command finishes running).
2856 @cindex @samp{call} defined
2857 Also, we have just introduced another jargon term, the word @dfn{call}.
2858 When you evaluate a list in which the first symbol is a function, you
2859 are calling that function. The use of the term comes from the notion of
2860 the function as an entity that can do something for you if you `call'
2861 it---just as a plumber is an entity who can fix a leak if you call him
2864 @node Buffer Size & Locations
2865 @section Buffer Size and the Location of Point
2866 @cindex Size of buffer
2868 @cindex Point location
2869 @cindex Location of point
2871 Finally, let's look at several rather simple functions,
2872 @code{buffer-size}, @code{point}, @code{point-min}, and
2873 @code{point-max}. These give information about the size of a buffer and
2874 the location of point within it.
2876 The function @code{buffer-size} tells you the size of the current
2877 buffer; that is, the function returns a count of the number of
2878 characters in the buffer.
2885 You can evaluate this in the usual way, by positioning the
2886 cursor after the expression and typing @kbd{C-x C-e}.
2888 @cindex @samp{point} defined
2889 In Emacs, the current position of the cursor is called @dfn{point}.
2890 The expression @code{(point)} returns a number that tells you where the
2891 cursor is located as a count of the number of characters from the
2892 beginning of the buffer up to point.
2895 You can see the character count for point in this buffer by evaluating
2896 the following expression in the usual way:
2903 As I write this, the value of @code{point} is 65724. The @code{point}
2904 function is frequently used in some of the examples later in this
2908 The value of point depends, of course, on its location within the
2909 buffer. If you evaluate point in this spot, the number will be larger:
2916 For me, the value of point in this location is 66043, which means that
2917 there are 319 characters (including spaces) between the two
2918 expressions. (Doubtless, you will see different numbers, since I will
2919 have edited this since I first evaluated point.)
2921 @cindex @samp{narrowing} defined
2922 The function @code{point-min} is somewhat similar to @code{point}, but
2923 it returns the value of the minimum permissible value of point in the
2924 current buffer. This is the number 1 unless @dfn{narrowing} is in
2925 effect. (Narrowing is a mechanism whereby you can restrict yourself,
2926 or a program, to operations on just a part of a buffer.
2927 @xref{Narrowing & Widening, , Narrowing and Widening}.) Likewise, the
2928 function @code{point-max} returns the value of the maximum permissible
2929 value of point in the current buffer.
2931 @node Evaluation Exercise
2934 Find a file with which you are working and move towards its middle.
2935 Find its buffer name, file name, length, and your position in the file.
2937 @node Writing Defuns
2938 @chapter How To Write Function Definitions
2939 @cindex Definition writing
2940 @cindex Function definition writing
2941 @cindex Writing a function definition
2943 When the Lisp interpreter evaluates a list, it looks to see whether the
2944 first symbol on the list has a function definition attached to it; or,
2945 put another way, whether the symbol points to a function definition. If
2946 it does, the computer carries out the instructions in the definition. A
2947 symbol that has a function definition is called, simply, a function
2948 (although, properly speaking, the definition is the function and the
2949 symbol refers to it.)
2952 * Primitive Functions::
2953 * defun:: The @code{defun} macro.
2954 * Install:: Install a function definition.
2955 * Interactive:: Making a function interactive.
2956 * Interactive Options:: Different options for @code{interactive}.
2957 * Permanent Installation:: Installing code permanently.
2958 * let:: Creating and initializing local variables.
2960 * else:: If--then--else expressions.
2961 * Truth & Falsehood:: What Lisp considers false and true.
2962 * save-excursion:: Keeping track of point, mark, and buffer.
2968 @node Primitive Functions
2969 @unnumberedsec An Aside about Primitive Functions
2971 @cindex Primitive functions
2972 @cindex Functions, primitive
2974 @cindex C language primitives
2975 @cindex Primitives written in C
2976 All functions are defined in terms of other functions, except for a few
2977 @dfn{primitive} functions that are written in the C programming
2978 language. When you write functions' definitions, you will write them in
2979 Emacs Lisp and use other functions as your building blocks. Some of the
2980 functions you will use will themselves be written in Emacs Lisp (perhaps
2981 by you) and some will be primitives written in C@. The primitive
2982 functions are used exactly like those written in Emacs Lisp and behave
2983 like them. They are written in C so we can easily run GNU Emacs on any
2984 computer that has sufficient power and can run C.
2986 Let me re-emphasize this: when you write code in Emacs Lisp, you do not
2987 distinguish between the use of functions written in C and the use of
2988 functions written in Emacs Lisp. The difference is irrelevant. I
2989 mention the distinction only because it is interesting to know. Indeed,
2990 unless you investigate, you won't know whether an already-written
2991 function is written in Emacs Lisp or C.
2994 @section The @code{defun} Macro
2997 @cindex @samp{function definition} defined
2998 In Lisp, a symbol such as @code{mark-whole-buffer} has code attached to
2999 it that tells the computer what to do when the function is called.
3000 This code is called the @dfn{function definition} and is created by
3001 evaluating a Lisp expression that starts with the symbol @code{defun}
3002 (which is an abbreviation for @emph{define function}).
3004 In subsequent sections, we will look at function definitions from the
3005 Emacs source code, such as @code{mark-whole-buffer}. In this section,
3006 we will describe a simple function definition so you can see how it
3007 looks. This function definition uses arithmetic because it makes for a
3008 simple example. Some people dislike examples using arithmetic; however,
3009 if you are such a person, do not despair. Hardly any of the code we
3010 will study in the remainder of this introduction involves arithmetic or
3011 mathematics. The examples mostly involve text in one way or another.
3013 A function definition has up to five parts following the word
3018 The name of the symbol to which the function definition should be
3022 A list of the arguments that will be passed to the function. If no
3023 arguments will be passed to the function, this is an empty list,
3027 Documentation describing the function. (Technically optional, but
3028 strongly recommended.)
3031 Optionally, an expression to make the function interactive so you can
3032 use it by typing @kbd{M-x} and then the name of the function; or by
3033 typing an appropriate key or keychord.
3035 @cindex @samp{body} defined
3037 The code that instructs the computer what to do: the @dfn{body} of the
3038 function definition.
3041 It is helpful to think of the five parts of a function definition as
3042 being organized in a template, with slots for each part:
3046 (defun @var{function-name} (@var{arguments}@dots{})
3047 "@var{optional-documentation}@dots{}"
3048 (interactive @var{argument-passing-info}) ; @r{optional}
3053 As an example, here is the code for a function that multiplies its
3054 argument by 7. (This example is not interactive. @xref{Interactive,
3055 , Making a Function Interactive}, for that information.)
3059 (defun multiply-by-seven (number)
3060 "Multiply NUMBER by seven."
3065 This definition begins with a parenthesis and the symbol @code{defun},
3066 followed by the name of the function.
3068 @cindex @samp{argument list} defined
3069 The name of the function is followed by a list that contains the
3070 arguments that will be passed to the function. This list is called
3071 the @dfn{argument list}. In this example, the list has only one
3072 element, the symbol, @code{number}. When the function is used, the
3073 symbol will be bound to the value that is used as the argument to the
3076 Instead of choosing the word @code{number} for the name of the argument,
3077 I could have picked any other name. For example, I could have chosen
3078 the word @code{multiplicand}. I picked the word `number' because it
3079 tells what kind of value is intended for this slot; but I could just as
3080 well have chosen the word `multiplicand' to indicate the role that the
3081 value placed in this slot will play in the workings of the function. I
3082 could have called it @code{foogle}, but that would have been a bad
3083 choice because it would not tell humans what it means. The choice of
3084 name is up to the programmer and should be chosen to make the meaning of
3087 Indeed, you can choose any name you wish for a symbol in an argument
3088 list, even the name of a symbol used in some other function: the name
3089 you use in an argument list is private to that particular definition.
3090 In that definition, the name refers to a different entity than any use
3091 of the same name outside the function definition. Suppose you have a
3092 nick-name `Shorty' in your family; when your family members refer to
3093 `Shorty', they mean you. But outside your family, in a movie, for
3094 example, the name `Shorty' refers to someone else. Because a name in an
3095 argument list is private to the function definition, you can change the
3096 value of such a symbol inside the body of a function without changing
3097 its value outside the function. The effect is similar to that produced
3098 by a @code{let} expression. (@xref{let, , @code{let}}.)
3101 Note also that we discuss the word `number' in two different ways: as a
3102 symbol that appears in the code, and as the name of something that will
3103 be replaced by a something else during the evaluation of the function.
3104 In the first case, @code{number} is a symbol, not a number; it happens
3105 that within the function, it is a variable who value is the number in
3106 question, but our primary interest in it is as a symbol. On the other
3107 hand, when we are talking about the function, our interest is that we
3108 will substitute a number for the word @var{number}. To keep this
3109 distinction clear, we use different typography for the two
3110 circumstances. When we talk about this function, or about how it works,
3111 we refer to this number by writing @var{number}. In the function
3112 itself, we refer to it by writing @code{number}.
3115 The argument list is followed by the documentation string that
3116 describes the function. This is what you see when you type
3117 @w{@kbd{C-h f}} and the name of a function. Incidentally, when you
3118 write a documentation string like this, you should make the first line
3119 a complete sentence since some commands, such as @code{apropos}, print
3120 only the first line of a multi-line documentation string. Also, you
3121 should not indent the second line of a documentation string, if you
3122 have one, because that looks odd when you use @kbd{C-h f}
3123 (@code{describe-function}). The documentation string is optional, but
3124 it is so useful, it should be included in almost every function you
3127 @findex * @r{(multiplication)}
3128 The third line of the example consists of the body of the function
3129 definition. (Most functions' definitions, of course, are longer than
3130 this.) In this function, the body is the list, @code{(* 7 number)}, which
3131 says to multiply the value of @var{number} by 7. (In Emacs Lisp,
3132 @code{*} is the function for multiplication, just as @code{+} is the
3133 function for addition.)
3135 When you use the @code{multiply-by-seven} function, the argument
3136 @code{number} evaluates to the actual number you want used. Here is an
3137 example that shows how @code{multiply-by-seven} is used; but don't try
3138 to evaluate this yet!
3141 (multiply-by-seven 3)
3145 The symbol @code{number}, specified in the function definition in the
3146 next section, is given or ``bound to'' the value 3 in the actual use of
3147 the function. Note that although @code{number} was inside parentheses
3148 in the function definition, the argument passed to the
3149 @code{multiply-by-seven} function is not in parentheses. The
3150 parentheses are written in the function definition so the computer can
3151 figure out where the argument list ends and the rest of the function
3154 If you evaluate this example, you are likely to get an error message.
3155 (Go ahead, try it!) This is because we have written the function
3156 definition, but not yet told the computer about the definition---we have
3157 not yet installed (or `loaded') the function definition in Emacs.
3158 Installing a function is the process that tells the Lisp interpreter the
3159 definition of the function. Installation is described in the next
3163 @section Install a Function Definition
3164 @cindex Install a Function Definition
3165 @cindex Definition installation
3166 @cindex Function definition installation
3168 If you are reading this inside of Info in Emacs, you can try out the
3169 @code{multiply-by-seven} function by first evaluating the function
3170 definition and then evaluating @code{(multiply-by-seven 3)}. A copy of
3171 the function definition follows. Place the cursor after the last
3172 parenthesis of the function definition and type @kbd{C-x C-e}. When you
3173 do this, @code{multiply-by-seven} will appear in the echo area. (What
3174 this means is that when a function definition is evaluated, the value it
3175 returns is the name of the defined function.) At the same time, this
3176 action installs the function definition.
3180 (defun multiply-by-seven (number)
3181 "Multiply NUMBER by seven."
3187 By evaluating this @code{defun}, you have just installed
3188 @code{multiply-by-seven} in Emacs. The function is now just as much a
3189 part of Emacs as @code{forward-word} or any other editing function you
3190 use. (@code{multiply-by-seven} will stay installed until you quit
3191 Emacs. To reload code automatically whenever you start Emacs, see
3192 @ref{Permanent Installation, , Installing Code Permanently}.)
3195 * Effect of installation::
3196 * Change a defun:: How to change a function definition.
3200 @node Effect of installation
3201 @unnumberedsubsec The effect of installation
3204 You can see the effect of installing @code{multiply-by-seven} by
3205 evaluating the following sample. Place the cursor after the following
3206 expression and type @kbd{C-x C-e}. The number 21 will appear in the
3210 (multiply-by-seven 3)
3213 If you wish, you can read the documentation for the function by typing
3214 @kbd{C-h f} (@code{describe-function}) and then the name of the
3215 function, @code{multiply-by-seven}. When you do this, a
3216 @file{*Help*} window will appear on your screen that says:
3220 multiply-by-seven is a Lisp function.
3221 (multiply-by-seven NUMBER)
3223 Multiply NUMBER by seven.
3228 (To return to a single window on your screen, type @kbd{C-x 1}.)
3230 @node Change a defun
3231 @subsection Change a Function Definition
3232 @cindex Changing a function definition
3233 @cindex Function definition, how to change
3234 @cindex Definition, how to change
3236 If you want to change the code in @code{multiply-by-seven}, just rewrite
3237 it. To install the new version in place of the old one, evaluate the
3238 function definition again. This is how you modify code in Emacs. It is
3241 As an example, you can change the @code{multiply-by-seven} function to
3242 add the number to itself seven times instead of multiplying the number
3243 by seven. It produces the same answer, but by a different path. At
3244 the same time, we will add a comment to the code; a comment is text
3245 that the Lisp interpreter ignores, but that a human reader may find
3246 useful or enlightening. The comment is that this is the ``second
3251 (defun multiply-by-seven (number) ; @r{Second version.}
3252 "Multiply NUMBER by seven."
3253 (+ number number number number number number number))
3257 @cindex Comments in Lisp code
3258 The comment follows a semicolon, @samp{;}. In Lisp, everything on a
3259 line that follows a semicolon is a comment. The end of the line is the
3260 end of the comment. To stretch a comment over two or more lines, begin
3261 each line with a semicolon.
3263 @xref{Beginning a .emacs File, , Beginning a @file{.emacs}
3264 File}, and @ref{Comments, , Comments, elisp, The GNU Emacs Lisp
3265 Reference Manual}, for more about comments.
3267 You can install this version of the @code{multiply-by-seven} function by
3268 evaluating it in the same way you evaluated the first function: place
3269 the cursor after the last parenthesis and type @kbd{C-x C-e}.
3271 In summary, this is how you write code in Emacs Lisp: you write a
3272 function; install it; test it; and then make fixes or enhancements and
3276 @section Make a Function Interactive
3277 @cindex Interactive functions
3280 You make a function interactive by placing a list that begins with
3281 the special form @code{interactive} immediately after the
3282 documentation. A user can invoke an interactive function by typing
3283 @kbd{M-x} and then the name of the function; or by typing the keys to
3284 which it is bound, for example, by typing @kbd{C-n} for
3285 @code{next-line} or @kbd{C-x h} for @code{mark-whole-buffer}.
3287 Interestingly, when you call an interactive function interactively,
3288 the value returned is not automatically displayed in the echo area.
3289 This is because you often call an interactive function for its side
3290 effects, such as moving forward by a word or line, and not for the
3291 value returned. If the returned value were displayed in the echo area
3292 each time you typed a key, it would be very distracting.
3295 * Interactive multiply-by-seven:: An overview.
3296 * multiply-by-seven in detail:: The interactive version.
3300 @node Interactive multiply-by-seven
3301 @unnumberedsubsec An Interactive @code{multiply-by-seven}, An Overview
3304 Both the use of the special form @code{interactive} and one way to
3305 display a value in the echo area can be illustrated by creating an
3306 interactive version of @code{multiply-by-seven}.
3313 (defun multiply-by-seven (number) ; @r{Interactive version.}
3314 "Multiply NUMBER by seven."
3316 (message "The result is %d" (* 7 number)))
3321 You can install this code by placing your cursor after it and typing
3322 @kbd{C-x C-e}. The name of the function will appear in your echo area.
3323 Then, you can use this code by typing @kbd{C-u} and a number and then
3324 typing @kbd{M-x multiply-by-seven} and pressing @key{RET}. The phrase
3325 @samp{The result is @dots{}} followed by the product will appear in the
3328 Speaking more generally, you invoke a function like this in either of two
3333 By typing a prefix argument that contains the number to be passed, and
3334 then typing @kbd{M-x} and the name of the function, as with
3335 @kbd{C-u 3 M-x forward-sentence}; or,
3338 By typing whatever key or keychord the function is bound to, as with
3343 Both the examples just mentioned work identically to move point forward
3344 three sentences. (Since @code{multiply-by-seven} is not bound to a key,
3345 it could not be used as an example of key binding.)
3347 (@xref{Keybindings, , Some Keybindings}, to learn how to bind a command
3350 A prefix argument is passed to an interactive function by typing the
3351 @key{META} key followed by a number, for example, @kbd{M-3 M-e}, or by
3352 typing @kbd{C-u} and then a number, for example, @kbd{C-u 3 M-e} (if you
3353 type @kbd{C-u} without a number, it defaults to 4).
3355 @node multiply-by-seven in detail
3356 @subsection An Interactive @code{multiply-by-seven}
3358 Let's look at the use of the special form @code{interactive} and then at
3359 the function @code{message} in the interactive version of
3360 @code{multiply-by-seven}. You will recall that the function definition
3365 (defun multiply-by-seven (number) ; @r{Interactive version.}
3366 "Multiply NUMBER by seven."
3368 (message "The result is %d" (* 7 number)))
3372 In this function, the expression, @code{(interactive "p")}, is a list of
3373 two elements. The @code{"p"} tells Emacs to pass the prefix argument to
3374 the function and use its value for the argument of the function.
3377 The argument will be a number. This means that the symbol
3378 @code{number} will be bound to a number in the line:
3381 (message "The result is %d" (* 7 number))
3386 For example, if your prefix argument is 5, the Lisp interpreter will
3387 evaluate the line as if it were:
3390 (message "The result is %d" (* 7 5))
3394 (If you are reading this in GNU Emacs, you can evaluate this expression
3395 yourself.) First, the interpreter will evaluate the inner list, which
3396 is @code{(* 7 5)}. This returns a value of 35. Next, it
3397 will evaluate the outer list, passing the values of the second and
3398 subsequent elements of the list to the function @code{message}.
3400 As we have seen, @code{message} is an Emacs Lisp function especially
3401 designed for sending a one line message to a user. (@xref{message, ,
3402 The @code{message} function}.) In summary, the @code{message}
3403 function prints its first argument in the echo area as is, except for
3404 occurrences of @samp{%d} or @samp{%s} (and various other %-sequences
3405 which we have not mentioned). When it sees a control sequence, the
3406 function looks to the second or subsequent arguments and prints the
3407 value of the argument in the location in the string where the control
3408 sequence is located.
3410 In the interactive @code{multiply-by-seven} function, the control string
3411 is @samp{%d}, which requires a number, and the value returned by
3412 evaluating @code{(* 7 5)} is the number 35. Consequently, the number 35
3413 is printed in place of the @samp{%d} and the message is @samp{The result
3416 (Note that when you call the function @code{multiply-by-seven}, the
3417 message is printed without quotes, but when you call @code{message}, the
3418 text is printed in double quotes. This is because the value returned by
3419 @code{message} is what appears in the echo area when you evaluate an
3420 expression whose first element is @code{message}; but when embedded in a
3421 function, @code{message} prints the text as a side effect without
3424 @node Interactive Options
3425 @section Different Options for @code{interactive}
3426 @cindex Options for @code{interactive}
3427 @cindex Interactive options
3429 In the example, @code{multiply-by-seven} used @code{"p"} as the
3430 argument to @code{interactive}. This argument told Emacs to interpret
3431 your typing either @kbd{C-u} followed by a number or @key{META}
3432 followed by a number as a command to pass that number to the function
3433 as its argument. Emacs has more than twenty characters predefined for
3434 use with @code{interactive}. In almost every case, one of these
3435 options will enable you to pass the right information interactively to
3436 a function. (@xref{Interactive Codes, , Code Characters for
3437 @code{interactive}, elisp, The GNU Emacs Lisp Reference Manual}.)
3440 Consider the function @code{zap-to-char}. Its interactive expression
3444 (interactive "p\ncZap to char: ")
3447 The first part of the argument to @code{interactive} is @samp{p}, with
3448 which you are already familiar. This argument tells Emacs to
3449 interpret a `prefix', as a number to be passed to the function. You
3450 can specify a prefix either by typing @kbd{C-u} followed by a number
3451 or by typing @key{META} followed by a number. The prefix is the
3452 number of specified characters. Thus, if your prefix is three and the
3453 specified character is @samp{x}, then you will delete all the text up
3454 to and including the third next @samp{x}. If you do not set a prefix,
3455 then you delete all the text up to and including the specified
3456 character, but no more.
3458 The @samp{c} tells the function the name of the character to which to delete.
3460 More formally, a function with two or more arguments can have
3461 information passed to each argument by adding parts to the string that
3462 follows @code{interactive}. When you do this, the information is
3463 passed to each argument in the same order it is specified in the
3464 @code{interactive} list. In the string, each part is separated from
3465 the next part by a @samp{\n}, which is a newline. For example, you
3466 can follow @samp{p} with a @samp{\n} and an @samp{cZap to char:@: }.
3467 This causes Emacs to pass the value of the prefix argument (if there
3468 is one) and the character.
3470 In this case, the function definition looks like the following, where
3471 @code{arg} and @code{char} are the symbols to which @code{interactive}
3472 binds the prefix argument and the specified character:
3476 (defun @var{name-of-function} (arg char)
3477 "@var{documentation}@dots{}"
3478 (interactive "p\ncZap to char: ")
3479 @var{body-of-function}@dots{})
3484 (The space after the colon in the prompt makes it look better when you
3485 are prompted. @xref{copy-to-buffer, , The Definition of
3486 @code{copy-to-buffer}}, for an example.)
3488 When a function does not take arguments, @code{interactive} does not
3489 require any. Such a function contains the simple expression
3490 @code{(interactive)}. The @code{mark-whole-buffer} function is like
3493 Alternatively, if the special letter-codes are not right for your
3494 application, you can pass your own arguments to @code{interactive} as
3497 @xref{append-to-buffer, , The Definition of @code{append-to-buffer}},
3498 for an example. @xref{Using Interactive, , Using @code{Interactive},
3499 elisp, The GNU Emacs Lisp Reference Manual}, for a more complete
3500 explanation about this technique.
3502 @node Permanent Installation
3503 @section Install Code Permanently
3504 @cindex Install code permanently
3505 @cindex Permanent code installation
3506 @cindex Code installation
3508 When you install a function definition by evaluating it, it will stay
3509 installed until you quit Emacs. The next time you start a new session
3510 of Emacs, the function will not be installed unless you evaluate the
3511 function definition again.
3513 At some point, you may want to have code installed automatically
3514 whenever you start a new session of Emacs. There are several ways of
3519 If you have code that is just for yourself, you can put the code for the
3520 function definition in your @file{.emacs} initialization file. When you
3521 start Emacs, your @file{.emacs} file is automatically evaluated and all
3522 the function definitions within it are installed.
3523 @xref{Emacs Initialization, , Your @file{.emacs} File}.
3526 Alternatively, you can put the function definitions that you want
3527 installed in one or more files of their own and use the @code{load}
3528 function to cause Emacs to evaluate and thereby install each of the
3529 functions in the files.
3530 @xref{Loading Files, , Loading Files}.
3533 Thirdly, if you have code that your whole site will use, it is usual
3534 to put it in a file called @file{site-init.el} that is loaded when
3535 Emacs is built. This makes the code available to everyone who uses
3536 your machine. (See the @file{INSTALL} file that is part of the Emacs
3540 Finally, if you have code that everyone who uses Emacs may want, you
3541 can post it on a computer network or send a copy to the Free Software
3542 Foundation. (When you do this, please license the code and its
3543 documentation under a license that permits other people to run, copy,
3544 study, modify, and redistribute the code and which protects you from
3545 having your work taken from you.) If you send a copy of your code to
3546 the Free Software Foundation, and properly protect yourself and
3547 others, it may be included in the next release of Emacs. In large
3548 part, this is how Emacs has grown over the past years, by donations.
3554 The @code{let} expression is a special form in Lisp that you will need
3555 to use in most function definitions.
3557 @code{let} is used to attach or bind a symbol to a value in such a way
3558 that the Lisp interpreter will not confuse the variable with a
3559 variable of the same name that is not part of the function.
3561 To understand why the @code{let} special form is necessary, consider
3562 the situation in which you own a home that you generally refer to as
3563 `the house', as in the sentence, ``The house needs painting.'' If you
3564 are visiting a friend and your host refers to `the house', he is
3565 likely to be referring to @emph{his} house, not yours, that is, to a
3568 If your friend is referring to his house and you think he is referring
3569 to your house, you may be in for some confusion. The same thing could
3570 happen in Lisp if a variable that is used inside of one function has
3571 the same name as a variable that is used inside of another function,
3572 and the two are not intended to refer to the same value. The
3573 @code{let} special form prevents this kind of confusion.
3576 * Prevent confusion::
3577 * Parts of let Expression::
3578 * Sample let Expression::
3579 * Uninitialized let Variables::
3583 @node Prevent confusion
3584 @unnumberedsubsec @code{let} Prevents Confusion
3587 @cindex @samp{local variable} defined
3588 @cindex @samp{variable, local}, defined
3589 The @code{let} special form prevents confusion. @code{let} creates a
3590 name for a @dfn{local variable} that overshadows any use of the same
3591 name outside the @code{let} expression. This is like understanding
3592 that whenever your host refers to `the house', he means his house, not
3593 yours. (Symbols used in argument lists work the same way.
3594 @xref{defun, , The @code{defun} Macro}.)
3596 Local variables created by a @code{let} expression retain their value
3597 @emph{only} within the @code{let} expression itself (and within
3598 expressions called within the @code{let} expression); the local
3599 variables have no effect outside the @code{let} expression.
3601 Another way to think about @code{let} is that it is like a @code{setq}
3602 that is temporary and local. The values set by @code{let} are
3603 automatically undone when the @code{let} is finished. The setting
3604 only affects expressions that are inside the bounds of the @code{let}
3605 expression. In computer science jargon, we would say ``the binding of
3606 a symbol is visible only in functions called in the @code{let} form;
3607 in Emacs Lisp, scoping is dynamic, not lexical.''
3609 @code{let} can create more than one variable at once. Also,
3610 @code{let} gives each variable it creates an initial value, either a
3611 value specified by you, or @code{nil}. (In the jargon, this is called
3612 `binding the variable to the value'.) After @code{let} has created
3613 and bound the variables, it executes the code in the body of the
3614 @code{let}, and returns the value of the last expression in the body,
3615 as the value of the whole @code{let} expression. (`Execute' is a jargon
3616 term that means to evaluate a list; it comes from the use of the word
3617 meaning `to give practical effect to' (@cite{Oxford English
3618 Dictionary}). Since you evaluate an expression to perform an action,
3619 `execute' has evolved as a synonym to `evaluate'.)
3621 @node Parts of let Expression
3622 @subsection The Parts of a @code{let} Expression
3623 @cindex @code{let} expression, parts of
3624 @cindex Parts of @code{let} expression
3626 @cindex @samp{varlist} defined
3627 A @code{let} expression is a list of three parts. The first part is
3628 the symbol @code{let}. The second part is a list, called a
3629 @dfn{varlist}, each element of which is either a symbol by itself or a
3630 two-element list, the first element of which is a symbol. The third
3631 part of the @code{let} expression is the body of the @code{let}. The
3632 body usually consists of one or more lists.
3635 A template for a @code{let} expression looks like this:
3638 (let @var{varlist} @var{body}@dots{})
3642 The symbols in the varlist are the variables that are given initial
3643 values by the @code{let} special form. Symbols by themselves are given
3644 the initial value of @code{nil}; and each symbol that is the first
3645 element of a two-element list is bound to the value that is returned
3646 when the Lisp interpreter evaluates the second element.
3648 Thus, a varlist might look like this: @code{(thread (needles 3))}. In
3649 this case, in a @code{let} expression, Emacs binds the symbol
3650 @code{thread} to an initial value of @code{nil}, and binds the symbol
3651 @code{needles} to an initial value of 3.
3653 When you write a @code{let} expression, what you do is put the
3654 appropriate expressions in the slots of the @code{let} expression
3657 If the varlist is composed of two-element lists, as is often the case,
3658 the template for the @code{let} expression looks like this:
3662 (let ((@var{variable} @var{value})
3663 (@var{variable} @var{value})
3669 @node Sample let Expression
3670 @subsection Sample @code{let} Expression
3671 @cindex Sample @code{let} expression
3672 @cindex @code{let} expression sample
3674 The following expression creates and gives initial values
3675 to the two variables @code{zebra} and @code{tiger}. The body of the
3676 @code{let} expression is a list which calls the @code{message} function.
3680 (let ((zebra 'stripes)
3682 (message "One kind of animal has %s and another is %s."
3687 Here, the varlist is @code{((zebra 'stripes) (tiger 'fierce))}.
3689 The two variables are @code{zebra} and @code{tiger}. Each variable is
3690 the first element of a two-element list and each value is the second
3691 element of its two-element list. In the varlist, Emacs binds the
3692 variable @code{zebra} to the value @code{stripes}@footnote{According
3693 to Jared Diamond in @cite{Guns, Germs, and Steel}, ``@dots{} zebras
3694 become impossibly dangerous as they grow older'' but the claim here is
3695 that they do not become fierce like a tiger. (1997, W. W. Norton and
3696 Co., ISBN 0-393-03894-2, page 171)}, and binds the
3697 variable @code{tiger} to the value @code{fierce}. In this example,
3698 both values are symbols preceded by a quote. The values could just as
3699 well have been another list or a string. The body of the @code{let}
3700 follows after the list holding the variables. In this example, the
3701 body is a list that uses the @code{message} function to print a string
3705 You may evaluate the example in the usual fashion, by placing the
3706 cursor after the last parenthesis and typing @kbd{C-x C-e}. When you do
3707 this, the following will appear in the echo area:
3710 "One kind of animal has stripes and another is fierce."
3713 As we have seen before, the @code{message} function prints its first
3714 argument, except for @samp{%s}. In this example, the value of the variable
3715 @code{zebra} is printed at the location of the first @samp{%s} and the
3716 value of the variable @code{tiger} is printed at the location of the
3719 @node Uninitialized let Variables
3720 @subsection Uninitialized Variables in a @code{let} Statement
3721 @cindex Uninitialized @code{let} variables
3722 @cindex @code{let} variables uninitialized
3724 If you do not bind the variables in a @code{let} statement to specific
3725 initial values, they will automatically be bound to an initial value of
3726 @code{nil}, as in the following expression:
3735 "Here are %d variables with %s, %s, and %s value."
3736 birch pine fir oak))
3741 Here, the varlist is @code{((birch 3) pine fir (oak 'some))}.
3744 If you evaluate this expression in the usual way, the following will
3745 appear in your echo area:
3748 "Here are 3 variables with nil, nil, and some value."
3752 In this example, Emacs binds the symbol @code{birch} to the number 3,
3753 binds the symbols @code{pine} and @code{fir} to @code{nil}, and binds
3754 the symbol @code{oak} to the value @code{some}.
3756 Note that in the first part of the @code{let}, the variables @code{pine}
3757 and @code{fir} stand alone as atoms that are not surrounded by
3758 parentheses; this is because they are being bound to @code{nil}, the
3759 empty list. But @code{oak} is bound to @code{some} and so is a part of
3760 the list @code{(oak 'some)}. Similarly, @code{birch} is bound to the
3761 number 3 and so is in a list with that number. (Since a number
3762 evaluates to itself, the number does not need to be quoted. Also, the
3763 number is printed in the message using a @samp{%d} rather than a
3764 @samp{%s}.) The four variables as a group are put into a list to
3765 delimit them from the body of the @code{let}.
3768 @section The @code{if} Special Form
3770 @cindex Conditional with @code{if}
3772 A third special form, in addition to @code{defun} and @code{let}, is the
3773 conditional @code{if}. This form is used to instruct the computer to
3774 make decisions. You can write function definitions without using
3775 @code{if}, but it is used often enough, and is important enough, to be
3776 included here. It is used, for example, in the code for the
3777 function @code{beginning-of-buffer}.
3779 The basic idea behind an @code{if}, is that ``@emph{if} a test is true,
3780 @emph{then} an expression is evaluated.'' If the test is not true, the
3781 expression is not evaluated. For example, you might make a decision
3782 such as, ``if it is warm and sunny, then go to the beach!''
3785 * if in more detail::
3786 * type-of-animal in detail:: An example of an @code{if} expression.
3790 @node if in more detail
3791 @unnumberedsubsec @code{if} in more detail
3794 @cindex @samp{if-part} defined
3795 @cindex @samp{then-part} defined
3796 An @code{if} expression written in Lisp does not use the word `then';
3797 the test and the action are the second and third elements of the list
3798 whose first element is @code{if}. Nonetheless, the test part of an
3799 @code{if} expression is often called the @dfn{if-part} and the second
3800 argument is often called the @dfn{then-part}.
3802 Also, when an @code{if} expression is written, the true-or-false-test
3803 is usually written on the same line as the symbol @code{if}, but the
3804 action to carry out if the test is true, the ``then-part'', is written
3805 on the second and subsequent lines. This makes the @code{if}
3806 expression easier to read.
3810 (if @var{true-or-false-test}
3811 @var{action-to-carry-out-if-test-is-true})
3816 The true-or-false-test will be an expression that
3817 is evaluated by the Lisp interpreter.
3819 Here is an example that you can evaluate in the usual manner. The test
3820 is whether the number 5 is greater than the number 4. Since it is, the
3821 message @samp{5 is greater than 4!} will be printed.
3825 (if (> 5 4) ; @r{if-part}
3826 (message "5 is greater than 4!")) ; @r{then-part}
3831 (The function @code{>} tests whether its first argument is greater than
3832 its second argument and returns true if it is.)
3833 @findex > (greater than)
3835 Of course, in actual use, the test in an @code{if} expression will not
3836 be fixed for all time as it is by the expression @code{(> 5 4)}.
3837 Instead, at least one of the variables used in the test will be bound to
3838 a value that is not known ahead of time. (If the value were known ahead
3839 of time, we would not need to run the test!)
3841 For example, the value may be bound to an argument of a function
3842 definition. In the following function definition, the character of the
3843 animal is a value that is passed to the function. If the value bound to
3844 @code{characteristic} is @code{fierce}, then the message, @samp{It's a
3845 tiger!} will be printed; otherwise, @code{nil} will be returned.
3849 (defun type-of-animal (characteristic)
3850 "Print message in echo area depending on CHARACTERISTIC.
3851 If the CHARACTERISTIC is the symbol `fierce',
3852 then warn of a tiger."
3853 (if (equal characteristic 'fierce)
3854 (message "It's a tiger!")))
3860 If you are reading this inside of GNU Emacs, you can evaluate the
3861 function definition in the usual way to install it in Emacs, and then you
3862 can evaluate the following two expressions to see the results:
3866 (type-of-animal 'fierce)
3868 (type-of-animal 'zebra)
3873 @c Following sentences rewritten to prevent overfull hbox.
3875 When you evaluate @code{(type-of-animal 'fierce)}, you will see the
3876 following message printed in the echo area: @code{"It's a tiger!"}; and
3877 when you evaluate @code{(type-of-animal 'zebra)} you will see @code{nil}
3878 printed in the echo area.
3880 @node type-of-animal in detail
3881 @subsection The @code{type-of-animal} Function in Detail
3883 Let's look at the @code{type-of-animal} function in detail.
3885 The function definition for @code{type-of-animal} was written by filling
3886 the slots of two templates, one for a function definition as a whole, and
3887 a second for an @code{if} expression.
3890 The template for every function that is not interactive is:
3894 (defun @var{name-of-function} (@var{argument-list})
3895 "@var{documentation}@dots{}"
3901 The parts of the function that match this template look like this:
3905 (defun type-of-animal (characteristic)
3906 "Print message in echo area depending on CHARACTERISTIC.
3907 If the CHARACTERISTIC is the symbol `fierce',
3908 then warn of a tiger."
3909 @var{body: the} @code{if} @var{expression})
3913 The name of function is @code{type-of-animal}; it is passed the value
3914 of one argument. The argument list is followed by a multi-line
3915 documentation string. The documentation string is included in the
3916 example because it is a good habit to write documentation string for
3917 every function definition. The body of the function definition
3918 consists of the @code{if} expression.
3921 The template for an @code{if} expression looks like this:
3925 (if @var{true-or-false-test}
3926 @var{action-to-carry-out-if-the-test-returns-true})
3931 In the @code{type-of-animal} function, the code for the @code{if}
3936 (if (equal characteristic 'fierce)
3937 (message "It's a tiger!")))
3942 Here, the true-or-false-test is the expression:
3945 (equal characteristic 'fierce)
3949 In Lisp, @code{equal} is a function that determines whether its first
3950 argument is equal to its second argument. The second argument is the
3951 quoted symbol @code{'fierce} and the first argument is the value of the
3952 symbol @code{characteristic}---in other words, the argument passed to
3955 In the first exercise of @code{type-of-animal}, the argument
3956 @code{fierce} is passed to @code{type-of-animal}. Since @code{fierce}
3957 is equal to @code{fierce}, the expression, @code{(equal characteristic
3958 'fierce)}, returns a value of true. When this happens, the @code{if}
3959 evaluates the second argument or then-part of the @code{if}:
3960 @code{(message "It's tiger!")}.
3962 On the other hand, in the second exercise of @code{type-of-animal}, the
3963 argument @code{zebra} is passed to @code{type-of-animal}. @code{zebra}
3964 is not equal to @code{fierce}, so the then-part is not evaluated and
3965 @code{nil} is returned by the @code{if} expression.
3968 @section If--then--else Expressions
3971 An @code{if} expression may have an optional third argument, called
3972 the @dfn{else-part}, for the case when the true-or-false-test returns
3973 false. When this happens, the second argument or then-part of the
3974 overall @code{if} expression is @emph{not} evaluated, but the third or
3975 else-part @emph{is} evaluated. You might think of this as the cloudy
3976 day alternative for the decision ``if it is warm and sunny, then go to
3977 the beach, else read a book!''.
3979 The word ``else'' is not written in the Lisp code; the else-part of an
3980 @code{if} expression comes after the then-part. In the written Lisp, the
3981 else-part is usually written to start on a line of its own and is
3982 indented less than the then-part:
3986 (if @var{true-or-false-test}
3987 @var{action-to-carry-out-if-the-test-returns-true}
3988 @var{action-to-carry-out-if-the-test-returns-false})
3992 For example, the following @code{if} expression prints the message @samp{4
3993 is not greater than 5!} when you evaluate it in the usual way:
3997 (if (> 4 5) ; @r{if-part}
3998 (message "4 falsely greater than 5!") ; @r{then-part}
3999 (message "4 is not greater than 5!")) ; @r{else-part}
4004 Note that the different levels of indentation make it easy to
4005 distinguish the then-part from the else-part. (GNU Emacs has several
4006 commands that automatically indent @code{if} expressions correctly.
4007 @xref{Typing Lists, , GNU Emacs Helps You Type Lists}.)
4009 We can extend the @code{type-of-animal} function to include an
4010 else-part by simply incorporating an additional part to the @code{if}
4014 You can see the consequences of doing this if you evaluate the following
4015 version of the @code{type-of-animal} function definition to install it
4016 and then evaluate the two subsequent expressions to pass different
4017 arguments to the function.
4021 (defun type-of-animal (characteristic) ; @r{Second version.}
4022 "Print message in echo area depending on CHARACTERISTIC.
4023 If the CHARACTERISTIC is the symbol `fierce',
4024 then warn of a tiger;
4025 else say it's not fierce."
4026 (if (equal characteristic 'fierce)
4027 (message "It's a tiger!")
4028 (message "It's not fierce!")))
4035 (type-of-animal 'fierce)
4037 (type-of-animal 'zebra)
4042 @c Following sentence rewritten to prevent overfull hbox.
4044 When you evaluate @code{(type-of-animal 'fierce)}, you will see the
4045 following message printed in the echo area: @code{"It's a tiger!"}; but
4046 when you evaluate @code{(type-of-animal 'zebra)}, you will see
4047 @code{"It's not fierce!"}.
4049 (Of course, if the @var{characteristic} were @code{ferocious}, the
4050 message @code{"It's not fierce!"} would be printed; and it would be
4051 misleading! When you write code, you need to take into account the
4052 possibility that some such argument will be tested by the @code{if}
4053 and write your program accordingly.)
4055 @node Truth & Falsehood
4056 @section Truth and Falsehood in Emacs Lisp
4057 @cindex Truth and falsehood in Emacs Lisp
4058 @cindex Falsehood and truth in Emacs Lisp
4061 There is an important aspect to the truth test in an @code{if}
4062 expression. So far, we have spoken of `true' and `false' as values of
4063 predicates as if they were new kinds of Emacs Lisp objects. In fact,
4064 `false' is just our old friend @code{nil}. Anything else---anything
4067 The expression that tests for truth is interpreted as @dfn{true}
4068 if the result of evaluating it is a value that is not @code{nil}. In
4069 other words, the result of the test is considered true if the value
4070 returned is a number such as 47, a string such as @code{"hello"}, or a
4071 symbol (other than @code{nil}) such as @code{flowers}, or a list (so
4072 long as it is not empty), or even a buffer!
4075 * nil explained:: @code{nil} has two meanings.
4080 @unnumberedsubsec An explanation of @code{nil}
4083 Before illustrating a test for truth, we need an explanation of @code{nil}.
4085 In Emacs Lisp, the symbol @code{nil} has two meanings. First, it means the
4086 empty list. Second, it means false and is the value returned when a
4087 true-or-false-test tests false. @code{nil} can be written as an empty
4088 list, @code{()}, or as @code{nil}. As far as the Lisp interpreter is
4089 concerned, @code{()} and @code{nil} are the same. Humans, however, tend
4090 to use @code{nil} for false and @code{()} for the empty list.
4092 In Emacs Lisp, any value that is not @code{nil}---is not the empty
4093 list---is considered true. This means that if an evaluation returns
4094 something that is not an empty list, an @code{if} expression will test
4095 true. For example, if a number is put in the slot for the test, it
4096 will be evaluated and will return itself, since that is what numbers
4097 do when evaluated. In this conditional, the @code{if} expression will
4098 test true. The expression tests false only when @code{nil}, an empty
4099 list, is returned by evaluating the expression.
4101 You can see this by evaluating the two expressions in the following examples.
4103 In the first example, the number 4 is evaluated as the test in the
4104 @code{if} expression and returns itself; consequently, the then-part
4105 of the expression is evaluated and returned: @samp{true} appears in
4106 the echo area. In the second example, the @code{nil} indicates false;
4107 consequently, the else-part of the expression is evaluated and
4108 returned: @samp{false} appears in the echo area.
4125 Incidentally, if some other useful value is not available for a test that
4126 returns true, then the Lisp interpreter will return the symbol @code{t}
4127 for true. For example, the expression @code{(> 5 4)} returns @code{t}
4128 when evaluated, as you can see by evaluating it in the usual way:
4136 On the other hand, this function returns @code{nil} if the test is false.
4142 @node save-excursion
4143 @section @code{save-excursion}
4144 @findex save-excursion
4145 @cindex Region, what it is
4146 @cindex Preserving point, mark, and buffer
4147 @cindex Point, mark, buffer preservation
4151 The @code{save-excursion} function is the third and final special form
4152 that we will discuss in this chapter.
4154 In Emacs Lisp programs used for editing, the @code{save-excursion}
4155 function is very common. It saves the location of point and mark,
4156 executes the body of the function, and then restores point and mark to
4157 their previous positions if their locations were changed. Its primary
4158 purpose is to keep the user from being surprised and disturbed by
4159 unexpected movement of point or mark.
4162 * Point and mark:: A review of various locations.
4163 * Template for save-excursion::
4167 @node Point and mark
4168 @unnumberedsubsec Point and Mark
4171 Before discussing @code{save-excursion}, however, it may be useful
4172 first to review what point and mark are in GNU Emacs. @dfn{Point} is
4173 the current location of the cursor. Wherever the cursor
4174 is, that is point. More precisely, on terminals where the cursor
4175 appears to be on top of a character, point is immediately before the
4176 character. In Emacs Lisp, point is an integer. The first character in
4177 a buffer is number one, the second is number two, and so on. The
4178 function @code{point} returns the current position of the cursor as a
4179 number. Each buffer has its own value for point.
4181 The @dfn{mark} is another position in the buffer; its value can be set
4182 with a command such as @kbd{C-@key{SPC}} (@code{set-mark-command}). If
4183 a mark has been set, you can use the command @kbd{C-x C-x}
4184 (@code{exchange-point-and-mark}) to cause the cursor to jump to the mark
4185 and set the mark to be the previous position of point. In addition, if
4186 you set another mark, the position of the previous mark is saved in the
4187 mark ring. Many mark positions can be saved this way. You can jump the
4188 cursor to a saved mark by typing @kbd{C-u C-@key{SPC}} one or more
4191 The part of the buffer between point and mark is called @dfn{the
4192 region}. Numerous commands work on the region, including
4193 @code{center-region}, @code{count-lines-region}, @code{kill-region}, and
4194 @code{print-region}.
4196 The @code{save-excursion} special form saves the locations of point and
4197 mark and restores those positions after the code within the body of the
4198 special form is evaluated by the Lisp interpreter. Thus, if point were
4199 in the beginning of a piece of text and some code moved point to the end
4200 of the buffer, the @code{save-excursion} would put point back to where
4201 it was before, after the expressions in the body of the function were
4204 In Emacs, a function frequently moves point as part of its internal
4205 workings even though a user would not expect this. For example,
4206 @code{count-lines-region} moves point. To prevent the user from being
4207 bothered by jumps that are both unexpected and (from the user's point of
4208 view) unnecessary, @code{save-excursion} is often used to keep point and
4209 mark in the location expected by the user. The use of
4210 @code{save-excursion} is good housekeeping.
4212 To make sure the house stays clean, @code{save-excursion} restores the
4213 values of point and mark even if something goes wrong in the code inside
4214 of it (or, to be more precise and to use the proper jargon, ``in case of
4215 abnormal exit''). This feature is very helpful.
4217 In addition to recording the values of point and mark,
4218 @code{save-excursion} keeps track of the current buffer, and restores
4219 it, too. This means you can write code that will change the buffer and
4220 have @code{save-excursion} switch you back to the original buffer.
4221 This is how @code{save-excursion} is used in @code{append-to-buffer}.
4222 (@xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.)
4224 @node Template for save-excursion
4225 @subsection Template for a @code{save-excursion} Expression
4228 The template for code using @code{save-excursion} is simple:
4238 The body of the function is one or more expressions that will be
4239 evaluated in sequence by the Lisp interpreter. If there is more than
4240 one expression in the body, the value of the last one will be returned
4241 as the value of the @code{save-excursion} function. The other
4242 expressions in the body are evaluated only for their side effects; and
4243 @code{save-excursion} itself is used only for its side effect (which
4244 is restoring the positions of point and mark).
4247 In more detail, the template for a @code{save-excursion} expression
4253 @var{first-expression-in-body}
4254 @var{second-expression-in-body}
4255 @var{third-expression-in-body}
4257 @var{last-expression-in-body})
4262 An expression, of course, may be a symbol on its own or a list.
4264 In Emacs Lisp code, a @code{save-excursion} expression often occurs
4265 within the body of a @code{let} expression. It looks like this:
4278 In the last few chapters we have introduced a macro and a fair number
4279 of functions and special forms. Here they are described in brief,
4280 along with a few similar functions that have not been mentioned yet.
4283 @item eval-last-sexp
4284 Evaluate the last symbolic expression before the current location of
4285 point. The value is printed in the echo area unless the function is
4286 invoked with an argument; in that case, the output is printed in the
4287 current buffer. This command is normally bound to @kbd{C-x C-e}.
4290 Define function. This macro has up to five parts: the name, a
4291 template for the arguments that will be passed to the function,
4292 documentation, an optional interactive declaration, and the body of
4296 For example, in an early version of Emacs, the function definition was
4297 as follows. (It is slightly more complex now that it seeks the first
4298 non-whitespace character rather than the first visible character.)
4302 (defun back-to-indentation ()
4303 "Move point to first visible character on line."
4305 (beginning-of-line 1)
4306 (skip-chars-forward " \t"))
4313 (defun backward-to-indentation (&optional arg)
4314 "Move backward ARG lines and position at first nonblank character."
4316 (forward-line (- (or arg 1)))
4317 (skip-chars-forward " \t"))
4319 (defun back-to-indentation ()
4320 "Move point to the first non-whitespace character on this line."
4322 (beginning-of-line 1)
4323 (skip-syntax-forward " " (line-end-position))
4324 ;; Move back over chars that have whitespace syntax but have the p flag.
4325 (backward-prefix-chars))
4329 Declare to the interpreter that the function can be used
4330 interactively. This special form may be followed by a string with one
4331 or more parts that pass the information to the arguments of the
4332 function, in sequence. These parts may also tell the interpreter to
4333 prompt for information. Parts of the string are separated by
4334 newlines, @samp{\n}.
4337 Common code characters are:
4341 The name of an existing buffer.
4344 The name of an existing file.
4347 The numeric prefix argument. (Note that this `p' is lower case.)
4350 Point and the mark, as two numeric arguments, smallest first. This
4351 is the only code letter that specifies two successive arguments
4355 @xref{Interactive Codes, , Code Characters for @samp{interactive},
4356 elisp, The GNU Emacs Lisp Reference Manual}, for a complete list of
4360 Declare that a list of variables is for use within the body of the
4361 @code{let} and give them an initial value, either @code{nil} or a
4362 specified value; then evaluate the rest of the expressions in the body
4363 of the @code{let} and return the value of the last one. Inside the
4364 body of the @code{let}, the Lisp interpreter does not see the values of
4365 the variables of the same names that are bound outside of the
4373 (let ((foo (buffer-name))
4374 (bar (buffer-size)))
4376 "This buffer is %s and has %d characters."
4381 @item save-excursion
4382 Record the values of point and mark and the current buffer before
4383 evaluating the body of this special form. Restore the values of point
4384 and mark and buffer afterward.
4391 (message "We are %d characters into this buffer."
4394 (goto-char (point-min)) (point))))
4399 Evaluate the first argument to the function; if it is true, evaluate
4400 the second argument; else evaluate the third argument, if there is one.
4402 The @code{if} special form is called a @dfn{conditional}. There are
4403 other conditionals in Emacs Lisp, but @code{if} is perhaps the most
4411 (if (= 22 emacs-major-version)
4412 (message "This is version 22 Emacs")
4413 (message "This is not version 22 Emacs"))
4422 The @code{<} function tests whether its first argument is smaller than
4423 its second argument. A corresponding function, @code{>}, tests whether
4424 the first argument is greater than the second. Likewise, @code{<=}
4425 tests whether the first argument is less than or equal to the second and
4426 @code{>=} tests whether the first argument is greater than or equal to
4427 the second. In all cases, both arguments must be numbers or markers
4428 (markers indicate positions in buffers).
4432 The @code{=} function tests whether two arguments, both numbers or
4438 Test whether two objects are the same. @code{equal} uses one meaning
4439 of the word `same' and @code{eq} uses another: @code{equal} returns
4440 true if the two objects have a similar structure and contents, such as
4441 two copies of the same book. On the other hand, @code{eq}, returns
4442 true if both arguments are actually the same object.
4451 The @code{string-lessp} function tests whether its first argument is
4452 smaller than the second argument. A shorter, alternative name for the
4453 same function (a @code{defalias}) is @code{string<}.
4455 The arguments to @code{string-lessp} must be strings or symbols; the
4456 ordering is lexicographic, so case is significant. The print names of
4457 symbols are used instead of the symbols themselves.
4459 @cindex @samp{empty string} defined
4460 An empty string, @samp{""}, a string with no characters in it, is
4461 smaller than any string of characters.
4463 @code{string-equal} provides the corresponding test for equality. Its
4464 shorter, alternative name is @code{string=}. There are no string test
4465 functions that correspond to @var{>}, @code{>=}, or @code{<=}.
4468 Print a message in the echo area. The first argument is a string that
4469 can contain @samp{%s}, @samp{%d}, or @samp{%c} to print the value of
4470 arguments that follow the string. The argument used by @samp{%s} must
4471 be a string or a symbol; the argument used by @samp{%d} must be a
4472 number. The argument used by @samp{%c} must be an @sc{ascii} code
4473 number; it will be printed as the character with that @sc{ascii} code.
4474 (Various other %-sequences have not been mentioned.)
4478 The @code{setq} function sets the value of its first argument to the
4479 value of the second argument. The first argument is automatically
4480 quoted by @code{setq}. It does the same for succeeding pairs of
4481 arguments. Another function, @code{set}, takes only two arguments and
4482 evaluates both of them before setting the value returned by its first
4483 argument to the value returned by its second argument.
4486 Without an argument, return the name of the buffer, as a string.
4488 @item buffer-file-name
4489 Without an argument, return the name of the file the buffer is
4492 @item current-buffer
4493 Return the buffer in which Emacs is active; it may not be
4494 the buffer that is visible on the screen.
4497 Return the most recently selected buffer (other than the buffer passed
4498 to @code{other-buffer} as an argument and other than the current
4501 @item switch-to-buffer
4502 Select a buffer for Emacs to be active in and display it in the current
4503 window so users can look at it. Usually bound to @kbd{C-x b}.
4506 Switch Emacs's attention to a buffer on which programs will run. Don't
4507 alter what the window is showing.
4510 Return the number of characters in the current buffer.
4513 Return the value of the current position of the cursor, as an
4514 integer counting the number of characters from the beginning of the
4518 Return the minimum permissible value of point in
4519 the current buffer. This is 1, unless narrowing is in effect.
4522 Return the value of the maximum permissible value of point in the
4523 current buffer. This is the end of the buffer, unless narrowing is in
4528 @node defun Exercises
4533 Write a non-interactive function that doubles the value of its
4534 argument, a number. Make that function interactive.
4537 Write a function that tests whether the current value of
4538 @code{fill-column} is greater than the argument passed to the function,
4539 and if so, prints an appropriate message.
4542 @node Buffer Walk Through
4543 @chapter A Few Buffer--Related Functions
4545 In this chapter we study in detail several of the functions used in GNU
4546 Emacs. This is called a ``walk-through''. These functions are used as
4547 examples of Lisp code, but are not imaginary examples; with the
4548 exception of the first, simplified function definition, these functions
4549 show the actual code used in GNU Emacs. You can learn a great deal from
4550 these definitions. The functions described here are all related to
4551 buffers. Later, we will study other functions.
4554 * Finding More:: How to find more information.
4555 * simplified-beginning-of-buffer:: Shows @code{goto-char},
4556 @code{point-min}, and @code{push-mark}.
4557 * mark-whole-buffer:: Almost the same as @code{beginning-of-buffer}.
4558 * append-to-buffer:: Uses @code{save-excursion} and
4559 @code{insert-buffer-substring}.
4560 * Buffer Related Review:: Review.
4561 * Buffer Exercises::
4565 @section Finding More Information
4567 @findex describe-function, @r{introduced}
4568 @cindex Find function documentation
4569 In this walk-through, I will describe each new function as we come to
4570 it, sometimes in detail and sometimes briefly. If you are interested,
4571 you can get the full documentation of any Emacs Lisp function at any
4572 time by typing @kbd{C-h f} and then the name of the function (and then
4573 @key{RET}). Similarly, you can get the full documentation for a
4574 variable by typing @kbd{C-h v} and then the name of the variable (and
4577 @cindex Find source of function
4578 @c In version 22, tells location both of C and of Emacs Lisp
4579 Also, @code{describe-function} will tell you the location of the
4580 function definition.
4582 Put point into the name of the file that contains the function and
4583 press the @key{RET} key. In this case, @key{RET} means
4584 @code{push-button} rather than `return' or `enter'. Emacs will take
4585 you directly to the function definition.
4590 If you move point over the file name and press
4591 the @key{RET} key, which in this case means @code{help-follow} rather
4592 than `return' or `enter', Emacs will take you directly to the function
4596 More generally, if you want to see a function in its original source
4597 file, you can use the @code{find-tag} function to jump to it.
4598 @code{find-tag} works with a wide variety of languages, not just
4599 Lisp, and C, and it works with non-programming text as well. For
4600 example, @code{find-tag} will jump to the various nodes in the
4601 Texinfo source file of this document.
4602 The @code{find-tag} function depends on `tags tables' that record
4603 the locations of the functions, variables, and other items to which
4604 @code{find-tag} jumps.
4606 To use the @code{find-tag} command, type @kbd{M-.} (i.e., press the
4607 period key while holding down the @key{META} key, or else type the
4608 @key{ESC} key and then type the period key), and then, at the prompt,
4609 type in the name of the function whose source code you want to see,
4610 such as @code{mark-whole-buffer}, and then type @key{RET}. Emacs will
4611 switch buffers and display the source code for the function on your
4612 screen. To switch back to your current buffer, type @kbd{C-x b
4613 @key{RET}}. (On some keyboards, the @key{META} key is labeled
4616 @c !!! 22.1.1 tags table location in this paragraph
4617 @cindex TAGS table, specifying
4619 Depending on how the initial default values of your copy of Emacs are
4620 set, you may also need to specify the location of your `tags table',
4621 which is a file called @file{TAGS}. For example, if you are
4622 interested in Emacs sources, the tags table you will most likely want,
4623 if it has already been created for you, will be in a subdirectory of
4624 the @file{/usr/local/share/emacs/} directory; thus you would use the
4625 @code{M-x visit-tags-table} command and specify a pathname such as
4626 @file{/usr/local/share/emacs/22.1.1/lisp/TAGS}. If the tags table
4627 has not already been created, you will have to create it yourself. It
4628 will be in a file such as @file{/usr/local/src/emacs/src/TAGS}.
4631 To create a @file{TAGS} file in a specific directory, switch to that
4632 directory in Emacs using @kbd{M-x cd} command, or list the directory
4633 with @kbd{C-x d} (@code{dired}). Then run the compile command, with
4634 @w{@code{etags *.el}} as the command to execute:
4637 M-x compile RET etags *.el RET
4640 For more information, see @ref{etags, , Create Your Own @file{TAGS} File}.
4642 After you become more familiar with Emacs Lisp, you will find that you will
4643 frequently use @code{find-tag} to navigate your way around source code;
4644 and you will create your own @file{TAGS} tables.
4646 @cindex Library, as term for `file'
4647 Incidentally, the files that contain Lisp code are conventionally
4648 called @dfn{libraries}. The metaphor is derived from that of a
4649 specialized library, such as a law library or an engineering library,
4650 rather than a general library. Each library, or file, contains
4651 functions that relate to a particular topic or activity, such as
4652 @file{abbrev.el} for handling abbreviations and other typing
4653 shortcuts, and @file{help.el} for on-line help. (Sometimes several
4654 libraries provide code for a single activity, as the various
4655 @file{rmail@dots{}} files provide code for reading electronic mail.)
4656 In @cite{The GNU Emacs Manual}, you will see sentences such as ``The
4657 @kbd{C-h p} command lets you search the standard Emacs Lisp libraries
4658 by topic keywords.''
4660 @node simplified-beginning-of-buffer
4661 @section A Simplified @code{beginning-of-buffer} Definition
4662 @findex simplified-beginning-of-buffer
4664 The @code{beginning-of-buffer} command is a good function to start with
4665 since you are likely to be familiar with it and it is easy to
4666 understand. Used as an interactive command, @code{beginning-of-buffer}
4667 moves the cursor to the beginning of the buffer, leaving the mark at the
4668 previous position. It is generally bound to @kbd{M-<}.
4670 In this section, we will discuss a shortened version of the function
4671 that shows how it is most frequently used. This shortened function
4672 works as written, but it does not contain the code for a complex option.
4673 In another section, we will describe the entire function.
4674 (@xref{beginning-of-buffer, , Complete Definition of
4675 @code{beginning-of-buffer}}.)
4677 Before looking at the code, let's consider what the function
4678 definition has to contain: it must include an expression that makes
4679 the function interactive so it can be called by typing @kbd{M-x
4680 beginning-of-buffer} or by typing a keychord such as @kbd{M-<}; it
4681 must include code to leave a mark at the original position in the
4682 buffer; and it must include code to move the cursor to the beginning
4686 Here is the complete text of the shortened version of the function:
4690 (defun simplified-beginning-of-buffer ()
4691 "Move point to the beginning of the buffer;
4692 leave mark at previous position."
4695 (goto-char (point-min)))
4699 Like all function definitions, this definition has five parts following
4700 the macro @code{defun}:
4704 The name: in this example, @code{simplified-beginning-of-buffer}.
4707 A list of the arguments: in this example, an empty list, @code{()},
4710 The documentation string.
4713 The interactive expression.
4720 In this function definition, the argument list is empty; this means that
4721 this function does not require any arguments. (When we look at the
4722 definition for the complete function, we will see that it may be passed
4723 an optional argument.)
4725 The interactive expression tells Emacs that the function is intended to
4726 be used interactively. In this example, @code{interactive} does not have
4727 an argument because @code{simplified-beginning-of-buffer} does not
4731 The body of the function consists of the two lines:
4736 (goto-char (point-min))
4740 The first of these lines is the expression, @code{(push-mark)}. When
4741 this expression is evaluated by the Lisp interpreter, it sets a mark at
4742 the current position of the cursor, wherever that may be. The position
4743 of this mark is saved in the mark ring.
4745 The next line is @code{(goto-char (point-min))}. This expression
4746 jumps the cursor to the minimum point in the buffer, that is, to the
4747 beginning of the buffer (or to the beginning of the accessible portion
4748 of the buffer if it is narrowed. @xref{Narrowing & Widening, ,
4749 Narrowing and Widening}.)
4751 The @code{push-mark} command sets a mark at the place where the cursor
4752 was located before it was moved to the beginning of the buffer by the
4753 @code{(goto-char (point-min))} expression. Consequently, you can, if
4754 you wish, go back to where you were originally by typing @kbd{C-x C-x}.
4756 That is all there is to the function definition!
4758 @findex describe-function
4759 When you are reading code such as this and come upon an unfamiliar
4760 function, such as @code{goto-char}, you can find out what it does by
4761 using the @code{describe-function} command. To use this command, type
4762 @kbd{C-h f} and then type in the name of the function and press
4763 @key{RET}. The @code{describe-function} command will print the
4764 function's documentation string in a @file{*Help*} window. For
4765 example, the documentation for @code{goto-char} is:
4769 Set point to POSITION, a number or marker.
4770 Beginning of buffer is position (point-min), end is (point-max).
4775 The function's one argument is the desired position.
4778 (The prompt for @code{describe-function} will offer you the symbol
4779 under or preceding the cursor, so you can save typing by positioning
4780 the cursor right over or after the function and then typing @kbd{C-h f
4783 The @code{end-of-buffer} function definition is written in the same way as
4784 the @code{beginning-of-buffer} definition except that the body of the
4785 function contains the expression @code{(goto-char (point-max))} in place
4786 of @code{(goto-char (point-min))}.
4788 @node mark-whole-buffer
4789 @section The Definition of @code{mark-whole-buffer}
4790 @findex mark-whole-buffer
4792 The @code{mark-whole-buffer} function is no harder to understand than the
4793 @code{simplified-beginning-of-buffer} function. In this case, however,
4794 we will look at the complete function, not a shortened version.
4796 The @code{mark-whole-buffer} function is not as commonly used as the
4797 @code{beginning-of-buffer} function, but is useful nonetheless: it
4798 marks a whole buffer as a region by putting point at the beginning and
4799 a mark at the end of the buffer. It is generally bound to @kbd{C-x
4803 * mark-whole-buffer overview::
4804 * Body of mark-whole-buffer:: Only three lines of code.
4808 @node mark-whole-buffer overview
4809 @unnumberedsubsec An overview of @code{mark-whole-buffer}
4813 In GNU Emacs 22, the code for the complete function looks like this:
4817 (defun mark-whole-buffer ()
4818 "Put point at beginning and mark at end of buffer.
4819 You probably should not use this function in Lisp programs;
4820 it is usually a mistake for a Lisp function to use any subroutine
4821 that uses or sets the mark."
4824 (push-mark (point-max) nil t)
4825 (goto-char (point-min)))
4830 Like all other functions, the @code{mark-whole-buffer} function fits
4831 into the template for a function definition. The template looks like
4836 (defun @var{name-of-function} (@var{argument-list})
4837 "@var{documentation}@dots{}"
4838 (@var{interactive-expression}@dots{})
4843 Here is how the function works: the name of the function is
4844 @code{mark-whole-buffer}; it is followed by an empty argument list,
4845 @samp{()}, which means that the function does not require arguments.
4846 The documentation comes next.
4848 The next line is an @code{(interactive)} expression that tells Emacs
4849 that the function will be used interactively. These details are similar
4850 to the @code{simplified-beginning-of-buffer} function described in the
4854 @node Body of mark-whole-buffer
4855 @subsection Body of @code{mark-whole-buffer}
4857 The body of the @code{mark-whole-buffer} function consists of three
4864 (push-mark (point-max) nil t)
4865 (goto-char (point-min))
4869 The first of these lines is the expression, @code{(push-mark (point))}.
4871 This line does exactly the same job as the first line of the body of
4872 the @code{simplified-beginning-of-buffer} function, which is written
4873 @code{(push-mark)}. In both cases, the Lisp interpreter sets a mark
4874 at the current position of the cursor.
4876 I don't know why the expression in @code{mark-whole-buffer} is written
4877 @code{(push-mark (point))} and the expression in
4878 @code{beginning-of-buffer} is written @code{(push-mark)}. Perhaps
4879 whoever wrote the code did not know that the arguments for
4880 @code{push-mark} are optional and that if @code{push-mark} is not
4881 passed an argument, the function automatically sets mark at the
4882 location of point by default. Or perhaps the expression was written
4883 so as to parallel the structure of the next line. In any case, the
4884 line causes Emacs to determine the position of point and set a mark
4887 In earlier versions of GNU Emacs, the next line of
4888 @code{mark-whole-buffer} was @code{(push-mark (point-max))}. This
4889 expression sets a mark at the point in the buffer that has the highest
4890 number. This will be the end of the buffer (or, if the buffer is
4891 narrowed, the end of the accessible portion of the buffer.
4892 @xref{Narrowing & Widening, , Narrowing and Widening}, for more about
4893 narrowing.) After this mark has been set, the previous mark, the one
4894 set at point, is no longer set, but Emacs remembers its position, just
4895 as all other recent marks are always remembered. This means that you
4896 can, if you wish, go back to that position by typing @kbd{C-u
4900 In GNU Emacs 22, the @code{(point-max)} is slightly more complicated.
4904 (push-mark (point-max) nil t)
4908 The expression works nearly the same as before. It sets a mark at the
4909 highest numbered place in the buffer that it can. However, in this
4910 version, @code{push-mark} has two additional arguments. The second
4911 argument to @code{push-mark} is @code{nil}. This tells the function
4912 it @emph{should} display a message that says `Mark set' when it pushes
4913 the mark. The third argument is @code{t}. This tells
4914 @code{push-mark} to activate the mark when Transient Mark mode is
4915 turned on. Transient Mark mode highlights the currently active
4916 region. It is often turned off.
4918 Finally, the last line of the function is @code{(goto-char
4919 (point-min)))}. This is written exactly the same way as it is written
4920 in @code{beginning-of-buffer}. The expression moves the cursor to
4921 the minimum point in the buffer, that is, to the beginning of the buffer
4922 (or to the beginning of the accessible portion of the buffer). As a
4923 result of this, point is placed at the beginning of the buffer and mark
4924 is set at the end of the buffer. The whole buffer is, therefore, the
4927 @node append-to-buffer
4928 @section The Definition of @code{append-to-buffer}
4929 @findex append-to-buffer
4931 The @code{append-to-buffer} command is more complex than the
4932 @code{mark-whole-buffer} command. What it does is copy the region
4933 (that is, the part of the buffer between point and mark) from the
4934 current buffer to a specified buffer.
4937 * append-to-buffer overview::
4938 * append interactive:: A two part interactive expression.
4939 * append-to-buffer body:: Incorporates a @code{let} expression.
4940 * append save-excursion:: How the @code{save-excursion} works.
4944 @node append-to-buffer overview
4945 @unnumberedsubsec An Overview of @code{append-to-buffer}
4948 @findex insert-buffer-substring
4949 The @code{append-to-buffer} command uses the
4950 @code{insert-buffer-substring} function to copy the region.
4951 @code{insert-buffer-substring} is described by its name: it takes a
4952 string of characters from part of a buffer, a ``substring'', and
4953 inserts them into another buffer.
4955 Most of @code{append-to-buffer} is
4956 concerned with setting up the conditions for
4957 @code{insert-buffer-substring} to work: the code must specify both the
4958 buffer to which the text will go, the window it comes from and goes
4959 to, and the region that will be copied.
4962 Here is the complete text of the function:
4966 (defun append-to-buffer (buffer start end)
4967 "Append to specified buffer the text of the region.
4968 It is inserted into that buffer before its point.
4972 When calling from a program, give three arguments:
4973 BUFFER (or buffer name), START and END.
4974 START and END specify the portion of the current buffer to be copied."
4976 (list (read-buffer "Append to buffer: " (other-buffer
4977 (current-buffer) t))
4978 (region-beginning) (region-end)))
4981 (let ((oldbuf (current-buffer)))
4983 (let* ((append-to (get-buffer-create buffer))
4984 (windows (get-buffer-window-list append-to t t))
4986 (set-buffer append-to)
4987 (setq point (point))
4988 (barf-if-buffer-read-only)
4989 (insert-buffer-substring oldbuf start end)
4990 (dolist (window windows)
4991 (when (= (window-point window) point)
4992 (set-window-point window (point))))))))
4996 The function can be understood by looking at it as a series of
4997 filled-in templates.
4999 The outermost template is for the function definition. In this
5000 function, it looks like this (with several slots filled in):
5004 (defun append-to-buffer (buffer start end)
5005 "@var{documentation}@dots{}"
5006 (interactive @dots{})
5011 The first line of the function includes its name and three arguments.
5012 The arguments are the @code{buffer} to which the text will be copied, and
5013 the @code{start} and @code{end} of the region in the current buffer that
5016 The next part of the function is the documentation, which is clear and
5017 complete. As is conventional, the three arguments are written in
5018 upper case so you will notice them easily. Even better, they are
5019 described in the same order as in the argument list.
5021 Note that the documentation distinguishes between a buffer and its
5022 name. (The function can handle either.)
5024 @node append interactive
5025 @subsection The @code{append-to-buffer} Interactive Expression
5027 Since the @code{append-to-buffer} function will be used interactively,
5028 the function must have an @code{interactive} expression. (For a
5029 review of @code{interactive}, see @ref{Interactive, , Making a
5030 Function Interactive}.) The expression reads as follows:
5036 "Append to buffer: "
5037 (other-buffer (current-buffer) t))
5044 This expression is not one with letters standing for parts, as
5045 described earlier. Instead, it starts a list with these parts:
5047 The first part of the list is an expression to read the name of a
5048 buffer and return it as a string. That is @code{read-buffer}. The
5049 function requires a prompt as its first argument, @samp{"Append to
5050 buffer: "}. Its second argument tells the command what value to
5051 provide if you don't specify anything.
5053 In this case that second argument is an expression containing the
5054 function @code{other-buffer}, an exception, and a @samp{t}, standing
5057 The first argument to @code{other-buffer}, the exception, is yet
5058 another function, @code{current-buffer}. That is not going to be
5059 returned. The second argument is the symbol for true, @code{t}. that
5060 tells @code{other-buffer} that it may show visible buffers (except in
5061 this case, it will not show the current buffer, which makes sense).
5064 The expression looks like this:
5067 (other-buffer (current-buffer) t)
5070 The second and third arguments to the @code{list} expression are
5071 @code{(region-beginning)} and @code{(region-end)}. These two
5072 functions specify the beginning and end of the text to be appended.
5075 Originally, the command used the letters @samp{B} and @samp{r}.
5076 The whole @code{interactive} expression looked like this:
5079 (interactive "BAppend to buffer:@: \nr")
5083 But when that was done, the default value of the buffer switched to
5084 was invisible. That was not wanted.
5086 (The prompt was separated from the second argument with a newline,
5087 @samp{\n}. It was followed by an @samp{r} that told Emacs to bind the
5088 two arguments that follow the symbol @code{buffer} in the function's
5089 argument list (that is, @code{start} and @code{end}) to the values of
5090 point and mark. That argument worked fine.)
5092 @node append-to-buffer body
5093 @subsection The Body of @code{append-to-buffer}
5096 in GNU Emacs 22 in /usr/local/src/emacs/lisp/simple.el
5098 (defun append-to-buffer (buffer start end)
5099 "Append to specified buffer the text of the region.
5100 It is inserted into that buffer before its point.
5102 When calling from a program, give three arguments:
5103 BUFFER (or buffer name), START and END.
5104 START and END specify the portion of the current buffer to be copied."
5106 (list (read-buffer "Append to buffer: " (other-buffer (current-buffer) t))
5107 (region-beginning) (region-end)))
5108 (let ((oldbuf (current-buffer)))
5110 (let* ((append-to (get-buffer-create buffer))
5111 (windows (get-buffer-window-list append-to t t))
5113 (set-buffer append-to)
5114 (setq point (point))
5115 (barf-if-buffer-read-only)
5116 (insert-buffer-substring oldbuf start end)
5117 (dolist (window windows)
5118 (when (= (window-point window) point)
5119 (set-window-point window (point))))))))
5122 The body of the @code{append-to-buffer} function begins with @code{let}.
5124 As we have seen before (@pxref{let, , @code{let}}), the purpose of a
5125 @code{let} expression is to create and give initial values to one or
5126 more variables that will only be used within the body of the
5127 @code{let}. This means that such a variable will not be confused with
5128 any variable of the same name outside the @code{let} expression.
5130 We can see how the @code{let} expression fits into the function as a
5131 whole by showing a template for @code{append-to-buffer} with the
5132 @code{let} expression in outline:
5136 (defun append-to-buffer (buffer start end)
5137 "@var{documentation}@dots{}"
5138 (interactive @dots{})
5139 (let ((@var{variable} @var{value}))
5144 The @code{let} expression has three elements:
5148 The symbol @code{let};
5151 A varlist containing, in this case, a single two-element list,
5152 @code{(@var{variable} @var{value})};
5155 The body of the @code{let} expression.
5159 In the @code{append-to-buffer} function, the varlist looks like this:
5162 (oldbuf (current-buffer))
5166 In this part of the @code{let} expression, the one variable,
5167 @code{oldbuf}, is bound to the value returned by the
5168 @code{(current-buffer)} expression. The variable, @code{oldbuf}, is
5169 used to keep track of the buffer in which you are working and from
5170 which you will copy.
5172 The element or elements of a varlist are surrounded by a set of
5173 parentheses so the Lisp interpreter can distinguish the varlist from
5174 the body of the @code{let}. As a consequence, the two-element list
5175 within the varlist is surrounded by a circumscribing set of parentheses.
5176 The line looks like this:
5180 (let ((oldbuf (current-buffer)))
5186 The two parentheses before @code{oldbuf} might surprise you if you did
5187 not realize that the first parenthesis before @code{oldbuf} marks the
5188 boundary of the varlist and the second parenthesis marks the beginning
5189 of the two-element list, @code{(oldbuf (current-buffer))}.
5191 @node append save-excursion
5192 @subsection @code{save-excursion} in @code{append-to-buffer}
5194 The body of the @code{let} expression in @code{append-to-buffer}
5195 consists of a @code{save-excursion} expression.
5197 The @code{save-excursion} function saves the locations of point and
5198 mark, and restores them to those positions after the expressions in the
5199 body of the @code{save-excursion} complete execution. In addition,
5200 @code{save-excursion} keeps track of the original buffer, and
5201 restores it. This is how @code{save-excursion} is used in
5202 @code{append-to-buffer}.
5205 @cindex Indentation for formatting
5206 @cindex Formatting convention
5207 Incidentally, it is worth noting here that a Lisp function is normally
5208 formatted so that everything that is enclosed in a multi-line spread is
5209 indented more to the right than the first symbol. In this function
5210 definition, the @code{let} is indented more than the @code{defun}, and
5211 the @code{save-excursion} is indented more than the @code{let}, like
5227 This formatting convention makes it easy to see that the lines in
5228 the body of the @code{save-excursion} are enclosed by the parentheses
5229 associated with @code{save-excursion}, just as the
5230 @code{save-excursion} itself is enclosed by the parentheses associated
5231 with the @code{let}:
5235 (let ((oldbuf (current-buffer)))
5238 (set-buffer @dots{})
5239 (insert-buffer-substring oldbuf start end)
5245 The use of the @code{save-excursion} function can be viewed as a process
5246 of filling in the slots of a template:
5251 @var{first-expression-in-body}
5252 @var{second-expression-in-body}
5254 @var{last-expression-in-body})
5260 In this function, the body of the @code{save-excursion} contains only
5261 one expression, the @code{let*} expression. You know about a
5262 @code{let} function. The @code{let*} function is different. It has a
5263 @samp{*} in its name. It enables Emacs to set each variable in its
5264 varlist in sequence, one after another.
5266 Its critical feature is that variables later in the varlist can make
5267 use of the values to which Emacs set variables earlier in the varlist.
5268 @xref{fwd-para let, , The @code{let*} expression}.
5270 We will skip functions like @code{let*} and focus on two: the
5271 @code{set-buffer} function and the @code{insert-buffer-substring}
5275 In the old days, the @code{set-buffer} expression was simply
5278 (set-buffer (get-buffer-create buffer))
5286 (set-buffer append-to)
5290 @code{append-to} is bound to @code{(get-buffer-create buffer)} earlier
5291 on in the @code{let*} expression. That extra binding would not be
5292 necessary except for that @code{append-to} is used later in the
5293 varlist as an argument to @code{get-buffer-window-list}.
5298 (let ((oldbuf (current-buffer)))
5300 (let* ((append-to (get-buffer-create buffer))
5301 (windows (get-buffer-window-list append-to t t))
5303 (set-buffer append-to)
5304 (setq point (point))
5305 (barf-if-buffer-read-only)
5306 (insert-buffer-substring oldbuf start end)
5307 (dolist (window windows)
5308 (when (= (window-point window) point)
5309 (set-window-point window (point))))))))
5312 The @code{append-to-buffer} function definition inserts text from the
5313 buffer in which you are currently to a named buffer. It happens that
5314 @code{insert-buffer-substring} copies text from another buffer to the
5315 current buffer, just the reverse---that is why the
5316 @code{append-to-buffer} definition starts out with a @code{let} that
5317 binds the local symbol @code{oldbuf} to the value returned by
5318 @code{current-buffer}.
5321 The @code{insert-buffer-substring} expression looks like this:
5324 (insert-buffer-substring oldbuf start end)
5328 The @code{insert-buffer-substring} function copies a string
5329 @emph{from} the buffer specified as its first argument and inserts the
5330 string into the present buffer. In this case, the argument to
5331 @code{insert-buffer-substring} is the value of the variable created
5332 and bound by the @code{let}, namely the value of @code{oldbuf}, which
5333 was the current buffer when you gave the @code{append-to-buffer}
5336 After @code{insert-buffer-substring} has done its work,
5337 @code{save-excursion} will restore the action to the original buffer
5338 and @code{append-to-buffer} will have done its job.
5341 Written in skeletal form, the workings of the body look like this:
5345 (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer})
5346 (save-excursion ; @r{Keep track of buffer.}
5348 @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer})
5350 @var{change-back-to-original-buffer-when-finished}
5351 @var{let-the-local-meaning-of-}@code{oldbuf}@var{-disappear-when-finished}
5355 In summary, @code{append-to-buffer} works as follows: it saves the
5356 value of the current buffer in the variable called @code{oldbuf}. It
5357 gets the new buffer (creating one if need be) and switches Emacs's
5358 attention to it. Using the value of @code{oldbuf}, it inserts the
5359 region of text from the old buffer into the new buffer; and then using
5360 @code{save-excursion}, it brings you back to your original buffer.
5362 In looking at @code{append-to-buffer}, you have explored a fairly
5363 complex function. It shows how to use @code{let} and
5364 @code{save-excursion}, and how to change to and come back from another
5365 buffer. Many function definitions use @code{let},
5366 @code{save-excursion}, and @code{set-buffer} this way.
5368 @node Buffer Related Review
5371 Here is a brief summary of the various functions discussed in this chapter.
5374 @item describe-function
5375 @itemx describe-variable
5376 Print the documentation for a function or variable.
5377 Conventionally bound to @kbd{C-h f} and @kbd{C-h v}.
5380 Find the file containing the source for a function or variable and
5381 switch buffers to it, positioning point at the beginning of the item.
5382 Conventionally bound to @kbd{M-.} (that's a period following the
5385 @item save-excursion
5386 Save the location of point and mark and restore their values after the
5387 arguments to @code{save-excursion} have been evaluated. Also, remember
5388 the current buffer and return to it.
5391 Set mark at a location and record the value of the previous mark on the
5392 mark ring. The mark is a location in the buffer that will keep its
5393 relative position even if text is added to or removed from the buffer.
5396 Set point to the location specified by the value of the argument, which
5397 can be a number, a marker, or an expression that returns the number of
5398 a position, such as @code{(point-min)}.
5400 @item insert-buffer-substring
5401 Copy a region of text from a buffer that is passed to the function as
5402 an argument and insert the region into the current buffer.
5404 @item mark-whole-buffer
5405 Mark the whole buffer as a region. Normally bound to @kbd{C-x h}.
5408 Switch the attention of Emacs to another buffer, but do not change the
5409 window being displayed. Used when the program rather than a human is
5410 to work on a different buffer.
5412 @item get-buffer-create
5414 Find a named buffer or create one if a buffer of that name does not
5415 exist. The @code{get-buffer} function returns @code{nil} if the named
5416 buffer does not exist.
5420 @node Buffer Exercises
5425 Write your own @code{simplified-end-of-buffer} function definition;
5426 then test it to see whether it works.
5429 Use @code{if} and @code{get-buffer} to write a function that prints a
5430 message telling you whether a buffer exists.
5433 Using @code{find-tag}, find the source for the @code{copy-to-buffer}
5438 @chapter A Few More Complex Functions
5440 In this chapter, we build on what we have learned in previous chapters
5441 by looking at more complex functions. The @code{copy-to-buffer}
5442 function illustrates use of two @code{save-excursion} expressions in
5443 one definition, while the @code{insert-buffer} function illustrates
5444 use of an asterisk in an @code{interactive} expression, use of
5445 @code{or}, and the important distinction between a name and the object
5446 to which the name refers.
5449 * copy-to-buffer:: With @code{set-buffer}, @code{get-buffer-create}.
5450 * insert-buffer:: Read-only, and with @code{or}.
5451 * beginning-of-buffer:: Shows @code{goto-char},
5452 @code{point-min}, and @code{push-mark}.
5453 * Second Buffer Related Review::
5454 * optional Exercise::
5457 @node copy-to-buffer
5458 @section The Definition of @code{copy-to-buffer}
5459 @findex copy-to-buffer
5461 After understanding how @code{append-to-buffer} works, it is easy to
5462 understand @code{copy-to-buffer}. This function copies text into a
5463 buffer, but instead of adding to the second buffer, it replaces all the
5464 previous text in the second buffer.
5467 The body of @code{copy-to-buffer} looks like this,
5472 (interactive "BCopy to buffer: \nr")
5473 (let ((oldbuf (current-buffer)))
5474 (with-current-buffer (get-buffer-create buffer)
5475 (barf-if-buffer-read-only)
5478 (insert-buffer-substring oldbuf start end)))))
5482 The @code{copy-to-buffer} function has a simpler @code{interactive}
5483 expression than @code{append-to-buffer}.
5486 The definition then says
5489 (with-current-buffer (get-buffer-create buffer) @dots{}
5492 First, look at the earliest inner expression; that is evaluated first.
5493 That expression starts with @code{get-buffer-create buffer}. The
5494 function tells the computer to use the buffer with the name specified
5495 as the one to which you are copying, or if such a buffer does not
5496 exist, to create it. Then, the @code{with-current-buffer} function
5497 evaluates its body with that buffer temporarily current.
5499 (This demonstrates another way to shift the computer's attention but
5500 not the user's. The @code{append-to-buffer} function showed how to do
5501 the same with @code{save-excursion} and @code{set-buffer}.
5502 @code{with-current-buffer} is a newer, and arguably easier,
5505 The @code{barf-if-buffer-read-only} function sends you an error
5506 message saying the buffer is read-only if you cannot modify it.
5508 The next line has the @code{erase-buffer} function as its sole
5509 contents. That function erases the buffer.
5511 Finally, the last two lines contain the @code{save-excursion}
5512 expression with @code{insert-buffer-substring} as its body.
5513 The @code{insert-buffer-substring} expression copies the text from
5514 the buffer you are in (and you have not seen the computer shift its
5515 attention, so you don't know that that buffer is now called
5518 Incidentally, this is what is meant by `replacement'. To replace text,
5519 Emacs erases the previous text and then inserts new text.
5522 In outline, the body of @code{copy-to-buffer} looks like this:
5526 (let (@var{bind-}@code{oldbuf}@var{-to-value-of-}@code{current-buffer})
5527 (@var{with-the-buffer-you-are-copying-to}
5528 (@var{but-do-not-erase-or-copy-to-a-read-only-buffer})
5531 @var{insert-substring-from-}@code{oldbuf}@var{-into-buffer})))
5536 @section The Definition of @code{insert-buffer}
5537 @findex insert-buffer
5539 @code{insert-buffer} is yet another buffer-related function. This
5540 command copies another buffer @emph{into} the current buffer. It is the
5541 reverse of @code{append-to-buffer} or @code{copy-to-buffer}, since they
5542 copy a region of text @emph{from} the current buffer to another buffer.
5544 Here is a discussion based on the original code. The code was
5545 simplified in 2003 and is harder to understand.
5547 (@xref{New insert-buffer, , New Body for @code{insert-buffer}}, to see
5548 a discussion of the new body.)
5550 In addition, this code illustrates the use of @code{interactive} with a
5551 buffer that might be @dfn{read-only} and the important distinction
5552 between the name of an object and the object actually referred to.
5555 * insert-buffer code::
5556 * insert-buffer interactive:: When you can read, but not write.
5557 * insert-buffer body:: The body has an @code{or} and a @code{let}.
5558 * if & or:: Using an @code{if} instead of an @code{or}.
5559 * Insert or:: How the @code{or} expression works.
5560 * Insert let:: Two @code{save-excursion} expressions.
5561 * New insert-buffer::
5565 @node insert-buffer code
5566 @unnumberedsubsec The Code for @code{insert-buffer}
5570 Here is the earlier code:
5574 (defun insert-buffer (buffer)
5575 "Insert after point the contents of BUFFER.
5576 Puts mark after the inserted text.
5577 BUFFER may be a buffer or a buffer name."
5578 (interactive "*bInsert buffer:@: ")
5581 (or (bufferp buffer)
5582 (setq buffer (get-buffer buffer)))
5583 (let (start end newmark)
5587 (setq start (point-min) end (point-max)))
5590 (insert-buffer-substring buffer start end)
5591 (setq newmark (point)))
5592 (push-mark newmark)))
5597 As with other function definitions, you can use a template to see an
5598 outline of the function:
5602 (defun insert-buffer (buffer)
5603 "@var{documentation}@dots{}"
5604 (interactive "*bInsert buffer:@: ")
5609 @node insert-buffer interactive
5610 @subsection The Interactive Expression in @code{insert-buffer}
5611 @findex interactive, @r{example use of}
5613 In @code{insert-buffer}, the argument to the @code{interactive}
5614 declaration has two parts, an asterisk, @samp{*}, and @samp{bInsert
5618 * Read-only buffer:: When a buffer cannot be modified.
5619 * b for interactive:: An existing buffer or else its name.
5622 @node Read-only buffer
5623 @unnumberedsubsubsec A Read-only Buffer
5624 @cindex Read-only buffer
5625 @cindex Asterisk for read-only buffer
5626 @findex * @r{for read-only buffer}
5628 The asterisk is for the situation when the current buffer is a
5629 read-only buffer---a buffer that cannot be modified. If
5630 @code{insert-buffer} is called when the current buffer is read-only, a
5631 message to this effect is printed in the echo area and the terminal
5632 may beep or blink at you; you will not be permitted to insert anything
5633 into current buffer. The asterisk does not need to be followed by a
5634 newline to separate it from the next argument.
5636 @node b for interactive
5637 @unnumberedsubsubsec @samp{b} in an Interactive Expression
5639 The next argument in the interactive expression starts with a lower
5640 case @samp{b}. (This is different from the code for
5641 @code{append-to-buffer}, which uses an upper-case @samp{B}.
5642 @xref{append-to-buffer, , The Definition of @code{append-to-buffer}}.)
5643 The lower-case @samp{b} tells the Lisp interpreter that the argument
5644 for @code{insert-buffer} should be an existing buffer or else its
5645 name. (The upper-case @samp{B} option provides for the possibility
5646 that the buffer does not exist.) Emacs will prompt you for the name
5647 of the buffer, offering you a default buffer, with name completion
5648 enabled. If the buffer does not exist, you receive a message that
5649 says ``No match''; your terminal may beep at you as well.
5651 The new and simplified code generates a list for @code{interactive}.
5652 It uses the @code{barf-if-buffer-read-only} and @code{read-buffer}
5653 functions with which we are already familiar and the @code{progn}
5654 special form with which we are not. (It will be described later.)
5656 @node insert-buffer body
5657 @subsection The Body of the @code{insert-buffer} Function
5659 The body of the @code{insert-buffer} function has two major parts: an
5660 @code{or} expression and a @code{let} expression. The purpose of the
5661 @code{or} expression is to ensure that the argument @code{buffer} is
5662 bound to a buffer and not just the name of a buffer. The body of the
5663 @code{let} expression contains the code which copies the other buffer
5664 into the current buffer.
5667 In outline, the two expressions fit into the @code{insert-buffer}
5672 (defun insert-buffer (buffer)
5673 "@var{documentation}@dots{}"
5674 (interactive "*bInsert buffer:@: ")
5679 (let (@var{varlist})
5680 @var{body-of-}@code{let}@dots{} )
5684 To understand how the @code{or} expression ensures that the argument
5685 @code{buffer} is bound to a buffer and not to the name of a buffer, it
5686 is first necessary to understand the @code{or} function.
5688 Before doing this, let me rewrite this part of the function using
5689 @code{if} so that you can see what is done in a manner that will be familiar.
5692 @subsection @code{insert-buffer} With an @code{if} Instead of an @code{or}
5694 The job to be done is to make sure the value of @code{buffer} is a
5695 buffer itself and not the name of a buffer. If the value is the name,
5696 then the buffer itself must be got.
5698 You can imagine yourself at a conference where an usher is wandering
5699 around holding a list with your name on it and looking for you: the
5700 usher is ``bound'' to your name, not to you; but when the usher finds
5701 you and takes your arm, the usher becomes ``bound'' to you.
5704 In Lisp, you might describe this situation like this:
5708 (if (not (holding-on-to-guest))
5709 (find-and-take-arm-of-guest))
5713 We want to do the same thing with a buffer---if we do not have the
5714 buffer itself, we want to get it.
5717 Using a predicate called @code{bufferp} that tells us whether we have a
5718 buffer (rather than its name), we can write the code like this:
5722 (if (not (bufferp buffer)) ; @r{if-part}
5723 (setq buffer (get-buffer buffer))) ; @r{then-part}
5728 Here, the true-or-false-test of the @code{if} expression is
5729 @w{@code{(not (bufferp buffer))}}; and the then-part is the expression
5730 @w{@code{(setq buffer (get-buffer buffer))}}.
5732 In the test, the function @code{bufferp} returns true if its argument is
5733 a buffer---but false if its argument is the name of the buffer. (The
5734 last character of the function name @code{bufferp} is the character
5735 @samp{p}; as we saw earlier, such use of @samp{p} is a convention that
5736 indicates that the function is a predicate, which is a term that means
5737 that the function will determine whether some property is true or false.
5738 @xref{Wrong Type of Argument, , Using the Wrong Type Object as an
5742 The function @code{not} precedes the expression @code{(bufferp buffer)},
5743 so the true-or-false-test looks like this:
5746 (not (bufferp buffer))
5750 @code{not} is a function that returns true if its argument is false
5751 and false if its argument is true. So if @code{(bufferp buffer)}
5752 returns true, the @code{not} expression returns false and vice-verse:
5753 what is ``not true'' is false and what is ``not false'' is true.
5755 Using this test, the @code{if} expression works as follows: when the
5756 value of the variable @code{buffer} is actually a buffer rather than
5757 its name, the true-or-false-test returns false and the @code{if}
5758 expression does not evaluate the then-part. This is fine, since we do
5759 not need to do anything to the variable @code{buffer} if it really is
5762 On the other hand, when the value of @code{buffer} is not a buffer
5763 itself, but the name of a buffer, the true-or-false-test returns true
5764 and the then-part of the expression is evaluated. In this case, the
5765 then-part is @code{(setq buffer (get-buffer buffer))}. This
5766 expression uses the @code{get-buffer} function to return an actual
5767 buffer itself, given its name. The @code{setq} then sets the variable
5768 @code{buffer} to the value of the buffer itself, replacing its previous
5769 value (which was the name of the buffer).
5772 @subsection The @code{or} in the Body
5774 The purpose of the @code{or} expression in the @code{insert-buffer}
5775 function is to ensure that the argument @code{buffer} is bound to a
5776 buffer and not just to the name of a buffer. The previous section shows
5777 how the job could have been done using an @code{if} expression.
5778 However, the @code{insert-buffer} function actually uses @code{or}.
5779 To understand this, it is necessary to understand how @code{or} works.
5782 An @code{or} function can have any number of arguments. It evaluates
5783 each argument in turn and returns the value of the first of its
5784 arguments that is not @code{nil}. Also, and this is a crucial feature
5785 of @code{or}, it does not evaluate any subsequent arguments after
5786 returning the first non-@code{nil} value.
5789 The @code{or} expression looks like this:
5793 (or (bufferp buffer)
5794 (setq buffer (get-buffer buffer)))
5799 The first argument to @code{or} is the expression @code{(bufferp buffer)}.
5800 This expression returns true (a non-@code{nil} value) if the buffer is
5801 actually a buffer, and not just the name of a buffer. In the @code{or}
5802 expression, if this is the case, the @code{or} expression returns this
5803 true value and does not evaluate the next expression---and this is fine
5804 with us, since we do not want to do anything to the value of
5805 @code{buffer} if it really is a buffer.
5807 On the other hand, if the value of @code{(bufferp buffer)} is @code{nil},
5808 which it will be if the value of @code{buffer} is the name of a buffer,
5809 the Lisp interpreter evaluates the next element of the @code{or}
5810 expression. This is the expression @code{(setq buffer (get-buffer
5811 buffer))}. This expression returns a non-@code{nil} value, which
5812 is the value to which it sets the variable @code{buffer}---and this
5813 value is a buffer itself, not the name of a buffer.
5815 The result of all this is that the symbol @code{buffer} is always
5816 bound to a buffer itself rather than to the name of a buffer. All
5817 this is necessary because the @code{set-buffer} function in a
5818 following line only works with a buffer itself, not with the name to a
5822 Incidentally, using @code{or}, the situation with the usher would be
5826 (or (holding-on-to-guest) (find-and-take-arm-of-guest))
5830 @subsection The @code{let} Expression in @code{insert-buffer}
5832 After ensuring that the variable @code{buffer} refers to a buffer itself
5833 and not just to the name of a buffer, the @code{insert-buffer function}
5834 continues with a @code{let} expression. This specifies three local
5835 variables, @code{start}, @code{end}, and @code{newmark} and binds them
5836 to the initial value @code{nil}. These variables are used inside the
5837 remainder of the @code{let} and temporarily hide any other occurrence of
5838 variables of the same name in Emacs until the end of the @code{let}.
5841 The body of the @code{let} contains two @code{save-excursion}
5842 expressions. First, we will look at the inner @code{save-excursion}
5843 expression in detail. The expression looks like this:
5849 (setq start (point-min) end (point-max)))
5854 The expression @code{(set-buffer buffer)} changes Emacs's attention
5855 from the current buffer to the one from which the text will copied.
5856 In that buffer, the variables @code{start} and @code{end} are set to
5857 the beginning and end of the buffer, using the commands
5858 @code{point-min} and @code{point-max}. Note that we have here an
5859 illustration of how @code{setq} is able to set two variables in the
5860 same expression. The first argument of @code{setq} is set to the
5861 value of its second, and its third argument is set to the value of its
5864 After the body of the inner @code{save-excursion} is evaluated, the
5865 @code{save-excursion} restores the original buffer, but @code{start} and
5866 @code{end} remain set to the values of the beginning and end of the
5867 buffer from which the text will be copied.
5870 The outer @code{save-excursion} expression looks like this:
5875 (@var{inner-}@code{save-excursion}@var{-expression}
5876 (@var{go-to-new-buffer-and-set-}@code{start}@var{-and-}@code{end})
5877 (insert-buffer-substring buffer start end)
5878 (setq newmark (point)))
5883 The @code{insert-buffer-substring} function copies the text
5884 @emph{into} the current buffer @emph{from} the region indicated by
5885 @code{start} and @code{end} in @code{buffer}. Since the whole of the
5886 second buffer lies between @code{start} and @code{end}, the whole of
5887 the second buffer is copied into the buffer you are editing. Next,
5888 the value of point, which will be at the end of the inserted text, is
5889 recorded in the variable @code{newmark}.
5891 After the body of the outer @code{save-excursion} is evaluated, point
5892 and mark are relocated to their original places.
5894 However, it is convenient to locate a mark at the end of the newly
5895 inserted text and locate point at its beginning. The @code{newmark}
5896 variable records the end of the inserted text. In the last line of
5897 the @code{let} expression, the @code{(push-mark newmark)} expression
5898 function sets a mark to this location. (The previous location of the
5899 mark is still accessible; it is recorded on the mark ring and you can
5900 go back to it with @kbd{C-u C-@key{SPC}}.) Meanwhile, point is
5901 located at the beginning of the inserted text, which is where it was
5902 before you called the insert function, the position of which was saved
5903 by the first @code{save-excursion}.
5906 The whole @code{let} expression looks like this:
5910 (let (start end newmark)
5914 (setq start (point-min) end (point-max)))
5915 (insert-buffer-substring buffer start end)
5916 (setq newmark (point)))
5917 (push-mark newmark))
5921 Like the @code{append-to-buffer} function, the @code{insert-buffer}
5922 function uses @code{let}, @code{save-excursion}, and
5923 @code{set-buffer}. In addition, the function illustrates one way to
5924 use @code{or}. All these functions are building blocks that we will
5925 find and use again and again.
5927 @node New insert-buffer
5928 @subsection New Body for @code{insert-buffer}
5929 @findex insert-buffer, new version body
5930 @findex new version body for insert-buffer
5932 The body in the GNU Emacs 22 version is more confusing than the original.
5935 It consists of two expressions,
5941 (insert-buffer-substring (get-buffer buffer))
5949 except, and this is what confuses novices, very important work is done
5950 inside the @code{push-mark} expression.
5952 The @code{get-buffer} function returns a buffer with the name
5953 provided. You will note that the function is @emph{not} called
5954 @code{get-buffer-create}; it does not create a buffer if one does not
5955 already exist. The buffer returned by @code{get-buffer}, an existing
5956 buffer, is passed to @code{insert-buffer-substring}, which inserts the
5957 whole of the buffer (since you did not specify anything else).
5959 The location into which the buffer is inserted is recorded by
5960 @code{push-mark}. Then the function returns @code{nil}, the value of
5961 its last command. Put another way, the @code{insert-buffer} function
5962 exists only to produce a side effect, inserting another buffer, not to
5965 @node beginning-of-buffer
5966 @section Complete Definition of @code{beginning-of-buffer}
5967 @findex beginning-of-buffer
5969 The basic structure of the @code{beginning-of-buffer} function has
5970 already been discussed. (@xref{simplified-beginning-of-buffer, , A
5971 Simplified @code{beginning-of-buffer} Definition}.)
5972 This section describes the complex part of the definition.
5974 As previously described, when invoked without an argument,
5975 @code{beginning-of-buffer} moves the cursor to the beginning of the
5976 buffer (in truth, the beginning of the accessible portion of the
5977 buffer), leaving the mark at the previous position. However, when the
5978 command is invoked with a number between one and ten, the function
5979 considers that number to be a fraction of the length of the buffer,
5980 measured in tenths, and Emacs moves the cursor that fraction of the
5981 way from the beginning of the buffer. Thus, you can either call this
5982 function with the key command @kbd{M-<}, which will move the cursor to
5983 the beginning of the buffer, or with a key command such as @kbd{C-u 7
5984 M-<} which will move the cursor to a point 70% of the way through the
5985 buffer. If a number bigger than ten is used for the argument, it
5986 moves to the end of the buffer.
5988 The @code{beginning-of-buffer} function can be called with or without an
5989 argument. The use of the argument is optional.
5992 * Optional Arguments::
5993 * beginning-of-buffer opt arg:: Example with optional argument.
5994 * beginning-of-buffer complete::
5997 @node Optional Arguments
5998 @subsection Optional Arguments
6000 Unless told otherwise, Lisp expects that a function with an argument in
6001 its function definition will be called with a value for that argument.
6002 If that does not happen, you get an error and a message that says
6003 @samp{Wrong number of arguments}.
6005 @cindex Optional arguments
6008 However, optional arguments are a feature of Lisp: a particular
6009 @dfn{keyword} is used to tell the Lisp interpreter that an argument is
6010 optional. The keyword is @code{&optional}. (The @samp{&} in front of
6011 @samp{optional} is part of the keyword.) In a function definition, if
6012 an argument follows the keyword @code{&optional}, no value need be
6013 passed to that argument when the function is called.
6016 The first line of the function definition of @code{beginning-of-buffer}
6017 therefore looks like this:
6020 (defun beginning-of-buffer (&optional arg)
6024 In outline, the whole function looks like this:
6028 (defun beginning-of-buffer (&optional arg)
6029 "@var{documentation}@dots{}"
6031 (or (@var{is-the-argument-a-cons-cell} arg)
6032 (and @var{are-both-transient-mark-mode-and-mark-active-true})
6034 (let (@var{determine-size-and-set-it})
6036 (@var{if-there-is-an-argument}
6037 @var{figure-out-where-to-go}
6044 The function is similar to the @code{simplified-beginning-of-buffer}
6045 function except that the @code{interactive} expression has @code{"P"}
6046 as an argument and the @code{goto-char} function is followed by an
6047 if-then-else expression that figures out where to put the cursor if
6048 there is an argument that is not a cons cell.
6050 (Since I do not explain a cons cell for many more chapters, please
6051 consider ignoring the function @code{consp}. @xref{List
6052 Implementation, , How Lists are Implemented}, and @ref{Cons Cell Type,
6053 , Cons Cell and List Types, elisp, The GNU Emacs Lisp Reference
6056 The @code{"P"} in the @code{interactive} expression tells Emacs to
6057 pass a prefix argument, if there is one, to the function in raw form.
6058 A prefix argument is made by typing the @key{META} key followed by a
6059 number, or by typing @kbd{C-u} and then a number. (If you don't type
6060 a number, @kbd{C-u} defaults to a cons cell with a 4. A lowercase
6061 @code{"p"} in the @code{interactive} expression causes the function to
6062 convert a prefix arg to a number.)
6064 The true-or-false-test of the @code{if} expression looks complex, but
6065 it is not: it checks whether @code{arg} has a value that is not
6066 @code{nil} and whether it is a cons cell. (That is what @code{consp}
6067 does; it checks whether its argument is a cons cell.) If @code{arg}
6068 has a value that is not @code{nil} (and is not a cons cell), which
6069 will be the case if @code{beginning-of-buffer} is called with a
6070 numeric argument, then this true-or-false-test will return true and
6071 the then-part of the @code{if} expression will be evaluated. On the
6072 other hand, if @code{beginning-of-buffer} is not called with an
6073 argument, the value of @code{arg} will be @code{nil} and the else-part
6074 of the @code{if} expression will be evaluated. The else-part is
6075 simply @code{point-min}, and when this is the outcome, the whole
6076 @code{goto-char} expression is @code{(goto-char (point-min))}, which
6077 is how we saw the @code{beginning-of-buffer} function in its
6080 @node beginning-of-buffer opt arg
6081 @subsection @code{beginning-of-buffer} with an Argument
6083 When @code{beginning-of-buffer} is called with an argument, an
6084 expression is evaluated which calculates what value to pass to
6085 @code{goto-char}. This expression is rather complicated at first sight.
6086 It includes an inner @code{if} expression and much arithmetic. It looks
6091 (if (> (buffer-size) 10000)
6092 ;; @r{Avoid overflow for large buffer sizes!}
6093 (* (prefix-numeric-value arg)
6098 size (prefix-numeric-value arg))) 10)))
6103 * Disentangle beginning-of-buffer::
6104 * Large buffer case::
6105 * Small buffer case::
6109 @node Disentangle beginning-of-buffer
6110 @unnumberedsubsubsec Disentangle @code{beginning-of-buffer}
6113 Like other complex-looking expressions, the conditional expression
6114 within @code{beginning-of-buffer} can be disentangled by looking at it
6115 as parts of a template, in this case, the template for an if-then-else
6116 expression. In skeletal form, the expression looks like this:
6120 (if (@var{buffer-is-large}
6121 @var{divide-buffer-size-by-10-and-multiply-by-arg}
6122 @var{else-use-alternate-calculation}
6126 The true-or-false-test of this inner @code{if} expression checks the
6127 size of the buffer. The reason for this is that the old version 18
6128 Emacs used numbers that are no bigger than eight million or so and in
6129 the computation that followed, the programmer feared that Emacs might
6130 try to use over-large numbers if the buffer were large. The term
6131 `overflow', mentioned in the comment, means numbers that are over
6132 large. More recent versions of Emacs use larger numbers, but this
6133 code has not been touched, if only because people now look at buffers
6134 that are far, far larger than ever before.
6136 There are two cases: if the buffer is large and if it is not.
6138 @node Large buffer case
6139 @unnumberedsubsubsec What happens in a large buffer
6141 In @code{beginning-of-buffer}, the inner @code{if} expression tests
6142 whether the size of the buffer is greater than 10,000 characters. To do
6143 this, it uses the @code{>} function and the computation of @code{size}
6144 that comes from the let expression.
6146 In the old days, the function @code{buffer-size} was used. Not only
6147 was that function called several times, it gave the size of the whole
6148 buffer, not the accessible part. The computation makes much more
6149 sense when it handles just the accessible part. (@xref{Narrowing &
6150 Widening, , Narrowing and Widening}, for more information on focusing
6151 attention to an `accessible' part.)
6154 The line looks like this:
6162 When the buffer is large, the then-part of the @code{if} expression is
6163 evaluated. It reads like this (after formatting for easy reading):
6168 (prefix-numeric-value arg)
6174 This expression is a multiplication, with two arguments to the function
6177 The first argument is @code{(prefix-numeric-value arg)}. When
6178 @code{"P"} is used as the argument for @code{interactive}, the value
6179 passed to the function as its argument is passed a ``raw prefix
6180 argument'', and not a number. (It is a number in a list.) To perform
6181 the arithmetic, a conversion is necessary, and
6182 @code{prefix-numeric-value} does the job.
6184 @findex / @r{(division)}
6186 The second argument is @code{(/ size 10)}. This expression divides
6187 the numeric value by ten---the numeric value of the size of the
6188 accessible portion of the buffer. This produces a number that tells
6189 how many characters make up one tenth of the buffer size. (In Lisp,
6190 @code{/} is used for division, just as @code{*} is used for
6194 In the multiplication expression as a whole, this amount is multiplied
6195 by the value of the prefix argument---the multiplication looks like this:
6199 (* @var{numeric-value-of-prefix-arg}
6200 @var{number-of-characters-in-one-tenth-of-the-accessible-buffer})
6205 If, for example, the prefix argument is @samp{7}, the one-tenth value
6206 will be multiplied by 7 to give a position 70% of the way through.
6209 The result of all this is that if the accessible portion of the buffer
6210 is large, the @code{goto-char} expression reads like this:
6214 (goto-char (* (prefix-numeric-value arg)
6219 This puts the cursor where we want it.
6221 @node Small buffer case
6222 @unnumberedsubsubsec What happens in a small buffer
6224 If the buffer contains fewer than 10,000 characters, a slightly
6225 different computation is performed. You might think this is not
6226 necessary, since the first computation could do the job. However, in
6227 a small buffer, the first method may not put the cursor on exactly the
6228 desired line; the second method does a better job.
6231 The code looks like this:
6233 @c Keep this on one line.
6235 (/ (+ 10 (* size (prefix-numeric-value arg))) 10))
6240 This is code in which you figure out what happens by discovering how the
6241 functions are embedded in parentheses. It is easier to read if you
6242 reformat it with each expression indented more deeply than its
6243 enclosing expression:
6251 (prefix-numeric-value arg)))
6258 Looking at parentheses, we see that the innermost operation is
6259 @code{(prefix-numeric-value arg)}, which converts the raw argument to
6260 a number. In the following expression, this number is multiplied by
6261 the size of the accessible portion of the buffer:
6264 (* size (prefix-numeric-value arg))
6268 This multiplication creates a number that may be larger than the size of
6269 the buffer---seven times larger if the argument is 7, for example. Ten
6270 is then added to this number and finally the large number is divided by
6271 ten to provide a value that is one character larger than the percentage
6272 position in the buffer.
6274 The number that results from all this is passed to @code{goto-char} and
6275 the cursor is moved to that point.
6278 @node beginning-of-buffer complete
6279 @subsection The Complete @code{beginning-of-buffer}
6282 Here is the complete text of the @code{beginning-of-buffer} function:
6288 (defun beginning-of-buffer (&optional arg)
6289 "Move point to the beginning of the buffer;
6290 leave mark at previous position.
6291 With \\[universal-argument] prefix,
6292 do not set mark at previous position.
6294 put point N/10 of the way from the beginning.
6296 If the buffer is narrowed,
6297 this command uses the beginning and size
6298 of the accessible part of the buffer.
6302 Don't use this command in Lisp programs!
6303 \(goto-char (point-min)) is faster
6304 and avoids clobbering the mark."
6307 (and transient-mark-mode mark-active)
6311 (let ((size (- (point-max) (point-min))))
6312 (goto-char (if (and arg (not (consp arg)))
6315 ;; Avoid overflow for large buffer sizes!
6316 (* (prefix-numeric-value arg)
6318 (/ (+ 10 (* size (prefix-numeric-value arg)))
6321 (if arg (forward-line 1)))
6326 From before GNU Emacs 22
6329 (defun beginning-of-buffer (&optional arg)
6330 "Move point to the beginning of the buffer;
6331 leave mark at previous position.
6332 With arg N, put point N/10 of the way
6333 from the true beginning.
6336 Don't use this in Lisp programs!
6337 \(goto-char (point-min)) is faster
6338 and does not set the mark."
6345 (if (> (buffer-size) 10000)
6346 ;; @r{Avoid overflow for large buffer sizes!}
6347 (* (prefix-numeric-value arg)
6348 (/ (buffer-size) 10))
6351 (/ (+ 10 (* (buffer-size)
6352 (prefix-numeric-value arg)))
6355 (if arg (forward-line 1)))
6361 Except for two small points, the previous discussion shows how this
6362 function works. The first point deals with a detail in the
6363 documentation string, and the second point concerns the last line of
6367 In the documentation string, there is reference to an expression:
6370 \\[universal-argument]
6374 A @samp{\\} is used before the first square bracket of this
6375 expression. This @samp{\\} tells the Lisp interpreter to substitute
6376 whatever key is currently bound to the @samp{[@dots{}]}. In the case
6377 of @code{universal-argument}, that is usually @kbd{C-u}, but it might
6378 be different. (@xref{Documentation Tips, , Tips for Documentation
6379 Strings, elisp, The GNU Emacs Lisp Reference Manual}, for more
6383 Finally, the last line of the @code{beginning-of-buffer} command says
6384 to move point to the beginning of the next line if the command is
6385 invoked with an argument:
6388 (if arg (forward-line 1)))
6392 This puts the cursor at the beginning of the first line after the
6393 appropriate tenths position in the buffer. This is a flourish that
6394 means that the cursor is always located @emph{at least} the requested
6395 tenths of the way through the buffer, which is a nicety that is,
6396 perhaps, not necessary, but which, if it did not occur, would be sure
6399 On the other hand, it also means that if you specify the command with
6400 a @kbd{C-u}, but without a number, that is to say, if the `raw prefix
6401 argument' is simply a cons cell, then the command puts you at the
6402 beginning of the second line @dots{} I don't know whether this is
6403 intended or whether no one has dealt with the code to avoid this
6406 @node Second Buffer Related Review
6409 Here is a brief summary of some of the topics covered in this chapter.
6413 Evaluate each argument in sequence, and return the value of the first
6414 argument that is not @code{nil}; if none return a value that is not
6415 @code{nil}, return @code{nil}. In brief, return the first true value
6416 of the arguments; return a true value if one @emph{or} any of the
6420 Evaluate each argument in sequence, and if any are @code{nil}, return
6421 @code{nil}; if none are @code{nil}, return the value of the last
6422 argument. In brief, return a true value only if all the arguments are
6423 true; return a true value if one @emph{and} each of the others is
6427 A keyword used to indicate that an argument to a function definition
6428 is optional; this means that the function can be evaluated without the
6429 argument, if desired.
6431 @item prefix-numeric-value
6432 Convert the `raw prefix argument' produced by @code{(interactive
6433 "P")} to a numeric value.
6436 Move point forward to the beginning of the next line, or if the argument
6437 is greater than one, forward that many lines. If it can't move as far
6438 forward as it is supposed to, @code{forward-line} goes forward as far as
6439 it can and then returns a count of the number of additional lines it was
6440 supposed to move but couldn't.
6443 Delete the entire contents of the current buffer.
6446 Return @code{t} if its argument is a buffer; otherwise return @code{nil}.
6449 @node optional Exercise
6450 @section @code{optional} Argument Exercise
6452 Write an interactive function with an optional argument that tests
6453 whether its argument, a number, is greater than or equal to, or else,
6454 less than the value of @code{fill-column}, and tells you which, in a
6455 message. However, if you do not pass an argument to the function, use
6456 56 as a default value.
6458 @node Narrowing & Widening
6459 @chapter Narrowing and Widening
6460 @cindex Focusing attention (narrowing)
6464 Narrowing is a feature of Emacs that makes it possible for you to focus
6465 on a specific part of a buffer, and work without accidentally changing
6466 other parts. Narrowing is normally disabled since it can confuse
6470 * Narrowing advantages:: The advantages of narrowing
6471 * save-restriction:: The @code{save-restriction} special form.
6472 * what-line:: The number of the line that point is on.
6477 @node Narrowing advantages
6478 @unnumberedsec The Advantages of Narrowing
6481 With narrowing, the rest of a buffer is made invisible, as if it weren't
6482 there. This is an advantage if, for example, you want to replace a word
6483 in one part of a buffer but not in another: you narrow to the part you want
6484 and the replacement is carried out only in that section, not in the rest
6485 of the buffer. Searches will only work within a narrowed region, not
6486 outside of one, so if you are fixing a part of a document, you can keep
6487 yourself from accidentally finding parts you do not need to fix by
6488 narrowing just to the region you want.
6489 (The key binding for @code{narrow-to-region} is @kbd{C-x n n}.)
6491 However, narrowing does make the rest of the buffer invisible, which
6492 can scare people who inadvertently invoke narrowing and think they
6493 have deleted a part of their file. Moreover, the @code{undo} command
6494 (which is usually bound to @kbd{C-x u}) does not turn off narrowing
6495 (nor should it), so people can become quite desperate if they do not
6496 know that they can return the rest of a buffer to visibility with the
6497 @code{widen} command.
6498 (The key binding for @code{widen} is @kbd{C-x n w}.)
6500 Narrowing is just as useful to the Lisp interpreter as to a human.
6501 Often, an Emacs Lisp function is designed to work on just part of a
6502 buffer; or conversely, an Emacs Lisp function needs to work on all of a
6503 buffer that has been narrowed. The @code{what-line} function, for
6504 example, removes the narrowing from a buffer, if it has any narrowing
6505 and when it has finished its job, restores the narrowing to what it was.
6506 On the other hand, the @code{count-lines} function
6507 uses narrowing to restrict itself to just that portion
6508 of the buffer in which it is interested and then restores the previous
6511 @node save-restriction
6512 @section The @code{save-restriction} Special Form
6513 @findex save-restriction
6515 In Emacs Lisp, you can use the @code{save-restriction} special form to
6516 keep track of whatever narrowing is in effect, if any. When the Lisp
6517 interpreter meets with @code{save-restriction}, it executes the code
6518 in the body of the @code{save-restriction} expression, and then undoes
6519 any changes to narrowing that the code caused. If, for example, the
6520 buffer is narrowed and the code that follows @code{save-restriction}
6521 gets rid of the narrowing, @code{save-restriction} returns the buffer
6522 to its narrowed region afterwards. In the @code{what-line} command,
6523 any narrowing the buffer may have is undone by the @code{widen}
6524 command that immediately follows the @code{save-restriction} command.
6525 Any original narrowing is restored just before the completion of the
6529 The template for a @code{save-restriction} expression is simple:
6539 The body of the @code{save-restriction} is one or more expressions that
6540 will be evaluated in sequence by the Lisp interpreter.
6542 Finally, a point to note: when you use both @code{save-excursion} and
6543 @code{save-restriction}, one right after the other, you should use
6544 @code{save-excursion} outermost. If you write them in reverse order,
6545 you may fail to record narrowing in the buffer to which Emacs switches
6546 after calling @code{save-excursion}. Thus, when written together,
6547 @code{save-excursion} and @code{save-restriction} should be written
6558 In other circumstances, when not written together, the
6559 @code{save-excursion} and @code{save-restriction} special forms must
6560 be written in the order appropriate to the function.
6576 /usr/local/src/emacs/lisp/simple.el
6579 "Print the current buffer line number and narrowed line number of point."
6581 (let ((start (point-min))
6582 (n (line-number-at-pos)))
6584 (message "Line %d" n)
6588 (message "line %d (narrowed line %d)"
6589 (+ n (line-number-at-pos start) -1) n))))))
6591 (defun line-number-at-pos (&optional pos)
6592 "Return (narrowed) buffer line number at position POS.
6593 If POS is nil, use current buffer location.
6594 Counting starts at (point-min), so the value refers
6595 to the contents of the accessible portion of the buffer."
6596 (let ((opoint (or pos (point))) start)
6598 (goto-char (point-min))
6599 (setq start (point))
6602 (1+ (count-lines start (point))))))
6604 (defun count-lines (start end)
6605 "Return number of lines between START and END.
6606 This is usually the number of newlines between them,
6607 but can be one more if START is not equal to END
6608 and the greater of them is not at the start of a line."
6611 (narrow-to-region start end)
6612 (goto-char (point-min))
6613 (if (eq selective-display t)
6616 (while (re-search-forward "[\n\C-m]" nil t 40)
6617 (setq done (+ 40 done)))
6618 (while (re-search-forward "[\n\C-m]" nil t 1)
6619 (setq done (+ 1 done)))
6620 (goto-char (point-max))
6621 (if (and (/= start end)
6625 (- (buffer-size) (forward-line (buffer-size)))))))
6629 @section @code{what-line}
6631 @cindex Widening, example of
6633 The @code{what-line} command tells you the number of the line in which
6634 the cursor is located. The function illustrates the use of the
6635 @code{save-restriction} and @code{save-excursion} commands. Here is the
6636 original text of the function:
6641 "Print the current line number (in the buffer) of point."
6648 (1+ (count-lines 1 (point)))))))
6652 (In recent versions of GNU Emacs, the @code{what-line} function has
6653 been expanded to tell you your line number in a narrowed buffer as
6654 well as your line number in a widened buffer. The recent version is
6655 more complex than the version shown here. If you feel adventurous,
6656 you might want to look at it after figuring out how this version
6657 works. You will probably need to use @kbd{C-h f}
6658 (@code{describe-function}). The newer version uses a conditional to
6659 determine whether the buffer has been narrowed.
6661 (Also, it uses @code{line-number-at-pos}, which among other simple
6662 expressions, such as @code{(goto-char (point-min))}, moves point to
6663 the beginning of the current line with @code{(forward-line 0)} rather
6664 than @code{beginning-of-line}.)
6666 The @code{what-line} function as shown here has a documentation line
6667 and is interactive, as you would expect. The next two lines use the
6668 functions @code{save-restriction} and @code{widen}.
6670 The @code{save-restriction} special form notes whatever narrowing is in
6671 effect, if any, in the current buffer and restores that narrowing after
6672 the code in the body of the @code{save-restriction} has been evaluated.
6674 The @code{save-restriction} special form is followed by @code{widen}.
6675 This function undoes any narrowing the current buffer may have had
6676 when @code{what-line} was called. (The narrowing that was there is
6677 the narrowing that @code{save-restriction} remembers.) This widening
6678 makes it possible for the line counting commands to count from the
6679 beginning of the buffer. Otherwise, they would have been limited to
6680 counting within the accessible region. Any original narrowing is
6681 restored just before the completion of the function by the
6682 @code{save-restriction} special form.
6684 The call to @code{widen} is followed by @code{save-excursion}, which
6685 saves the location of the cursor (i.e., of point) and of the mark, and
6686 restores them after the code in the body of the @code{save-excursion}
6687 uses the @code{beginning-of-line} function to move point.
6689 (Note that the @code{(widen)} expression comes between the
6690 @code{save-restriction} and @code{save-excursion} special forms. When
6691 you write the two @code{save- @dots{}} expressions in sequence, write
6692 @code{save-excursion} outermost.)
6695 The last two lines of the @code{what-line} function are functions to
6696 count the number of lines in the buffer and then print the number in the
6702 (1+ (count-lines 1 (point)))))))
6706 The @code{message} function prints a one-line message at the bottom of
6707 the Emacs screen. The first argument is inside of quotation marks and
6708 is printed as a string of characters. However, it may contain a
6709 @samp{%d} expression to print a following argument. @samp{%d} prints
6710 the argument as a decimal, so the message will say something such as
6714 The number that is printed in place of the @samp{%d} is computed by the
6715 last line of the function:
6718 (1+ (count-lines 1 (point)))
6724 (defun count-lines (start end)
6725 "Return number of lines between START and END.
6726 This is usually the number of newlines between them,
6727 but can be one more if START is not equal to END
6728 and the greater of them is not at the start of a line."
6731 (narrow-to-region start end)
6732 (goto-char (point-min))
6733 (if (eq selective-display t)
6736 (while (re-search-forward "[\n\C-m]" nil t 40)
6737 (setq done (+ 40 done)))
6738 (while (re-search-forward "[\n\C-m]" nil t 1)
6739 (setq done (+ 1 done)))
6740 (goto-char (point-max))
6741 (if (and (/= start end)
6745 (- (buffer-size) (forward-line (buffer-size)))))))
6749 What this does is count the lines from the first position of the
6750 buffer, indicated by the @code{1}, up to @code{(point)}, and then add
6751 one to that number. (The @code{1+} function adds one to its
6752 argument.) We add one to it because line 2 has only one line before
6753 it, and @code{count-lines} counts only the lines @emph{before} the
6756 After @code{count-lines} has done its job, and the message has been
6757 printed in the echo area, the @code{save-excursion} restores point and
6758 mark to their original positions; and @code{save-restriction} restores
6759 the original narrowing, if any.
6761 @node narrow Exercise
6762 @section Exercise with Narrowing
6764 Write a function that will display the first 60 characters of the
6765 current buffer, even if you have narrowed the buffer to its latter
6766 half so that the first line is inaccessible. Restore point, mark, and
6767 narrowing. For this exercise, you need to use a whole potpourri of
6768 functions, including @code{save-restriction}, @code{widen},
6769 @code{goto-char}, @code{point-min}, @code{message}, and
6770 @code{buffer-substring}.
6772 @cindex Properties, mention of @code{buffer-substring-no-properties}
6773 (@code{buffer-substring} is a previously unmentioned function you will
6774 have to investigate yourself; or perhaps you will have to use
6775 @code{buffer-substring-no-properties} or
6776 @code{filter-buffer-substring} @dots{}, yet other functions. Text
6777 properties are a feature otherwise not discussed here. @xref{Text
6778 Properties, , Text Properties, elisp, The GNU Emacs Lisp Reference
6781 Additionally, do you really need @code{goto-char} or @code{point-min}?
6782 Or can you write the function without them?
6784 @node car cdr & cons
6785 @chapter @code{car}, @code{cdr}, @code{cons}: Fundamental Functions
6786 @findex car, @r{introduced}
6787 @findex cdr, @r{introduced}
6789 In Lisp, @code{car}, @code{cdr}, and @code{cons} are fundamental
6790 functions. The @code{cons} function is used to construct lists, and
6791 the @code{car} and @code{cdr} functions are used to take them apart.
6793 In the walk through of the @code{copy-region-as-kill} function, we
6794 will see @code{cons} as well as two variants on @code{cdr},
6795 namely, @code{setcdr} and @code{nthcdr}. (@xref{copy-region-as-kill}.)
6798 * Strange Names:: An historical aside: why the strange names?
6799 * car & cdr:: Functions for extracting part of a list.
6800 * cons:: Constructing a list.
6801 * nthcdr:: Calling @code{cdr} repeatedly.
6803 * setcar:: Changing the first element of a list.
6804 * setcdr:: Changing the rest of a list.
6810 @unnumberedsec Strange Names
6813 The name of the @code{cons} function is not unreasonable: it is an
6814 abbreviation of the word `construct'. The origins of the names for
6815 @code{car} and @code{cdr}, on the other hand, are esoteric: @code{car}
6816 is an acronym from the phrase `Contents of the Address part of the
6817 Register'; and @code{cdr} (pronounced `could-er') is an acronym from
6818 the phrase `Contents of the Decrement part of the Register'. These
6819 phrases refer to specific pieces of hardware on the very early
6820 computer on which the original Lisp was developed. Besides being
6821 obsolete, the phrases have been completely irrelevant for more than 25
6822 years to anyone thinking about Lisp. Nonetheless, although a few
6823 brave scholars have begun to use more reasonable names for these
6824 functions, the old terms are still in use. In particular, since the
6825 terms are used in the Emacs Lisp source code, we will use them in this
6829 @section @code{car} and @code{cdr}
6831 The @sc{car} of a list is, quite simply, the first item in the list.
6832 Thus the @sc{car} of the list @code{(rose violet daisy buttercup)} is
6836 If you are reading this in Info in GNU Emacs, you can see this by
6837 evaluating the following:
6840 (car '(rose violet daisy buttercup))
6844 After evaluating the expression, @code{rose} will appear in the echo
6847 Clearly, a more reasonable name for the @code{car} function would be
6848 @code{first} and this is often suggested.
6850 @code{car} does not remove the first item from the list; it only reports
6851 what it is. After @code{car} has been applied to a list, the list is
6852 still the same as it was. In the jargon, @code{car} is
6853 `non-destructive'. This feature turns out to be important.
6855 The @sc{cdr} of a list is the rest of the list, that is, the
6856 @code{cdr} function returns the part of the list that follows the
6857 first item. Thus, while the @sc{car} of the list @code{'(rose violet
6858 daisy buttercup)} is @code{rose}, the rest of the list, the value
6859 returned by the @code{cdr} function, is @code{(violet daisy
6863 You can see this by evaluating the following in the usual way:
6866 (cdr '(rose violet daisy buttercup))
6870 When you evaluate this, @code{(violet daisy buttercup)} will appear in
6873 Like @code{car}, @code{cdr} does not remove any elements from the
6874 list---it just returns a report of what the second and subsequent
6877 Incidentally, in the example, the list of flowers is quoted. If it were
6878 not, the Lisp interpreter would try to evaluate the list by calling
6879 @code{rose} as a function. In this example, we do not want to do that.
6881 Clearly, a more reasonable name for @code{cdr} would be @code{rest}.
6883 (There is a lesson here: when you name new functions, consider very
6884 carefully what you are doing, since you may be stuck with the names
6885 for far longer than you expect. The reason this document perpetuates
6886 these names is that the Emacs Lisp source code uses them, and if I did
6887 not use them, you would have a hard time reading the code; but do,
6888 please, try to avoid using these terms yourself. The people who come
6889 after you will be grateful to you.)
6891 When @code{car} and @code{cdr} are applied to a list made up of symbols,
6892 such as the list @code{(pine fir oak maple)}, the element of the list
6893 returned by the function @code{car} is the symbol @code{pine} without
6894 any parentheses around it. @code{pine} is the first element in the
6895 list. However, the @sc{cdr} of the list is a list itself, @code{(fir
6896 oak maple)}, as you can see by evaluating the following expressions in
6901 (car '(pine fir oak maple))
6903 (cdr '(pine fir oak maple))
6907 On the other hand, in a list of lists, the first element is itself a
6908 list. @code{car} returns this first element as a list. For example,
6909 the following list contains three sub-lists, a list of carnivores, a
6910 list of herbivores and a list of sea mammals:
6914 (car '((lion tiger cheetah)
6915 (gazelle antelope zebra)
6916 (whale dolphin seal)))
6921 In this example, the first element or @sc{car} of the list is the list of
6922 carnivores, @code{(lion tiger cheetah)}, and the rest of the list is
6923 @code{((gazelle antelope zebra) (whale dolphin seal))}.
6927 (cdr '((lion tiger cheetah)
6928 (gazelle antelope zebra)
6929 (whale dolphin seal)))
6933 It is worth saying again that @code{car} and @code{cdr} are
6934 non-destructive---that is, they do not modify or change lists to which
6935 they are applied. This is very important for how they are used.
6937 Also, in the first chapter, in the discussion about atoms, I said that
6938 in Lisp, ``certain kinds of atom, such as an array, can be separated
6939 into parts; but the mechanism for doing this is different from the
6940 mechanism for splitting a list. As far as Lisp is concerned, the
6941 atoms of a list are unsplittable.'' (@xref{Lisp Atoms}.) The
6942 @code{car} and @code{cdr} functions are used for splitting lists and
6943 are considered fundamental to Lisp. Since they cannot split or gain
6944 access to the parts of an array, an array is considered an atom.
6945 Conversely, the other fundamental function, @code{cons}, can put
6946 together or construct a list, but not an array. (Arrays are handled
6947 by array-specific functions. @xref{Arrays, , Arrays, elisp, The GNU
6948 Emacs Lisp Reference Manual}.)
6951 @section @code{cons}
6952 @findex cons, @r{introduced}
6954 The @code{cons} function constructs lists; it is the inverse of
6955 @code{car} and @code{cdr}. For example, @code{cons} can be used to make
6956 a four element list from the three element list, @code{(fir oak maple)}:
6959 (cons 'pine '(fir oak maple))
6964 After evaluating this list, you will see
6967 (pine fir oak maple)
6971 appear in the echo area. @code{cons} causes the creation of a new
6972 list in which the element is followed by the elements of the original
6975 We often say that `@code{cons} puts a new element at the beginning of
6976 a list; it attaches or pushes elements onto the list', but this
6977 phrasing can be misleading, since @code{cons} does not change an
6978 existing list, but creates a new one.
6980 Like @code{car} and @code{cdr}, @code{cons} is non-destructive.
6984 * length:: How to find the length of a list.
6989 @unnumberedsubsec Build a list
6992 @code{cons} must have a list to attach to.@footnote{Actually, you can
6993 @code{cons} an element to an atom to produce a dotted pair. Dotted
6994 pairs are not discussed here; see @ref{Dotted Pair Notation, , Dotted
6995 Pair Notation, elisp, The GNU Emacs Lisp Reference Manual}.} You
6996 cannot start from absolutely nothing. If you are building a list, you
6997 need to provide at least an empty list at the beginning. Here is a
6998 series of @code{cons} expressions that build up a list of flowers. If
6999 you are reading this in Info in GNU Emacs, you can evaluate each of
7000 the expressions in the usual way; the value is printed in this text
7001 after @samp{@result{}}, which you may read as `evaluates to'.
7005 (cons 'buttercup ())
7006 @result{} (buttercup)
7010 (cons 'daisy '(buttercup))
7011 @result{} (daisy buttercup)
7015 (cons 'violet '(daisy buttercup))
7016 @result{} (violet daisy buttercup)
7020 (cons 'rose '(violet daisy buttercup))
7021 @result{} (rose violet daisy buttercup)
7026 In the first example, the empty list is shown as @code{()} and a list
7027 made up of @code{buttercup} followed by the empty list is constructed.
7028 As you can see, the empty list is not shown in the list that was
7029 constructed. All that you see is @code{(buttercup)}. The empty list is
7030 not counted as an element of a list because there is nothing in an empty
7031 list. Generally speaking, an empty list is invisible.
7033 The second example, @code{(cons 'daisy '(buttercup))} constructs a new,
7034 two element list by putting @code{daisy} in front of @code{buttercup};
7035 and the third example constructs a three element list by putting
7036 @code{violet} in front of @code{daisy} and @code{buttercup}.
7039 @subsection Find the Length of a List: @code{length}
7042 You can find out how many elements there are in a list by using the Lisp
7043 function @code{length}, as in the following examples:
7047 (length '(buttercup))
7052 (length '(daisy buttercup))
7057 (length (cons 'violet '(daisy buttercup)))
7063 In the third example, the @code{cons} function is used to construct a
7064 three element list which is then passed to the @code{length} function as
7068 We can also use @code{length} to count the number of elements in an
7079 As you would expect, the number of elements in an empty list is zero.
7081 An interesting experiment is to find out what happens if you try to find
7082 the length of no list at all; that is, if you try to call @code{length}
7083 without giving it an argument, not even an empty list:
7091 What you see, if you evaluate this, is the error message
7094 Lisp error: (wrong-number-of-arguments length 0)
7098 This means that the function receives the wrong number of
7099 arguments, zero, when it expects some other number of arguments. In
7100 this case, one argument is expected, the argument being a list whose
7101 length the function is measuring. (Note that @emph{one} list is
7102 @emph{one} argument, even if the list has many elements inside it.)
7104 The part of the error message that says @samp{length} is the name of
7108 @code{length} is still a subroutine, but you need C-h f to discover that.
7110 In an earlier version:
7111 This is written with a special notation, @samp{#<subr},
7112 that indicates that the function @code{length} is one of the primitive
7113 functions written in C rather than in Emacs Lisp. (@samp{subr} is an
7114 abbreviation for `subroutine'.) @xref{What Is a Function, , What Is a
7115 Function?, elisp , The GNU Emacs Lisp Reference Manual}, for more
7120 @section @code{nthcdr}
7123 The @code{nthcdr} function is associated with the @code{cdr} function.
7124 What it does is take the @sc{cdr} of a list repeatedly.
7126 If you take the @sc{cdr} of the list @code{(pine fir
7127 oak maple)}, you will be returned the list @code{(fir oak maple)}. If you
7128 repeat this on what was returned, you will be returned the list
7129 @code{(oak maple)}. (Of course, repeated @sc{cdr}ing on the original
7130 list will just give you the original @sc{cdr} since the function does
7131 not change the list. You need to evaluate the @sc{cdr} of the
7132 @sc{cdr} and so on.) If you continue this, eventually you will be
7133 returned an empty list, which in this case, instead of being shown as
7134 @code{()} is shown as @code{nil}.
7137 For review, here is a series of repeated @sc{cdr}s, the text following
7138 the @samp{@result{}} shows what is returned.
7142 (cdr '(pine fir oak maple))
7143 @result{}(fir oak maple)
7147 (cdr '(fir oak maple))
7148 @result{} (oak maple)
7173 You can also do several @sc{cdr}s without printing the values in
7178 (cdr (cdr '(pine fir oak maple)))
7179 @result{} (oak maple)
7184 In this example, the Lisp interpreter evaluates the innermost list first.
7185 The innermost list is quoted, so it just passes the list as it is to the
7186 innermost @code{cdr}. This @code{cdr} passes a list made up of the
7187 second and subsequent elements of the list to the outermost @code{cdr},
7188 which produces a list composed of the third and subsequent elements of
7189 the original list. In this example, the @code{cdr} function is repeated
7190 and returns a list that consists of the original list without its
7193 The @code{nthcdr} function does the same as repeating the call to
7194 @code{cdr}. In the following example, the argument 2 is passed to the
7195 function @code{nthcdr}, along with the list, and the value returned is
7196 the list without its first two items, which is exactly the same
7197 as repeating @code{cdr} twice on the list:
7201 (nthcdr 2 '(pine fir oak maple))
7202 @result{} (oak maple)
7207 Using the original four element list, we can see what happens when
7208 various numeric arguments are passed to @code{nthcdr}, including 0, 1,
7213 ;; @r{Leave the list as it was.}
7214 (nthcdr 0 '(pine fir oak maple))
7215 @result{} (pine fir oak maple)
7219 ;; @r{Return a copy without the first element.}
7220 (nthcdr 1 '(pine fir oak maple))
7221 @result{} (fir oak maple)
7225 ;; @r{Return a copy of the list without three elements.}
7226 (nthcdr 3 '(pine fir oak maple))
7231 ;; @r{Return a copy lacking all four elements.}
7232 (nthcdr 4 '(pine fir oak maple))
7237 ;; @r{Return a copy lacking all elements.}
7238 (nthcdr 5 '(pine fir oak maple))
7247 The @code{nthcdr} function takes the @sc{cdr} of a list repeatedly.
7248 The @code{nth} function takes the @sc{car} of the result returned by
7249 @code{nthcdr}. It returns the Nth element of the list.
7252 Thus, if it were not defined in C for speed, the definition of
7253 @code{nth} would be:
7258 "Returns the Nth element of LIST.
7259 N counts from zero. If LIST is not that long, nil is returned."
7260 (car (nthcdr n list)))
7265 (Originally, @code{nth} was defined in Emacs Lisp in @file{subr.el},
7266 but its definition was redone in C in the 1980s.)
7268 The @code{nth} function returns a single element of a list.
7269 This can be very convenient.
7271 Note that the elements are numbered from zero, not one. That is to
7272 say, the first element of a list, its @sc{car} is the zeroth element.
7273 This is called `zero-based' counting and often bothers people who
7274 are accustomed to the first element in a list being number one, which
7282 (nth 0 '("one" "two" "three"))
7285 (nth 1 '("one" "two" "three"))
7290 It is worth mentioning that @code{nth}, like @code{nthcdr} and
7291 @code{cdr}, does not change the original list---the function is
7292 non-destructive. This is in sharp contrast to the @code{setcar} and
7293 @code{setcdr} functions.
7296 @section @code{setcar}
7299 As you might guess from their names, the @code{setcar} and @code{setcdr}
7300 functions set the @sc{car} or the @sc{cdr} of a list to a new value.
7301 They actually change the original list, unlike @code{car} and @code{cdr}
7302 which leave the original list as it was. One way to find out how this
7303 works is to experiment. We will start with the @code{setcar} function.
7306 First, we can make a list and then set the value of a variable to the
7307 list, using the @code{setq} function. Here is a list of animals:
7310 (setq animals '(antelope giraffe lion tiger))
7314 If you are reading this in Info inside of GNU Emacs, you can evaluate
7315 this expression in the usual fashion, by positioning the cursor after
7316 the expression and typing @kbd{C-x C-e}. (I'm doing this right here
7317 as I write this. This is one of the advantages of having the
7318 interpreter built into the computing environment. Incidentally, when
7319 there is nothing on the line after the final parentheses, such as a
7320 comment, point can be on the next line. Thus, if your cursor is in
7321 the first column of the next line, you do not need to move it.
7322 Indeed, Emacs permits any amount of white space after the final
7326 When we evaluate the variable @code{animals}, we see that it is bound to
7327 the list @code{(antelope giraffe lion tiger)}:
7332 @result{} (antelope giraffe lion tiger)
7337 Put another way, the variable @code{animals} points to the list
7338 @code{(antelope giraffe lion tiger)}.
7340 Next, evaluate the function @code{setcar} while passing it two
7341 arguments, the variable @code{animals} and the quoted symbol
7342 @code{hippopotamus}; this is done by writing the three element list
7343 @code{(setcar animals 'hippopotamus)} and then evaluating it in the
7347 (setcar animals 'hippopotamus)
7352 After evaluating this expression, evaluate the variable @code{animals}
7353 again. You will see that the list of animals has changed:
7358 @result{} (hippopotamus giraffe lion tiger)
7363 The first element on the list, @code{antelope} is replaced by
7364 @code{hippopotamus}.
7366 So we can see that @code{setcar} did not add a new element to the list
7367 as @code{cons} would have; it replaced @code{antelope} with
7368 @code{hippopotamus}; it @emph{changed} the list.
7371 @section @code{setcdr}
7374 The @code{setcdr} function is similar to the @code{setcar} function,
7375 except that the function replaces the second and subsequent elements of
7376 a list rather than the first element.
7378 (To see how to change the last element of a list, look ahead to
7379 @ref{kill-new function, , The @code{kill-new} function}, which uses
7380 the @code{nthcdr} and @code{setcdr} functions.)
7383 To see how this works, set the value of the variable to a list of
7384 domesticated animals by evaluating the following expression:
7387 (setq domesticated-animals '(horse cow sheep goat))
7392 If you now evaluate the list, you will be returned the list
7393 @code{(horse cow sheep goat)}:
7397 domesticated-animals
7398 @result{} (horse cow sheep goat)
7403 Next, evaluate @code{setcdr} with two arguments, the name of the
7404 variable which has a list as its value, and the list to which the
7405 @sc{cdr} of the first list will be set;
7408 (setcdr domesticated-animals '(cat dog))
7412 If you evaluate this expression, the list @code{(cat dog)} will appear
7413 in the echo area. This is the value returned by the function. The
7414 result we are interested in is the ``side effect'', which we can see by
7415 evaluating the variable @code{domesticated-animals}:
7419 domesticated-animals
7420 @result{} (horse cat dog)
7425 Indeed, the list is changed from @code{(horse cow sheep goat)} to
7426 @code{(horse cat dog)}. The @sc{cdr} of the list is changed from
7427 @code{(cow sheep goat)} to @code{(cat dog)}.
7432 Construct a list of four birds by evaluating several expressions with
7433 @code{cons}. Find out what happens when you @code{cons} a list onto
7434 itself. Replace the first element of the list of four birds with a
7435 fish. Replace the rest of that list with a list of other fish.
7437 @node Cutting & Storing Text
7438 @chapter Cutting and Storing Text
7439 @cindex Cutting and storing text
7440 @cindex Storing and cutting text
7441 @cindex Killing text
7442 @cindex Clipping text
7443 @cindex Erasing text
7444 @cindex Deleting text
7446 Whenever you cut or clip text out of a buffer with a `kill' command in
7447 GNU Emacs, it is stored in a list and you can bring it back with a
7450 (The use of the word `kill' in Emacs for processes which specifically
7451 @emph{do not} destroy the values of the entities is an unfortunate
7452 historical accident. A much more appropriate word would be `clip' since
7453 that is what the kill commands do; they clip text out of a buffer and
7454 put it into storage from which it can be brought back. I have often
7455 been tempted to replace globally all occurrences of `kill' in the Emacs
7456 sources with `clip' and all occurrences of `killed' with `clipped'.)
7459 * Storing Text:: Text is stored in a list.
7460 * zap-to-char:: Cutting out text up to a character.
7461 * kill-region:: Cutting text out of a region.
7462 * copy-region-as-kill:: A definition for copying text.
7463 * Digression into C:: Minor note on C programming language macros.
7464 * defvar:: How to give a variable an initial value.
7465 * cons & search-fwd Review::
7466 * search Exercises::
7471 @unnumberedsec Storing Text in a List
7474 When text is cut out of a buffer, it is stored on a list. Successive
7475 pieces of text are stored on the list successively, so the list might
7479 ("a piece of text" "previous piece")
7484 The function @code{cons} can be used to create a new list from a piece
7485 of text (an `atom', to use the jargon) and an existing list, like
7490 (cons "another piece"
7491 '("a piece of text" "previous piece"))
7497 If you evaluate this expression, a list of three elements will appear in
7501 ("another piece" "a piece of text" "previous piece")
7504 With the @code{car} and @code{nthcdr} functions, you can retrieve
7505 whichever piece of text you want. For example, in the following code,
7506 @code{nthcdr 1 @dots{}} returns the list with the first item removed;
7507 and the @code{car} returns the first element of that remainder---the
7508 second element of the original list:
7512 (car (nthcdr 1 '("another piece"
7515 @result{} "a piece of text"
7519 The actual functions in Emacs are more complex than this, of course.
7520 The code for cutting and retrieving text has to be written so that
7521 Emacs can figure out which element in the list you want---the first,
7522 second, third, or whatever. In addition, when you get to the end of
7523 the list, Emacs should give you the first element of the list, rather
7524 than nothing at all.
7526 The list that holds the pieces of text is called the @dfn{kill ring}.
7527 This chapter leads up to a description of the kill ring and how it is
7528 used by first tracing how the @code{zap-to-char} function works. This
7529 function uses (or `calls') a function that invokes a function that
7530 manipulates the kill ring. Thus, before reaching the mountains, we
7531 climb the foothills.
7533 A subsequent chapter describes how text that is cut from the buffer is
7534 retrieved. @xref{Yanking, , Yanking Text Back}.
7537 @section @code{zap-to-char}
7540 @c FIXME remove obsolete stuff
7541 The @code{zap-to-char} function changed little between GNU Emacs
7542 version 19 and GNU Emacs version 22. However, @code{zap-to-char}
7543 calls another function, @code{kill-region}, which enjoyed a major
7546 The @code{kill-region} function in Emacs 19 is complex, but does not
7547 use code that is important at this time. We will skip it.
7549 The @code{kill-region} function in Emacs 22 is easier to read than the
7550 same function in Emacs 19 and introduces a very important concept,
7551 that of error handling. We will walk through the function.
7553 But first, let us look at the interactive @code{zap-to-char} function.
7556 * Complete zap-to-char:: The complete implementation.
7557 * zap-to-char interactive:: A three part interactive expression.
7558 * zap-to-char body:: A short overview.
7559 * search-forward:: How to search for a string.
7560 * progn:: The @code{progn} special form.
7561 * Summing up zap-to-char:: Using @code{point} and @code{search-forward}.
7565 @node Complete zap-to-char
7566 @unnumberedsubsec The Complete @code{zap-to-char} Implementation
7569 The @code{zap-to-char} function removes the text in the region between
7570 the location of the cursor (i.e., of point) up to and including the
7571 next occurrence of a specified character. The text that
7572 @code{zap-to-char} removes is put in the kill ring; and it can be
7573 retrieved from the kill ring by typing @kbd{C-y} (@code{yank}). If
7574 the command is given an argument, it removes text through that number
7575 of occurrences. Thus, if the cursor were at the beginning of this
7576 sentence and the character were @samp{s}, @samp{Thus} would be
7577 removed. If the argument were two, @samp{Thus, if the curs} would be
7578 removed, up to and including the @samp{s} in @samp{cursor}.
7580 If the specified character is not found, @code{zap-to-char} will say
7581 ``Search failed'', tell you the character you typed, and not remove
7584 In order to determine how much text to remove, @code{zap-to-char} uses
7585 a search function. Searches are used extensively in code that
7586 manipulates text, and we will focus attention on them as well as on the
7590 @c GNU Emacs version 19
7591 (defun zap-to-char (arg char) ; version 19 implementation
7592 "Kill up to and including ARG'th occurrence of CHAR.
7593 Goes backward if ARG is negative; error if CHAR not found."
7594 (interactive "*p\ncZap to char: ")
7595 (kill-region (point)
7598 (char-to-string char) nil nil arg)
7603 Here is the complete text of the version 22 implementation of the function:
7608 (defun zap-to-char (arg char)
7609 "Kill up to and including ARG'th occurrence of CHAR.
7610 Case is ignored if `case-fold-search' is non-nil in the current buffer.
7611 Goes backward if ARG is negative; error if CHAR not found."
7612 (interactive "p\ncZap to char: ")
7613 (if (char-table-p translation-table-for-input)
7614 (setq char (or (aref translation-table-for-input char) char)))
7615 (kill-region (point) (progn
7616 (search-forward (char-to-string char)
7622 The documentation is thorough. You do need to know the jargon meaning
7625 @node zap-to-char interactive
7626 @subsection The @code{interactive} Expression
7629 The interactive expression in the @code{zap-to-char} command looks like
7633 (interactive "p\ncZap to char: ")
7636 The part within quotation marks, @code{"p\ncZap to char:@: "}, specifies
7637 two different things. First, and most simply, is the @samp{p}.
7638 This part is separated from the next part by a newline, @samp{\n}.
7639 The @samp{p} means that the first argument to the function will be
7640 passed the value of a `processed prefix'. The prefix argument is
7641 passed by typing @kbd{C-u} and a number, or @kbd{M-} and a number. If
7642 the function is called interactively without a prefix, 1 is passed to
7645 The second part of @code{"p\ncZap to char:@: "} is
7646 @samp{cZap to char:@: }. In this part, the lower case @samp{c}
7647 indicates that @code{interactive} expects a prompt and that the
7648 argument will be a character. The prompt follows the @samp{c} and is
7649 the string @samp{Zap to char:@: } (with a space after the colon to
7652 What all this does is prepare the arguments to @code{zap-to-char} so they
7653 are of the right type, and give the user a prompt.
7655 In a read-only buffer, the @code{zap-to-char} function copies the text
7656 to the kill ring, but does not remove it. The echo area displays a
7657 message saying that the buffer is read-only. Also, the terminal may
7658 beep or blink at you.
7660 @node zap-to-char body
7661 @subsection The Body of @code{zap-to-char}
7663 The body of the @code{zap-to-char} function contains the code that
7664 kills (that is, removes) the text in the region from the current
7665 position of the cursor up to and including the specified character.
7667 The first part of the code looks like this:
7670 (if (char-table-p translation-table-for-input)
7671 (setq char (or (aref translation-table-for-input char) char)))
7672 (kill-region (point) (progn
7673 (search-forward (char-to-string char) nil nil arg)
7678 @code{char-table-p} is an hitherto unseen function. It determines
7679 whether its argument is a character table. When it is, it sets the
7680 character passed to @code{zap-to-char} to one of them, if that
7681 character exists, or to the character itself. (This becomes important
7682 for certain characters in non-European languages. The @code{aref}
7683 function extracts an element from an array. It is an array-specific
7684 function that is not described in this document. @xref{Arrays, ,
7685 Arrays, elisp, The GNU Emacs Lisp Reference Manual}.)
7688 @code{(point)} is the current position of the cursor.
7690 The next part of the code is an expression using @code{progn}. The body
7691 of the @code{progn} consists of calls to @code{search-forward} and
7694 It is easier to understand how @code{progn} works after learning about
7695 @code{search-forward}, so we will look at @code{search-forward} and
7696 then at @code{progn}.
7698 @node search-forward
7699 @subsection The @code{search-forward} Function
7700 @findex search-forward
7702 The @code{search-forward} function is used to locate the
7703 zapped-for-character in @code{zap-to-char}. If the search is
7704 successful, @code{search-forward} leaves point immediately after the
7705 last character in the target string. (In @code{zap-to-char}, the
7706 target string is just one character long. @code{zap-to-char} uses the
7707 function @code{char-to-string} to ensure that the computer treats that
7708 character as a string.) If the search is backwards,
7709 @code{search-forward} leaves point just before the first character in
7710 the target. Also, @code{search-forward} returns @code{t} for true.
7711 (Moving point is therefore a `side effect'.)
7714 In @code{zap-to-char}, the @code{search-forward} function looks like this:
7717 (search-forward (char-to-string char) nil nil arg)
7720 The @code{search-forward} function takes four arguments:
7724 The first argument is the target, what is searched for. This must be a
7725 string, such as @samp{"z"}.
7727 As it happens, the argument passed to @code{zap-to-char} is a single
7728 character. Because of the way computers are built, the Lisp
7729 interpreter may treat a single character as being different from a
7730 string of characters. Inside the computer, a single character has a
7731 different electronic format than a string of one character. (A single
7732 character can often be recorded in the computer using exactly one
7733 byte; but a string may be longer, and the computer needs to be ready
7734 for this.) Since the @code{search-forward} function searches for a
7735 string, the character that the @code{zap-to-char} function receives as
7736 its argument must be converted inside the computer from one format to
7737 the other; otherwise the @code{search-forward} function will fail.
7738 The @code{char-to-string} function is used to make this conversion.
7741 The second argument bounds the search; it is specified as a position in
7742 the buffer. In this case, the search can go to the end of the buffer,
7743 so no bound is set and the second argument is @code{nil}.
7746 The third argument tells the function what it should do if the search
7747 fails---it can signal an error (and print a message) or it can return
7748 @code{nil}. A @code{nil} as the third argument causes the function to
7749 signal an error when the search fails.
7752 The fourth argument to @code{search-forward} is the repeat count---how
7753 many occurrences of the string to look for. This argument is optional
7754 and if the function is called without a repeat count, this argument is
7755 passed the value 1. If this argument is negative, the search goes
7760 In template form, a @code{search-forward} expression looks like this:
7764 (search-forward "@var{target-string}"
7765 @var{limit-of-search}
7766 @var{what-to-do-if-search-fails}
7771 We will look at @code{progn} next.
7774 @subsection The @code{progn} Special Form
7777 @code{progn} is a special form that causes each of its arguments to be
7778 evaluated in sequence and then returns the value of the last one. The
7779 preceding expressions are evaluated only for the side effects they
7780 perform. The values produced by them are discarded.
7783 The template for a @code{progn} expression is very simple:
7792 In @code{zap-to-char}, the @code{progn} expression has to do two things:
7793 put point in exactly the right position; and return the location of
7794 point so that @code{kill-region} will know how far to kill to.
7796 The first argument to the @code{progn} is @code{search-forward}. When
7797 @code{search-forward} finds the string, the function leaves point
7798 immediately after the last character in the target string. (In this
7799 case the target string is just one character long.) If the search is
7800 backwards, @code{search-forward} leaves point just before the first
7801 character in the target. The movement of point is a side effect.
7803 The second and last argument to @code{progn} is the expression
7804 @code{(point)}. This expression returns the value of point, which in
7805 this case will be the location to which it has been moved by
7806 @code{search-forward}. (In the source, a line that tells the function
7807 to go to the previous character, if it is going forward, was commented
7808 out in 1999; I don't remember whether that feature or mis-feature was
7809 ever a part of the distributed source.) The value of @code{point} is
7810 returned by the @code{progn} expression and is passed to
7811 @code{kill-region} as @code{kill-region}'s second argument.
7813 @node Summing up zap-to-char
7814 @subsection Summing up @code{zap-to-char}
7816 Now that we have seen how @code{search-forward} and @code{progn} work,
7817 we can see how the @code{zap-to-char} function works as a whole.
7819 The first argument to @code{kill-region} is the position of the cursor
7820 when the @code{zap-to-char} command is given---the value of point at
7821 that time. Within the @code{progn}, the search function then moves
7822 point to just after the zapped-to-character and @code{point} returns the
7823 value of this location. The @code{kill-region} function puts together
7824 these two values of point, the first one as the beginning of the region
7825 and the second one as the end of the region, and removes the region.
7827 The @code{progn} special form is necessary because the
7828 @code{kill-region} command takes two arguments; and it would fail if
7829 @code{search-forward} and @code{point} expressions were written in
7830 sequence as two additional arguments. The @code{progn} expression is
7831 a single argument to @code{kill-region} and returns the one value that
7832 @code{kill-region} needs for its second argument.
7835 @section @code{kill-region}
7838 The @code{zap-to-char} function uses the @code{kill-region} function.
7839 This function clips text from a region and copies that text to
7840 the kill ring, from which it may be retrieved.
7845 (defun kill-region (beg end &optional yank-handler)
7846 "Kill (\"cut\") text between point and mark.
7847 This deletes the text from the buffer and saves it in the kill ring.
7848 The command \\[yank] can retrieve it from there.
7849 \(If you want to kill and then yank immediately, use \\[kill-ring-save].)
7851 If you want to append the killed region to the last killed text,
7852 use \\[append-next-kill] before \\[kill-region].
7854 If the buffer is read-only, Emacs will beep and refrain from deleting
7855 the text, but put the text in the kill ring anyway. This means that
7856 you can use the killing commands to copy text from a read-only buffer.
7858 This is the primitive for programs to kill text (as opposed to deleting it).
7859 Supply two arguments, character positions indicating the stretch of text
7861 Any command that calls this function is a \"kill command\".
7862 If the previous command was also a kill command,
7863 the text killed this time appends to the text killed last time
7864 to make one entry in the kill ring.
7866 In Lisp code, optional third arg YANK-HANDLER, if non-nil,
7867 specifies the yank-handler text property to be set on the killed
7868 text. See `insert-for-yank'."
7869 ;; Pass point first, then mark, because the order matters
7870 ;; when calling kill-append.
7871 (interactive (list (point) (mark)))
7872 (unless (and beg end)
7873 (error "The mark is not set now, so there is no region"))
7875 (let ((string (filter-buffer-substring beg end t)))
7876 (when string ;STRING is nil if BEG = END
7877 ;; Add that string to the kill ring, one way or another.
7878 (if (eq last-command 'kill-region)
7879 (kill-append string (< end beg) yank-handler)
7880 (kill-new string nil yank-handler)))
7881 (when (or string (eq last-command 'kill-region))
7882 (setq this-command 'kill-region))
7884 ((buffer-read-only text-read-only)
7885 ;; The code above failed because the buffer, or some of the characters
7886 ;; in the region, are read-only.
7887 ;; We should beep, in case the user just isn't aware of this.
7888 ;; However, there's no harm in putting
7889 ;; the region's text in the kill ring, anyway.
7890 (copy-region-as-kill beg end)
7891 ;; Set this-command now, so it will be set even if we get an error.
7892 (setq this-command 'kill-region)
7893 ;; This should barf, if appropriate, and give us the correct error.
7894 (if kill-read-only-ok
7895 (progn (message "Read only text copied to kill ring") nil)
7896 ;; Signal an error if the buffer is read-only.
7897 (barf-if-buffer-read-only)
7898 ;; If the buffer isn't read-only, the text is.
7899 (signal 'text-read-only (list (current-buffer)))))))
7902 The Emacs 22 version of that function uses @code{condition-case} and
7903 @code{copy-region-as-kill}, both of which we will explain.
7904 @code{condition-case} is an important special form.
7906 In essence, the @code{kill-region} function calls
7907 @code{condition-case}, which takes three arguments. In this function,
7908 the first argument does nothing. The second argument contains the
7909 code that does the work when all goes well. The third argument
7910 contains the code that is called in the event of an error.
7913 * Complete kill-region:: The function definition.
7914 * condition-case:: Dealing with a problem.
7919 @node Complete kill-region
7920 @unnumberedsubsec The Complete @code{kill-region} Definition
7924 We will go through the @code{condition-case} code in a moment. First,
7925 let us look at the definition of @code{kill-region}, with comments
7931 (defun kill-region (beg end)
7932 "Kill (\"cut\") text between point and mark.
7933 This deletes the text from the buffer and saves it in the kill ring.
7934 The command \\[yank] can retrieve it from there. @dots{} "
7938 ;; @bullet{} Since order matters, pass point first.
7939 (interactive (list (point) (mark)))
7940 ;; @bullet{} And tell us if we cannot cut the text.
7941 ;; `unless' is an `if' without a then-part.
7942 (unless (and beg end)
7943 (error "The mark is not set now, so there is no region"))
7947 ;; @bullet{} `condition-case' takes three arguments.
7948 ;; If the first argument is nil, as it is here,
7949 ;; information about the error signal is not
7950 ;; stored for use by another function.
7955 ;; @bullet{} The second argument to `condition-case' tells the
7956 ;; Lisp interpreter what to do when all goes well.
7960 ;; It starts with a `let' function that extracts the string
7961 ;; and tests whether it exists. If so (that is what the
7962 ;; `when' checks), it calls an `if' function that determines
7963 ;; whether the previous command was another call to
7964 ;; `kill-region'; if it was, then the new text is appended to
7965 ;; the previous text; if not, then a different function,
7966 ;; `kill-new', is called.
7970 ;; The `kill-append' function concatenates the new string and
7971 ;; the old. The `kill-new' function inserts text into a new
7972 ;; item in the kill ring.
7976 ;; `when' is an `if' without an else-part. The second `when'
7977 ;; again checks whether the current string exists; in
7978 ;; addition, it checks whether the previous command was
7979 ;; another call to `kill-region'. If one or the other
7980 ;; condition is true, then it sets the current command to
7981 ;; be `kill-region'.
7984 (let ((string (filter-buffer-substring beg end t)))
7985 (when string ;STRING is nil if BEG = END
7986 ;; Add that string to the kill ring, one way or another.
7987 (if (eq last-command 'kill-region)
7990 ;; @minus{} `yank-handler' is an optional argument to
7991 ;; `kill-region' that tells the `kill-append' and
7992 ;; `kill-new' functions how deal with properties
7993 ;; added to the text, such as `bold' or `italics'.
7994 (kill-append string (< end beg) yank-handler)
7995 (kill-new string nil yank-handler)))
7996 (when (or string (eq last-command 'kill-region))
7997 (setq this-command 'kill-region))
8002 ;; @bullet{} The third argument to `condition-case' tells the interpreter
8003 ;; what to do with an error.
8006 ;; The third argument has a conditions part and a body part.
8007 ;; If the conditions are met (in this case,
8008 ;; if text or buffer are read-only)
8009 ;; then the body is executed.
8012 ;; The first part of the third argument is the following:
8013 ((buffer-read-only text-read-only) ;; the if-part
8014 ;; @dots{} the then-part
8015 (copy-region-as-kill beg end)
8018 ;; Next, also as part of the then-part, set this-command, so
8019 ;; it will be set in an error
8020 (setq this-command 'kill-region)
8021 ;; Finally, in the then-part, send a message if you may copy
8022 ;; the text to the kill ring without signaling an error, but
8023 ;; don't if you may not.
8026 (if kill-read-only-ok
8027 (progn (message "Read only text copied to kill ring") nil)
8028 (barf-if-buffer-read-only)
8029 ;; If the buffer isn't read-only, the text is.
8030 (signal 'text-read-only (list (current-buffer)))))
8038 (defun kill-region (beg end)
8039 "Kill between point and mark.
8040 The text is deleted but saved in the kill ring."
8045 ;; 1. `condition-case' takes three arguments.
8046 ;; If the first argument is nil, as it is here,
8047 ;; information about the error signal is not
8048 ;; stored for use by another function.
8053 ;; 2. The second argument to `condition-case'
8054 ;; tells the Lisp interpreter what to do when all goes well.
8058 ;; The `delete-and-extract-region' function usually does the
8059 ;; work. If the beginning and ending of the region are both
8060 ;; the same, then the variable `string' will be empty, or nil
8061 (let ((string (delete-and-extract-region beg end)))
8065 ;; `when' is an `if' clause that cannot take an `else-part'.
8066 ;; Emacs normally sets the value of `last-command' to the
8067 ;; previous command.
8070 ;; `kill-append' concatenates the new string and the old.
8071 ;; `kill-new' inserts text into a new item in the kill ring.
8073 (if (eq last-command 'kill-region)
8074 ;; if true, prepend string
8075 (kill-append string (< end beg))
8077 (setq this-command 'kill-region))
8081 ;; 3. The third argument to `condition-case' tells the interpreter
8082 ;; what to do with an error.
8085 ;; The third argument has a conditions part and a body part.
8086 ;; If the conditions are met (in this case,
8087 ;; if text or buffer are read-only)
8088 ;; then the body is executed.
8091 ((buffer-read-only text-read-only) ;; this is the if-part
8093 (copy-region-as-kill beg end)
8096 (if kill-read-only-ok ;; usually this variable is nil
8097 (message "Read only text copied to kill ring")
8098 ;; or else, signal an error if the buffer is read-only;
8099 (barf-if-buffer-read-only)
8100 ;; and, in any case, signal that the text is read-only.
8101 (signal 'text-read-only (list (current-buffer)))))))
8106 @node condition-case
8107 @subsection @code{condition-case}
8108 @findex condition-case
8110 As we have seen earlier (@pxref{Making Errors, , Generate an Error
8111 Message}), when the Emacs Lisp interpreter has trouble evaluating an
8112 expression, it provides you with help; in the jargon, this is called
8113 ``signaling an error''. Usually, the computer stops the program and
8114 shows you a message.
8116 However, some programs undertake complicated actions. They should not
8117 simply stop on an error. In the @code{kill-region} function, the most
8118 likely error is that you will try to kill text that is read-only and
8119 cannot be removed. So the @code{kill-region} function contains code
8120 to handle this circumstance. This code, which makes up the body of
8121 the @code{kill-region} function, is inside of a @code{condition-case}
8125 The template for @code{condition-case} looks like this:
8132 @var{error-handler}@dots{})
8136 The second argument, @var{bodyform}, is straightforward. The
8137 @code{condition-case} special form causes the Lisp interpreter to
8138 evaluate the code in @var{bodyform}. If no error occurs, the special
8139 form returns the code's value and produces the side-effects, if any.
8141 In short, the @var{bodyform} part of a @code{condition-case}
8142 expression determines what should happen when everything works
8145 However, if an error occurs, among its other actions, the function
8146 generating the error signal will define one or more error condition
8149 An error handler is the third argument to @code{condition case}.
8150 An error handler has two parts, a @var{condition-name} and a
8151 @var{body}. If the @var{condition-name} part of an error handler
8152 matches a condition name generated by an error, then the @var{body}
8153 part of the error handler is run.
8155 As you will expect, the @var{condition-name} part of an error handler
8156 may be either a single condition name or a list of condition names.
8158 Also, a complete @code{condition-case} expression may contain more
8159 than one error handler. When an error occurs, the first applicable
8162 Lastly, the first argument to the @code{condition-case} expression,
8163 the @var{var} argument, is sometimes bound to a variable that
8164 contains information about the error. However, if that argument is
8165 nil, as is the case in @code{kill-region}, that information is
8169 In brief, in the @code{kill-region} function, the code
8170 @code{condition-case} works like this:
8174 @var{If no errors}, @var{run only this code}
8175 @var{but}, @var{if errors}, @var{run this other code}.
8182 copy-region-as-kill is short, 12 lines, and uses
8183 filter-buffer-substring, which is longer, 39 lines
8184 and has delete-and-extract-region in it.
8185 delete-and-extract-region is written in C.
8187 see Initializing a Variable with @code{defvar}
8189 Initializing a Variable with @code{defvar} includes line 8350
8193 @subsection Lisp macro
8197 The part of the @code{condition-case} expression that is evaluated in
8198 the expectation that all goes well has a @code{when}. The code uses
8199 @code{when} to determine whether the @code{string} variable points to
8202 A @code{when} expression is simply a programmers' convenience. It is
8203 an @code{if} without the possibility of an else clause. In your mind,
8204 you can replace @code{when} with @code{if} and understand what goes
8205 on. That is what the Lisp interpreter does.
8207 Technically speaking, @code{when} is a Lisp macro. A Lisp macro
8208 enables you to define new control constructs and other language
8209 features. It tells the interpreter how to compute another Lisp
8210 expression which will in turn compute the value. In this case, the
8211 `other expression' is an @code{if} expression.
8213 The @code{kill-region} function definition also has an @code{unless}
8214 macro; it is the converse of @code{when}. The @code{unless} macro is
8215 an @code{if} without a then clause
8217 For more about Lisp macros, see @ref{Macros, , Macros, elisp, The GNU
8218 Emacs Lisp Reference Manual}. The C programming language also
8219 provides macros. These are different, but also useful.
8222 We will briefly look at C macros in
8223 @ref{Digression into C}.
8227 Regarding the @code{when} macro, in the @code{condition-case}
8228 expression, when the string has content, then another conditional
8229 expression is executed. This is an @code{if} with both a then-part
8234 (if (eq last-command 'kill-region)
8235 (kill-append string (< end beg) yank-handler)
8236 (kill-new string nil yank-handler))
8240 The then-part is evaluated if the previous command was another call to
8241 @code{kill-region}; if not, the else-part is evaluated.
8243 @code{yank-handler} is an optional argument to @code{kill-region} that
8244 tells the @code{kill-append} and @code{kill-new} functions how deal
8245 with properties added to the text, such as `bold' or `italics'.
8247 @code{last-command} is a variable that comes with Emacs that we have
8248 not seen before. Normally, whenever a function is executed, Emacs
8249 sets the value of @code{last-command} to the previous command.
8252 In this segment of the definition, the @code{if} expression checks
8253 whether the previous command was @code{kill-region}. If it was,
8256 (kill-append string (< end beg) yank-handler)
8260 concatenates a copy of the newly clipped text to the just previously
8261 clipped text in the kill ring.
8263 @node copy-region-as-kill
8264 @section @code{copy-region-as-kill}
8265 @findex copy-region-as-kill
8268 The @code{copy-region-as-kill} function copies a region of text from a
8269 buffer and (via either @code{kill-append} or @code{kill-new}) saves it
8270 in the @code{kill-ring}.
8272 If you call @code{copy-region-as-kill} immediately after a
8273 @code{kill-region} command, Emacs appends the newly copied text to the
8274 previously copied text. This means that if you yank back the text, you
8275 get it all, from both this and the previous operation. On the other
8276 hand, if some other command precedes the @code{copy-region-as-kill},
8277 the function copies the text into a separate entry in the kill ring.
8280 * Complete copy-region-as-kill:: The complete function definition.
8281 * copy-region-as-kill body:: The body of @code{copy-region-as-kill}.
8285 @node Complete copy-region-as-kill
8286 @unnumberedsubsec The complete @code{copy-region-as-kill} function definition
8290 Here is the complete text of the version 22 @code{copy-region-as-kill}
8295 (defun copy-region-as-kill (beg end)
8296 "Save the region as if killed, but don't kill it.
8297 In Transient Mark mode, deactivate the mark.
8298 If `interprogram-cut-function' is non-nil, also save the text for a window
8299 system cut and paste."
8303 (if (eq last-command 'kill-region)
8304 (kill-append (filter-buffer-substring beg end) (< end beg))
8305 (kill-new (filter-buffer-substring beg end)))
8308 (if transient-mark-mode
8309 (setq deactivate-mark t))
8315 As usual, this function can be divided into its component parts:
8319 (defun copy-region-as-kill (@var{argument-list})
8320 "@var{documentation}@dots{}"
8326 The arguments are @code{beg} and @code{end} and the function is
8327 interactive with @code{"r"}, so the two arguments must refer to the
8328 beginning and end of the region. If you have been reading though this
8329 document from the beginning, understanding these parts of a function is
8330 almost becoming routine.
8332 The documentation is somewhat confusing unless you remember that the
8333 word `kill' has a meaning different from usual. The `Transient Mark'
8334 and @code{interprogram-cut-function} comments explain certain
8337 After you once set a mark, a buffer always contains a region. If you
8338 wish, you can use Transient Mark mode to highlight the region
8339 temporarily. (No one wants to highlight the region all the time, so
8340 Transient Mark mode highlights it only at appropriate times. Many
8341 people turn off Transient Mark mode, so the region is never
8344 Also, a windowing system allows you to copy, cut, and paste among
8345 different programs. In the X windowing system, for example, the
8346 @code{interprogram-cut-function} function is @code{x-select-text},
8347 which works with the windowing system's equivalent of the Emacs kill
8350 The body of the @code{copy-region-as-kill} function starts with an
8351 @code{if} clause. What this clause does is distinguish between two
8352 different situations: whether or not this command is executed
8353 immediately after a previous @code{kill-region} command. In the first
8354 case, the new region is appended to the previously copied text.
8355 Otherwise, it is inserted into the beginning of the kill ring as a
8356 separate piece of text from the previous piece.
8358 The last two lines of the function prevent the region from lighting up
8359 if Transient Mark mode is turned on.
8361 The body of @code{copy-region-as-kill} merits discussion in detail.
8363 @node copy-region-as-kill body
8364 @subsection The Body of @code{copy-region-as-kill}
8366 The @code{copy-region-as-kill} function works in much the same way as
8367 the @code{kill-region} function. Both are written so that two or more
8368 kills in a row combine their text into a single entry. If you yank
8369 back the text from the kill ring, you get it all in one piece.
8370 Moreover, kills that kill forward from the current position of the
8371 cursor are added to the end of the previously copied text and commands
8372 that copy text backwards add it to the beginning of the previously
8373 copied text. This way, the words in the text stay in the proper
8376 Like @code{kill-region}, the @code{copy-region-as-kill} function makes
8377 use of the @code{last-command} variable that keeps track of the
8378 previous Emacs command.
8381 * last-command & this-command::
8382 * kill-append function::
8383 * kill-new function::
8387 @node last-command & this-command
8388 @unnumberedsubsubsec @code{last-command} and @code{this-command}
8391 Normally, whenever a function is executed, Emacs sets the value of
8392 @code{this-command} to the function being executed (which in this case
8393 would be @code{copy-region-as-kill}). At the same time, Emacs sets
8394 the value of @code{last-command} to the previous value of
8395 @code{this-command}.
8397 In the first part of the body of the @code{copy-region-as-kill}
8398 function, an @code{if} expression determines whether the value of
8399 @code{last-command} is @code{kill-region}. If so, the then-part of
8400 the @code{if} expression is evaluated; it uses the @code{kill-append}
8401 function to concatenate the text copied at this call to the function
8402 with the text already in the first element (the @sc{car}) of the kill
8403 ring. On the other hand, if the value of @code{last-command} is not
8404 @code{kill-region}, then the @code{copy-region-as-kill} function
8405 attaches a new element to the kill ring using the @code{kill-new}
8409 The @code{if} expression reads as follows; it uses @code{eq}:
8413 (if (eq last-command 'kill-region)
8415 (kill-append (filter-buffer-substring beg end) (< end beg))
8417 (kill-new (filter-buffer-substring beg end)))
8421 @findex filter-buffer-substring
8422 (The @code{filter-buffer-substring} function returns a filtered
8423 substring of the buffer, if any. Optionally---the arguments are not
8424 here, so neither is done---the function may delete the initial text or
8425 return the text without its properties; this function is a replacement
8426 for the older @code{buffer-substring} function, which came before text
8427 properties were implemented.)
8429 @findex eq @r{(example of use)}
8431 The @code{eq} function tests whether its first argument is the same Lisp
8432 object as its second argument. The @code{eq} function is similar to the
8433 @code{equal} function in that it is used to test for equality, but
8434 differs in that it determines whether two representations are actually
8435 the same object inside the computer, but with different names.
8436 @code{equal} determines whether the structure and contents of two
8437 expressions are the same.
8439 If the previous command was @code{kill-region}, then the Emacs Lisp
8440 interpreter calls the @code{kill-append} function
8442 @node kill-append function
8443 @unnumberedsubsubsec The @code{kill-append} function
8447 The @code{kill-append} function looks like this:
8452 (defun kill-append (string before-p &optional yank-handler)
8453 "Append STRING to the end of the latest kill in the kill ring.
8454 If BEFORE-P is non-nil, prepend STRING to the kill.
8456 (let* ((cur (car kill-ring)))
8457 (kill-new (if before-p (concat string cur) (concat cur string))
8458 (or (= (length cur) 0)
8460 (get-text-property 0 'yank-handler cur)))
8467 (defun kill-append (string before-p)
8468 "Append STRING to the end of the latest kill in the kill ring.
8469 If BEFORE-P is non-nil, prepend STRING to the kill.
8470 If `interprogram-cut-function' is set, pass the resulting kill to
8472 (kill-new (if before-p
8473 (concat string (car kill-ring))
8474 (concat (car kill-ring) string))
8479 The @code{kill-append} function is fairly straightforward. It uses
8480 the @code{kill-new} function, which we will discuss in more detail in
8483 (Also, the function provides an optional argument called
8484 @code{yank-handler}; when invoked, this argument tells the function
8485 how to deal with properties added to the text, such as `bold' or
8488 @c !!! bug in GNU Emacs 22 version of kill-append ?
8489 It has a @code{let*} function to set the value of the first element of
8490 the kill ring to @code{cur}. (I do not know why the function does not
8491 use @code{let} instead; only one value is set in the expression.
8492 Perhaps this is a bug that produces no problems?)
8494 Consider the conditional that is one of the two arguments to
8495 @code{kill-new}. It uses @code{concat} to concatenate the new text to
8496 the @sc{car} of the kill ring. Whether it prepends or appends the
8497 text depends on the results of an @code{if} expression:
8501 (if before-p ; @r{if-part}
8502 (concat string cur) ; @r{then-part}
8503 (concat cur string)) ; @r{else-part}
8508 If the region being killed is before the region that was killed in the
8509 last command, then it should be prepended before the material that was
8510 saved in the previous kill; and conversely, if the killed text follows
8511 what was just killed, it should be appended after the previous text.
8512 The @code{if} expression depends on the predicate @code{before-p} to
8513 decide whether the newly saved text should be put before or after the
8514 previously saved text.
8516 The symbol @code{before-p} is the name of one of the arguments to
8517 @code{kill-append}. When the @code{kill-append} function is
8518 evaluated, it is bound to the value returned by evaluating the actual
8519 argument. In this case, this is the expression @code{(< end beg)}.
8520 This expression does not directly determine whether the killed text in
8521 this command is located before or after the kill text of the last
8522 command; what it does is determine whether the value of the variable
8523 @code{end} is less than the value of the variable @code{beg}. If it
8524 is, it means that the user is most likely heading towards the
8525 beginning of the buffer. Also, the result of evaluating the predicate
8526 expression, @code{(< end beg)}, will be true and the text will be
8527 prepended before the previous text. On the other hand, if the value of
8528 the variable @code{end} is greater than the value of the variable
8529 @code{beg}, the text will be appended after the previous text.
8532 When the newly saved text will be prepended, then the string with the new
8533 text will be concatenated before the old text:
8541 But if the text will be appended, it will be concatenated
8545 (concat cur string))
8548 To understand how this works, we first need to review the
8549 @code{concat} function. The @code{concat} function links together or
8550 unites two strings of text. The result is a string. For example:
8554 (concat "abc" "def")
8560 (car '("first element" "second element")))
8561 @result{} "new first element"
8564 '("first element" "second element")) " modified")
8565 @result{} "first element modified"
8569 We can now make sense of @code{kill-append}: it modifies the contents
8570 of the kill ring. The kill ring is a list, each element of which is
8571 saved text. The @code{kill-append} function uses the @code{kill-new}
8572 function which in turn uses the @code{setcar} function.
8574 @node kill-new function
8575 @unnumberedsubsubsec The @code{kill-new} function
8578 @c in GNU Emacs 22, additional documentation to kill-new:
8580 Optional third arguments YANK-HANDLER controls how the STRING is later
8581 inserted into a buffer; see `insert-for-yank' for details.
8582 When a yank handler is specified, STRING must be non-empty (the yank
8583 handler, if non-nil, is stored as a `yank-handler' text property on STRING).
8585 When the yank handler has a non-nil PARAM element, the original STRING
8586 argument is not used by `insert-for-yank'. However, since Lisp code
8587 may access and use elements from the kill ring directly, the STRING
8588 argument should still be a \"useful\" string for such uses."
8591 The @code{kill-new} function looks like this:
8595 (defun kill-new (string &optional replace yank-handler)
8596 "Make STRING the latest kill in the kill ring.
8597 Set `kill-ring-yank-pointer' to point to it.
8599 If `interprogram-cut-function' is non-nil, apply it to STRING.
8600 Optional second argument REPLACE non-nil means that STRING will replace
8601 the front of the kill ring, rather than being added to the list.
8605 (if (> (length string) 0)
8607 (put-text-property 0 (length string)
8608 'yank-handler yank-handler string))
8610 (signal 'args-out-of-range
8611 (list string "yank-handler specified for empty string"))))
8614 (if (fboundp 'menu-bar-update-yank-menu)
8615 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8618 (if (and replace kill-ring)
8619 (setcar kill-ring string)
8620 (push string kill-ring)
8621 (if (> (length kill-ring) kill-ring-max)
8622 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8625 (setq kill-ring-yank-pointer kill-ring)
8626 (if interprogram-cut-function
8627 (funcall interprogram-cut-function string (not replace))))
8632 (defun kill-new (string &optional replace)
8633 "Make STRING the latest kill in the kill ring.
8634 Set the kill-ring-yank pointer to point to it.
8635 If `interprogram-cut-function' is non-nil, apply it to STRING.
8636 Optional second argument REPLACE non-nil means that STRING will replace
8637 the front of the kill ring, rather than being added to the list."
8638 (and (fboundp 'menu-bar-update-yank-menu)
8639 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8640 (if (and replace kill-ring)
8641 (setcar kill-ring string)
8642 (setq kill-ring (cons string kill-ring))
8643 (if (> (length kill-ring) kill-ring-max)
8644 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8645 (setq kill-ring-yank-pointer kill-ring)
8646 (if interprogram-cut-function
8647 (funcall interprogram-cut-function string (not replace))))
8650 (Notice that the function is not interactive.)
8652 As usual, we can look at this function in parts.
8654 The function definition has an optional @code{yank-handler} argument,
8655 which when invoked tells the function how to deal with properties
8656 added to the text, such as `bold' or `italics'. We will skip that.
8659 The first line of the documentation makes sense:
8662 Make STRING the latest kill in the kill ring.
8666 Let's skip over the rest of the documentation for the moment.
8669 Also, let's skip over the initial @code{if} expression and those lines
8670 of code involving @code{menu-bar-update-yank-menu}. We will explain
8674 The critical lines are these:
8678 (if (and replace kill-ring)
8680 (setcar kill-ring string)
8684 (push string kill-ring)
8687 (setq kill-ring (cons string kill-ring))
8688 (if (> (length kill-ring) kill-ring-max)
8689 ;; @r{avoid overly long kill ring}
8690 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil)))
8693 (setq kill-ring-yank-pointer kill-ring)
8694 (if interprogram-cut-function
8695 (funcall interprogram-cut-function string (not replace))))
8699 The conditional test is @w{@code{(and replace kill-ring)}}.
8700 This will be true when two conditions are met: the kill ring has
8701 something in it, and the @code{replace} variable is true.
8704 When the @code{kill-append} function sets @code{replace} to be true
8705 and when the kill ring has at least one item in it, the @code{setcar}
8706 expression is executed:
8709 (setcar kill-ring string)
8712 The @code{setcar} function actually changes the first element of the
8713 @code{kill-ring} list to the value of @code{string}. It replaces the
8717 On the other hand, if the kill ring is empty, or replace is false, the
8718 else-part of the condition is executed:
8721 (push string kill-ring)
8726 @code{push} puts its first argument onto the second. It is similar to
8730 (setq kill-ring (cons string kill-ring))
8738 (add-to-list kill-ring string)
8742 When it is false, the expression first constructs a new version of the
8743 kill ring by prepending @code{string} to the existing kill ring as a
8744 new element (that is what the @code{push} does). Then it executes a
8745 second @code{if} clause. This second @code{if} clause keeps the kill
8746 ring from growing too long.
8748 Let's look at these two expressions in order.
8750 The @code{push} line of the else-part sets the new value of the kill
8751 ring to what results from adding the string being killed to the old
8754 We can see how this works with an example.
8760 (setq example-list '("here is a clause" "another clause"))
8765 After evaluating this expression with @kbd{C-x C-e}, you can evaluate
8766 @code{example-list} and see what it returns:
8771 @result{} ("here is a clause" "another clause")
8777 Now, we can add a new element on to this list by evaluating the
8778 following expression:
8779 @findex push, @r{example}
8782 (push "a third clause" example-list)
8787 When we evaluate @code{example-list}, we find its value is:
8792 @result{} ("a third clause" "here is a clause" "another clause")
8797 Thus, the third clause is added to the list by @code{push}.
8800 Now for the second part of the @code{if} clause. This expression
8801 keeps the kill ring from growing too long. It looks like this:
8805 (if (> (length kill-ring) kill-ring-max)
8806 (setcdr (nthcdr (1- kill-ring-max) kill-ring) nil))
8810 The code checks whether the length of the kill ring is greater than
8811 the maximum permitted length. This is the value of
8812 @code{kill-ring-max} (which is 60, by default). If the length of the
8813 kill ring is too long, then this code sets the last element of the
8814 kill ring to @code{nil}. It does this by using two functions,
8815 @code{nthcdr} and @code{setcdr}.
8817 We looked at @code{setcdr} earlier (@pxref{setcdr, , @code{setcdr}}).
8818 It sets the @sc{cdr} of a list, just as @code{setcar} sets the
8819 @sc{car} of a list. In this case, however, @code{setcdr} will not be
8820 setting the @sc{cdr} of the whole kill ring; the @code{nthcdr}
8821 function is used to cause it to set the @sc{cdr} of the next to last
8822 element of the kill ring---this means that since the @sc{cdr} of the
8823 next to last element is the last element of the kill ring, it will set
8824 the last element of the kill ring.
8826 @findex nthcdr, @r{example}
8827 The @code{nthcdr} function works by repeatedly taking the @sc{cdr} of a
8828 list---it takes the @sc{cdr} of the @sc{cdr} of the @sc{cdr}
8829 @dots{} It does this @var{N} times and returns the results.
8830 (@xref{nthcdr, , @code{nthcdr}}.)
8832 @findex setcdr, @r{example}
8833 Thus, if we had a four element list that was supposed to be three
8834 elements long, we could set the @sc{cdr} of the next to last element
8835 to @code{nil}, and thereby shorten the list. (If you set the last
8836 element to some other value than @code{nil}, which you could do, then
8837 you would not have shortened the list. @xref{setcdr, ,
8840 You can see shortening by evaluating the following three expressions
8841 in turn. First set the value of @code{trees} to @code{(maple oak pine
8842 birch)}, then set the @sc{cdr} of its second @sc{cdr} to @code{nil}
8843 and then find the value of @code{trees}:
8847 (setq trees '(maple oak pine birch))
8848 @result{} (maple oak pine birch)
8852 (setcdr (nthcdr 2 trees) nil)
8856 @result{} (maple oak pine)
8861 (The value returned by the @code{setcdr} expression is @code{nil} since
8862 that is what the @sc{cdr} is set to.)
8864 To repeat, in @code{kill-new}, the @code{nthcdr} function takes the
8865 @sc{cdr} a number of times that is one less than the maximum permitted
8866 size of the kill ring and @code{setcdr} sets the @sc{cdr} of that
8867 element (which will be the rest of the elements in the kill ring) to
8868 @code{nil}. This prevents the kill ring from growing too long.
8871 The next to last expression in the @code{kill-new} function is
8874 (setq kill-ring-yank-pointer kill-ring)
8877 The @code{kill-ring-yank-pointer} is a global variable that is set to be
8878 the @code{kill-ring}.
8880 Even though the @code{kill-ring-yank-pointer} is called a
8881 @samp{pointer}, it is a variable just like the kill ring. However, the
8882 name has been chosen to help humans understand how the variable is used.
8885 Now, to return to an early expression in the body of the function:
8889 (if (fboundp 'menu-bar-update-yank-menu)
8890 (menu-bar-update-yank-menu string (and replace (car kill-ring))))
8895 It starts with an @code{if} expression
8897 In this case, the expression tests first to see whether
8898 @code{menu-bar-update-yank-menu} exists as a function, and if so,
8899 calls it. The @code{fboundp} function returns true if the symbol it
8900 is testing has a function definition that `is not void'. If the
8901 symbol's function definition were void, we would receive an error
8902 message, as we did when we created errors intentionally (@pxref{Making
8903 Errors, , Generate an Error Message}).
8906 The then-part contains an expression whose first element is the
8907 function @code{and}.
8910 The @code{and} special form evaluates each of its arguments until one
8911 of the arguments returns a value of @code{nil}, in which case the
8912 @code{and} expression returns @code{nil}; however, if none of the
8913 arguments returns a value of @code{nil}, the value resulting from
8914 evaluating the last argument is returned. (Since such a value is not
8915 @code{nil}, it is considered true in Emacs Lisp.) In other words, an
8916 @code{and} expression returns a true value only if all its arguments
8917 are true. (@xref{Second Buffer Related Review}.)
8919 The expression determines whether the second argument to
8920 @code{menu-bar-update-yank-menu} is true or not.
8922 ;; If we're supposed to be extending an existing string, and that
8923 ;; string really is at the front of the menu, then update it in place.
8926 @code{menu-bar-update-yank-menu} is one of the functions that make it
8927 possible to use the `Select and Paste' menu in the Edit item of a menu
8928 bar; using a mouse, you can look at the various pieces of text you
8929 have saved and select one piece to paste.
8931 The last expression in the @code{kill-new} function adds the newly
8932 copied string to whatever facility exists for copying and pasting
8933 among different programs running in a windowing system. In the X
8934 Windowing system, for example, the @code{x-select-text} function takes
8935 the string and stores it in memory operated by X@. You can paste the
8936 string in another program, such as an Xterm.
8939 The expression looks like this:
8943 (if interprogram-cut-function
8944 (funcall interprogram-cut-function string (not replace))))
8948 If an @code{interprogram-cut-function} exists, then Emacs executes
8949 @code{funcall}, which in turn calls its first argument as a function
8950 and passes the remaining arguments to it. (Incidentally, as far as I
8951 can see, this @code{if} expression could be replaced by an @code{and}
8952 expression similar to the one in the first part of the function.)
8954 We are not going to discuss windowing systems and other programs
8955 further, but merely note that this is a mechanism that enables GNU
8956 Emacs to work easily and well with other programs.
8958 This code for placing text in the kill ring, either concatenated with
8959 an existing element or as a new element, leads us to the code for
8960 bringing back text that has been cut out of the buffer---the yank
8961 commands. However, before discussing the yank commands, it is better
8962 to learn how lists are implemented in a computer. This will make
8963 clear such mysteries as the use of the term `pointer'. But before
8964 that, we will digress into C.
8967 @c is this true in Emacs 22? Does not seems to be
8969 (If the @w{@code{(< end beg))}}
8970 expression is true, @code{kill-append} prepends the string to the just
8971 previously clipped text. For a detailed discussion, see
8972 @ref{kill-append function, , The @code{kill-append} function}.)
8974 If you then yank back the text, i.e., `paste' it, you get both
8975 pieces of text at once. That way, if you delete two words in a row,
8976 and then yank them back, you get both words, in their proper order,
8977 with one yank. (The @w{@code{(< end beg))}} expression makes sure the
8980 On the other hand, if the previous command is not @code{kill-region},
8981 then the @code{kill-new} function is called, which adds the text to
8982 the kill ring as the latest item, and sets the
8983 @code{kill-ring-yank-pointer} variable to point to it.
8987 @c Evidently, changed for Emacs 22. The zap-to-char command does not
8988 @c use the delete-and-extract-region function
8990 2006 Oct 26, the Digression into C is now OK but should come after
8991 copy-region-as-kill and filter-buffer-substring
8995 copy-region-as-kill is short, 12 lines, and uses
8996 filter-buffer-substring, which is longer, 39 lines
8997 and has delete-and-extract-region in it.
8998 delete-and-extract-region is written in C.
9000 see Initializing a Variable with @code{defvar}
9003 @node Digression into C
9004 @section Digression into C
9005 @findex delete-and-extract-region
9006 @cindex C, a digression into
9007 @cindex Digression into C
9009 The @code{copy-region-as-kill} function (@pxref{copy-region-as-kill, ,
9010 @code{copy-region-as-kill}}) uses the @code{filter-buffer-substring}
9011 function, which in turn uses the @code{delete-and-extract-region}
9012 function. It removes the contents of a region and you cannot get them
9015 Unlike the other code discussed here, the
9016 @code{delete-and-extract-region} function is not written in Emacs
9017 Lisp; it is written in C and is one of the primitives of the GNU Emacs
9018 system. Since it is very simple, I will digress briefly from Lisp and
9021 @c GNU Emacs 24 in src/editfns.c
9022 @c the DEFUN for delete-and-extract-region
9025 Like many of the other Emacs primitives,
9026 @code{delete-and-extract-region} is written as an instance of a C
9027 macro, a macro being a template for code. The complete macro looks
9032 DEFUN ("delete-and-extract-region", Fdelete_and_extract_region,
9033 Sdelete_and_extract_region, 2, 2, 0,
9034 doc: /* Delete the text between START and END and return it. */)
9035 (Lisp_Object start, Lisp_Object end)
9037 validate_region (&start, &end);
9038 if (XINT (start) == XINT (end))
9039 return empty_unibyte_string;
9040 return del_range_1 (XINT (start), XINT (end), 1, 1);
9045 Without going into the details of the macro writing process, let me
9046 point out that this macro starts with the word @code{DEFUN}. The word
9047 @code{DEFUN} was chosen since the code serves the same purpose as
9048 @code{defun} does in Lisp. (The @code{DEFUN} C macro is defined in
9049 @file{emacs/src/lisp.h}.)
9051 The word @code{DEFUN} is followed by seven parts inside of
9056 The first part is the name given to the function in Lisp,
9057 @code{delete-and-extract-region}.
9060 The second part is the name of the function in C,
9061 @code{Fdelete_and_extract_region}. By convention, it starts with
9062 @samp{F}. Since C does not use hyphens in names, underscores are used
9066 The third part is the name for the C constant structure that records
9067 information on this function for internal use. It is the name of the
9068 function in C but begins with an @samp{S} instead of an @samp{F}.
9071 The fourth and fifth parts specify the minimum and maximum number of
9072 arguments the function can have. This function demands exactly 2
9076 The sixth part is nearly like the argument that follows the
9077 @code{interactive} declaration in a function written in Lisp: a letter
9078 followed, perhaps, by a prompt. The only difference from the Lisp is
9079 when the macro is called with no arguments. Then you write a @code{0}
9080 (which is a `null string'), as in this macro.
9082 If you were to specify arguments, you would place them between
9083 quotation marks. The C macro for @code{goto-char} includes
9084 @code{"NGoto char: "} in this position to indicate that the function
9085 expects a raw prefix, in this case, a numerical location in a buffer,
9086 and provides a prompt.
9089 The seventh part is a documentation string, just like the one for a
9090 function written in Emacs Lisp. This is written as a C comment. (When
9091 you build Emacs, the program @command{lib-src/make-docfile} extracts
9092 these comments and uses them to make the ``real'' documentation.)
9096 In a C macro, the formal parameters come next, with a statement of
9097 what kind of object they are, followed by what might be called the `body'
9098 of the macro. For @code{delete-and-extract-region} the `body'
9099 consists of the following four lines:
9103 validate_region (&start, &end);
9104 if (XINT (start) == XINT (end))
9105 return empty_unibyte_string;
9106 return del_range_1 (XINT (start), XINT (end), 1, 1);
9110 The @code{validate_region} function checks whether the values
9111 passed as the beginning and end of the region are the proper type and
9112 are within range. If the beginning and end positions are the same,
9113 then return an empty string.
9115 The @code{del_range_1} function actually deletes the text. It is a
9116 complex function we will not look into. It updates the buffer and
9117 does other things. However, it is worth looking at the two arguments
9118 passed to @code{del_range}. These are @w{@code{XINT (start)}} and
9119 @w{@code{XINT (end)}}.
9121 As far as the C language is concerned, @code{start} and @code{end} are
9122 two integers that mark the beginning and end of the region to be
9123 deleted@footnote{More precisely, and requiring more expert knowledge
9124 to understand, the two integers are of type `Lisp_Object', which can
9125 also be a C union instead of an integer type.}.
9127 In early versions of Emacs, these two numbers were thirty-two bits
9128 long, but the code is slowly being generalized to handle other
9129 lengths. Three of the available bits are used to specify the type of
9130 information; the remaining bits are used as `content'.
9132 @samp{XINT} is a C macro that extracts the relevant number from the
9133 longer collection of bits; the three other bits are discarded.
9136 The command in @code{delete-and-extract-region} looks like this:
9139 del_range_1 (XINT (start), XINT (end), 1, 1);
9143 It deletes the region between the beginning position, @code{start},
9144 and the ending position, @code{end}.
9146 From the point of view of the person writing Lisp, Emacs is all very
9147 simple; but hidden underneath is a great deal of complexity to make it
9151 @section Initializing a Variable with @code{defvar}
9153 @cindex Initializing a variable
9154 @cindex Variable initialization
9159 copy-region-as-kill is short, 12 lines, and uses
9160 filter-buffer-substring, which is longer, 39 lines
9161 and has delete-and-extract-region in it.
9162 delete-and-extract-region is written in C.
9164 see Initializing a Variable with @code{defvar}
9168 The @code{copy-region-as-kill} function is written in Emacs Lisp. Two
9169 functions within it, @code{kill-append} and @code{kill-new}, copy a
9170 region in a buffer and save it in a variable called the
9171 @code{kill-ring}. This section describes how the @code{kill-ring}
9172 variable is created and initialized using the @code{defvar} special
9175 (Again we note that the term @code{kill-ring} is a misnomer. The text
9176 that is clipped out of the buffer can be brought back; it is not a ring
9177 of corpses, but a ring of resurrectable text.)
9179 In Emacs Lisp, a variable such as the @code{kill-ring} is created and
9180 given an initial value by using the @code{defvar} special form. The
9181 name comes from ``define variable''.
9183 The @code{defvar} special form is similar to @code{setq} in that it sets
9184 the value of a variable. It is unlike @code{setq} in two ways: first,
9185 it only sets the value of the variable if the variable does not already
9186 have a value. If the variable already has a value, @code{defvar} does
9187 not override the existing value. Second, @code{defvar} has a
9188 documentation string.
9190 (There is a related macro, @code{defcustom}, designed for variables
9191 that people customize. It has more features than @code{defvar}.
9192 (@xref{defcustom, , Setting Variables with @code{defcustom}}.)
9195 * See variable current value::
9196 * defvar and asterisk::
9200 @node See variable current value
9201 @unnumberedsubsec Seeing the Current Value of a Variable
9204 You can see the current value of a variable, any variable, by using
9205 the @code{describe-variable} function, which is usually invoked by
9206 typing @kbd{C-h v}. If you type @kbd{C-h v} and then @code{kill-ring}
9207 (followed by @key{RET}) when prompted, you will see what is in your
9208 current kill ring---this may be quite a lot! Conversely, if you have
9209 been doing nothing this Emacs session except read this document, you
9210 may have nothing in it. Also, you will see the documentation for
9216 List of killed text sequences.
9217 Since the kill ring is supposed to interact nicely with cut-and-paste
9218 facilities offered by window systems, use of this variable should
9221 interact nicely with `interprogram-cut-function' and
9222 `interprogram-paste-function'. The functions `kill-new',
9223 `kill-append', and `current-kill' are supposed to implement this
9224 interaction; you may want to use them instead of manipulating the kill
9230 The kill ring is defined by a @code{defvar} in the following way:
9234 (defvar kill-ring nil
9235 "List of killed text sequences.
9241 In this variable definition, the variable is given an initial value of
9242 @code{nil}, which makes sense, since if you have saved nothing, you want
9243 nothing back if you give a @code{yank} command. The documentation
9244 string is written just like the documentation string of a @code{defun}.
9245 As with the documentation string of the @code{defun}, the first line of
9246 the documentation should be a complete sentence, since some commands,
9247 like @code{apropos}, print only the first line of documentation.
9248 Succeeding lines should not be indented; otherwise they look odd when
9249 you use @kbd{C-h v} (@code{describe-variable}).
9251 @node defvar and asterisk
9252 @subsection @code{defvar} and an asterisk
9253 @findex defvar @r{for a user customizable variable}
9254 @findex defvar @r{with an asterisk}
9256 In the past, Emacs used the @code{defvar} special form both for
9257 internal variables that you would not expect a user to change and for
9258 variables that you do expect a user to change. Although you can still
9259 use @code{defvar} for user customizable variables, please use
9260 @code{defcustom} instead, since it provides a path into
9261 the Customization commands. (@xref{defcustom, , Specifying Variables
9262 using @code{defcustom}}.)
9264 When you specified a variable using the @code{defvar} special form,
9265 you could distinguish a variable that a user might want to change from
9266 others by typing an asterisk, @samp{*}, in the first column of its
9267 documentation string. For example:
9271 (defvar shell-command-default-error-buffer nil
9272 "*Buffer name for `shell-command' @dots{} error output.
9277 @findex set-variable
9279 You could (and still can) use the @code{set-variable} command to
9280 change the value of @code{shell-command-default-error-buffer}
9281 temporarily. However, options set using @code{set-variable} are set
9282 only for the duration of your editing session. The new values are not
9283 saved between sessions. Each time Emacs starts, it reads the original
9284 value, unless you change the value within your @file{.emacs} file,
9285 either by setting it manually or by using @code{customize}.
9286 @xref{Emacs Initialization, , Your @file{.emacs} File}.
9288 For me, the major use of the @code{set-variable} command is to suggest
9289 variables that I might want to set in my @file{.emacs} file. There
9290 are now more than 700 such variables, far too many to remember
9291 readily. Fortunately, you can press @key{TAB} after calling the
9292 @code{M-x set-variable} command to see the list of variables.
9293 (@xref{Examining, , Examining and Setting Variables, emacs,
9294 The GNU Emacs Manual}.)
9297 @node cons & search-fwd Review
9300 Here is a brief summary of some recently introduced functions.
9305 @code{car} returns the first element of a list; @code{cdr} returns the
9306 second and subsequent elements of a list.
9313 (car '(1 2 3 4 5 6 7))
9315 (cdr '(1 2 3 4 5 6 7))
9316 @result{} (2 3 4 5 6 7)
9321 @code{cons} constructs a list by prepending its first argument to its
9335 @code{funcall} evaluates its first argument as a function. It passes
9336 its remaining arguments to its first argument.
9339 Return the result of taking @sc{cdr} `n' times on a list.
9347 The `rest of the rest', as it were.
9354 (nthcdr 3 '(1 2 3 4 5 6 7))
9361 @code{setcar} changes the first element of a list; @code{setcdr}
9362 changes the second and subsequent elements of a list.
9369 (setq triple '(1 2 3))
9376 (setcdr triple '("foo" "bar"))
9379 @result{} (37 "foo" "bar")
9384 Evaluate each argument in sequence and then return the value of the
9397 @item save-restriction
9398 Record whatever narrowing is in effect in the current buffer, if any,
9399 and restore that narrowing after evaluating the arguments.
9401 @item search-forward
9402 Search for a string, and if the string is found, move point. With a
9403 regular expression, use the similar @code{re-search-forward}.
9404 (@xref{Regexp Search, , Regular Expression Searches}, for an
9405 explanation of regular expression patterns and searches.)
9409 @code{search-forward} and @code{re-search-forward} take four
9414 The string or regular expression to search for.
9417 Optionally, the limit of the search.
9420 Optionally, what to do if the search fails, return @code{nil} or an
9424 Optionally, how many times to repeat the search; if negative, the
9425 search goes backwards.
9429 @itemx delete-and-extract-region
9430 @itemx copy-region-as-kill
9432 @code{kill-region} cuts the text between point and mark from the
9433 buffer and stores that text in the kill ring, so you can get it back
9436 @code{copy-region-as-kill} copies the text between point and mark into
9437 the kill ring, from which you can get it by yanking. The function
9438 does not cut or remove the text from the buffer.
9441 @code{delete-and-extract-region} removes the text between point and
9442 mark from the buffer and throws it away. You cannot get it back.
9443 (This is not an interactive command.)
9446 @node search Exercises
9447 @section Searching Exercises
9451 Write an interactive function that searches for a string. If the
9452 search finds the string, leave point after it and display a message
9453 that says ``Found!''. (Do not use @code{search-forward} for the name
9454 of this function; if you do, you will overwrite the existing version of
9455 @code{search-forward} that comes with Emacs. Use a name such as
9456 @code{test-search} instead.)
9459 Write a function that prints the third element of the kill ring in the
9460 echo area, if any; if the kill ring does not contain a third element,
9461 print an appropriate message.
9464 @node List Implementation
9465 @chapter How Lists are Implemented
9466 @cindex Lists in a computer
9468 In Lisp, atoms are recorded in a straightforward fashion; if the
9469 implementation is not straightforward in practice, it is, nonetheless,
9470 straightforward in theory. The atom @samp{rose}, for example, is
9471 recorded as the four contiguous letters @samp{r}, @samp{o}, @samp{s},
9472 @samp{e}. A list, on the other hand, is kept differently. The mechanism
9473 is equally simple, but it takes a moment to get used to the idea. A
9474 list is kept using a series of pairs of pointers. In the series, the
9475 first pointer in each pair points to an atom or to another list, and the
9476 second pointer in each pair points to the next pair, or to the symbol
9477 @code{nil}, which marks the end of the list.
9479 A pointer itself is quite simply the electronic address of what is
9480 pointed to. Hence, a list is kept as a series of electronic addresses.
9483 * Lists diagrammed::
9484 * Symbols as Chest:: Exploring a powerful metaphor.
9489 @node Lists diagrammed
9490 @unnumberedsec Lists diagrammed
9493 For example, the list @code{(rose violet buttercup)} has three elements,
9494 @samp{rose}, @samp{violet}, and @samp{buttercup}. In the computer, the
9495 electronic address of @samp{rose} is recorded in a segment of computer
9496 memory along with the address that gives the electronic address of where
9497 the atom @samp{violet} is located; and that address (the one that tells
9498 where @samp{violet} is located) is kept along with an address that tells
9499 where the address for the atom @samp{buttercup} is located.
9502 This sounds more complicated than it is and is easier seen in a diagram:
9504 @c clear print-postscript-figures
9505 @c !!! cons-cell-diagram #1
9509 ___ ___ ___ ___ ___ ___
9510 |___|___|--> |___|___|--> |___|___|--> nil
9513 --> rose --> violet --> buttercup
9517 @ifset print-postscript-figures
9520 @center @image{cons-1}
9524 @ifclear print-postscript-figures
9528 ___ ___ ___ ___ ___ ___
9529 |___|___|--> |___|___|--> |___|___|--> nil
9532 --> rose --> violet --> buttercup
9539 In the diagram, each box represents a word of computer memory that
9540 holds a Lisp object, usually in the form of a memory address. The boxes,
9541 i.e., the addresses, are in pairs. Each arrow points to what the address
9542 is the address of, either an atom or another pair of addresses. The
9543 first box is the electronic address of @samp{rose} and the arrow points
9544 to @samp{rose}; the second box is the address of the next pair of boxes,
9545 the first part of which is the address of @samp{violet} and the second
9546 part of which is the address of the next pair. The very last box
9547 points to the symbol @code{nil}, which marks the end of the list.
9550 When a variable is set to a list with a function such as @code{setq},
9551 it stores the address of the first box in the variable. Thus,
9552 evaluation of the expression
9555 (setq bouquet '(rose violet buttercup))
9560 creates a situation like this:
9562 @c cons-cell-diagram #2
9568 | ___ ___ ___ ___ ___ ___
9569 --> |___|___|--> |___|___|--> |___|___|--> nil
9572 --> rose --> violet --> buttercup
9576 @ifset print-postscript-figures
9579 @center @image{cons-2}
9583 @ifclear print-postscript-figures
9589 | ___ ___ ___ ___ ___ ___
9590 --> |___|___|--> |___|___|--> |___|___|--> nil
9593 --> rose --> violet --> buttercup
9600 In this example, the symbol @code{bouquet} holds the address of the first
9604 This same list can be illustrated in a different sort of box notation
9607 @c cons-cell-diagram #2a
9613 | -------------- --------------- ----------------
9614 | | car | cdr | | car | cdr | | car | cdr |
9615 -->| rose | o------->| violet | o------->| butter- | nil |
9616 | | | | | | | cup | |
9617 -------------- --------------- ----------------
9621 @ifset print-postscript-figures
9624 @center @image{cons-2a}
9628 @ifclear print-postscript-figures
9634 | -------------- --------------- ----------------
9635 | | car | cdr | | car | cdr | | car | cdr |
9636 -->| rose | o------->| violet | o------->| butter- | nil |
9637 | | | | | | | cup | |
9638 -------------- --------------- ----------------
9644 (Symbols consist of more than pairs of addresses, but the structure of
9645 a symbol is made up of addresses. Indeed, the symbol @code{bouquet}
9646 consists of a group of address-boxes, one of which is the address of
9647 the printed word @samp{bouquet}, a second of which is the address of a
9648 function definition attached to the symbol, if any, a third of which
9649 is the address of the first pair of address-boxes for the list
9650 @code{(rose violet buttercup)}, and so on. Here we are showing that
9651 the symbol's third address-box points to the first pair of
9652 address-boxes for the list.)
9654 If a symbol is set to the @sc{cdr} of a list, the list itself is not
9655 changed; the symbol simply has an address further down the list. (In
9656 the jargon, @sc{car} and @sc{cdr} are `non-destructive'.) Thus,
9657 evaluation of the following expression
9660 (setq flowers (cdr bouquet))
9667 @c cons-cell-diagram #3
9674 | ___ ___ | ___ ___ ___ ___
9675 --> | | | --> | | | | | |
9676 |___|___|----> |___|___|--> |___|___|--> nil
9679 --> rose --> violet --> buttercup
9684 @ifset print-postscript-figures
9687 @center @image{cons-3}
9691 @ifclear print-postscript-figures
9698 | ___ ___ | ___ ___ ___ ___
9699 --> | | | --> | | | | | |
9700 |___|___|----> |___|___|--> |___|___|--> nil
9703 --> rose --> violet --> buttercup
9711 The value of @code{flowers} is @code{(violet buttercup)}, which is
9712 to say, the symbol @code{flowers} holds the address of the pair of
9713 address-boxes, the first of which holds the address of @code{violet},
9714 and the second of which holds the address of @code{buttercup}.
9716 A pair of address-boxes is called a @dfn{cons cell} or @dfn{dotted
9717 pair}. @xref{Cons Cell Type, , Cons Cell and List Types, elisp, The GNU Emacs Lisp
9718 Reference Manual}, and @ref{Dotted Pair Notation, , Dotted Pair
9719 Notation, elisp, The GNU Emacs Lisp Reference Manual}, for more
9720 information about cons cells and dotted pairs.
9723 The function @code{cons} adds a new pair of addresses to the front of
9724 a series of addresses like that shown above. For example, evaluating
9728 (setq bouquet (cons 'lily bouquet))
9735 @c cons-cell-diagram #4
9742 | ___ ___ ___ ___ | ___ ___ ___ ___
9743 --> | | | | | | --> | | | | | |
9744 |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil
9747 --> lily --> rose --> violet --> buttercup
9752 @ifset print-postscript-figures
9755 @center @image{cons-4}
9759 @ifclear print-postscript-figures
9766 | ___ ___ ___ ___ | ___ ___ ___ ___
9767 --> | | | | | | --> | | | | | |
9768 |___|___|----> |___|___|----> |___|___|---->|___|___|--> nil
9771 --> lily --> rose --> violet --> buttercup
9780 However, this does not change the value of the symbol
9781 @code{flowers}, as you can see by evaluating the following,
9784 (eq (cdr (cdr bouquet)) flowers)
9788 which returns @code{t} for true.
9790 Until it is reset, @code{flowers} still has the value
9791 @code{(violet buttercup)}; that is, it has the address of the cons
9792 cell whose first address is of @code{violet}. Also, this does not
9793 alter any of the pre-existing cons cells; they are all still there.
9795 Thus, in Lisp, to get the @sc{cdr} of a list, you just get the address
9796 of the next cons cell in the series; to get the @sc{car} of a list,
9797 you get the address of the first element of the list; to @code{cons} a
9798 new element on a list, you add a new cons cell to the front of the list.
9799 That is all there is to it! The underlying structure of Lisp is
9802 And what does the last address in a series of cons cells refer to? It
9803 is the address of the empty list, of @code{nil}.
9805 In summary, when a Lisp variable is set to a value, it is provided with
9806 the address of the list to which the variable refers.
9808 @node Symbols as Chest
9809 @section Symbols as a Chest of Drawers
9810 @cindex Symbols as a Chest of Drawers
9811 @cindex Chest of Drawers, metaphor for a symbol
9812 @cindex Drawers, Chest of, metaphor for a symbol
9814 In an earlier section, I suggested that you might imagine a symbol as
9815 being a chest of drawers. The function definition is put in one
9816 drawer, the value in another, and so on. What is put in the drawer
9817 holding the value can be changed without affecting the contents of the
9818 drawer holding the function definition, and vice-verse.
9820 Actually, what is put in each drawer is the address of the value or
9821 function definition. It is as if you found an old chest in the attic,
9822 and in one of its drawers you found a map giving you directions to
9823 where the buried treasure lies.
9825 (In addition to its name, symbol definition, and variable value, a
9826 symbol has a `drawer' for a @dfn{property list} which can be used to
9827 record other information. Property lists are not discussed here; see
9828 @ref{Property Lists, , Property Lists, elisp, The GNU Emacs Lisp
9832 Here is a fanciful representation:
9834 @c chest-of-drawers diagram
9839 Chest of Drawers Contents of Drawers
9843 ---------------------
9844 | directions to | [map to]
9845 | symbol name | bouquet
9847 +---------------------+
9849 | symbol definition | [none]
9851 +---------------------+
9852 | directions to | [map to]
9853 | variable value | (rose violet buttercup)
9855 +---------------------+
9857 | property list | [not described here]
9859 +---------------------+
9865 @ifset print-postscript-figures
9868 @center @image{drawers}
9872 @ifclear print-postscript-figures
9877 Chest of Drawers Contents of Drawers
9881 ---------------------
9882 | directions to | [map to]
9883 | symbol name | bouquet
9885 +---------------------+
9887 | symbol definition | [none]
9889 +---------------------+
9890 | directions to | [map to]
9891 | variable value | (rose violet buttercup)
9893 +---------------------+
9895 | property list | [not described here]
9897 +---------------------+
9908 Set @code{flowers} to @code{violet} and @code{buttercup}. Cons two
9909 more flowers on to this list and set this new list to
9910 @code{more-flowers}. Set the @sc{car} of @code{flowers} to a fish.
9911 What does the @code{more-flowers} list now contain?
9914 @chapter Yanking Text Back
9916 @cindex Text retrieval
9917 @cindex Retrieving text
9918 @cindex Pasting text
9920 Whenever you cut text out of a buffer with a `kill' command in GNU Emacs,
9921 you can bring it back with a `yank' command. The text that is cut out of
9922 the buffer is put in the kill ring and the yank commands insert the
9923 appropriate contents of the kill ring back into a buffer (not necessarily
9924 the original buffer).
9926 A simple @kbd{C-y} (@code{yank}) command inserts the first item from
9927 the kill ring into the current buffer. If the @kbd{C-y} command is
9928 followed immediately by @kbd{M-y}, the first element is replaced by
9929 the second element. Successive @kbd{M-y} commands replace the second
9930 element with the third, fourth, or fifth element, and so on. When the
9931 last element in the kill ring is reached, it is replaced by the first
9932 element and the cycle is repeated. (Thus the kill ring is called a
9933 `ring' rather than just a `list'. However, the actual data structure
9934 that holds the text is a list.
9935 @xref{Kill Ring, , Handling the Kill Ring}, for the details of how the
9936 list is handled as a ring.)
9939 * Kill Ring Overview::
9940 * kill-ring-yank-pointer:: The kill ring is a list.
9941 * yank nthcdr Exercises:: The @code{kill-ring-yank-pointer} variable.
9944 @node Kill Ring Overview
9945 @section Kill Ring Overview
9946 @cindex Kill ring overview
9948 The kill ring is a list of textual strings. This is what it looks like:
9951 ("some text" "a different piece of text" "yet more text")
9954 If this were the contents of my kill ring and I pressed @kbd{C-y}, the
9955 string of characters saying @samp{some text} would be inserted in this
9956 buffer where my cursor is located.
9958 The @code{yank} command is also used for duplicating text by copying it.
9959 The copied text is not cut from the buffer, but a copy of it is put on the
9960 kill ring and is inserted by yanking it back.
9962 Three functions are used for bringing text back from the kill ring:
9963 @code{yank}, which is usually bound to @kbd{C-y}; @code{yank-pop},
9964 which is usually bound to @kbd{M-y}; and @code{rotate-yank-pointer},
9965 which is used by the two other functions.
9967 These functions refer to the kill ring through a variable called the
9968 @code{kill-ring-yank-pointer}. Indeed, the insertion code for both the
9969 @code{yank} and @code{yank-pop} functions is:
9972 (insert (car kill-ring-yank-pointer))
9976 (Well, no more. In GNU Emacs 22, the function has been replaced by
9977 @code{insert-for-yank} which calls @code{insert-for-yank-1}
9978 repetitively for each @code{yank-handler} segment. In turn,
9979 @code{insert-for-yank-1} strips text properties from the inserted text
9980 according to @code{yank-excluded-properties}. Otherwise, it is just
9981 like @code{insert}. We will stick with plain @code{insert} since it
9982 is easier to understand.)
9984 To begin to understand how @code{yank} and @code{yank-pop} work, it is
9985 first necessary to look at the @code{kill-ring-yank-pointer} variable.
9987 @node kill-ring-yank-pointer
9988 @section The @code{kill-ring-yank-pointer} Variable
9990 @code{kill-ring-yank-pointer} is a variable, just as @code{kill-ring} is
9991 a variable. It points to something by being bound to the value of what
9992 it points to, like any other Lisp variable.
9995 Thus, if the value of the kill ring is:
9998 ("some text" "a different piece of text" "yet more text")
10003 and the @code{kill-ring-yank-pointer} points to the second clause, the
10004 value of @code{kill-ring-yank-pointer} is:
10007 ("a different piece of text" "yet more text")
10010 As explained in the previous chapter (@pxref{List Implementation}), the
10011 computer does not keep two different copies of the text being pointed to
10012 by both the @code{kill-ring} and the @code{kill-ring-yank-pointer}. The
10013 words ``a different piece of text'' and ``yet more text'' are not
10014 duplicated. Instead, the two Lisp variables point to the same pieces of
10015 text. Here is a diagram:
10017 @c cons-cell-diagram #5
10021 kill-ring kill-ring-yank-pointer
10023 | ___ ___ | ___ ___ ___ ___
10024 ---> | | | --> | | | | | |
10025 |___|___|----> |___|___|--> |___|___|--> nil
10028 | | --> "yet more text"
10030 | --> "a different piece of text"
10037 @ifset print-postscript-figures
10040 @center @image{cons-5}
10044 @ifclear print-postscript-figures
10048 kill-ring kill-ring-yank-pointer
10050 | ___ ___ | ___ ___ ___ ___
10051 ---> | | | --> | | | | | |
10052 |___|___|----> |___|___|--> |___|___|--> nil
10055 | | --> "yet more text"
10057 | --> "a different piece of text
10066 Both the variable @code{kill-ring} and the variable
10067 @code{kill-ring-yank-pointer} are pointers. But the kill ring itself is
10068 usually described as if it were actually what it is composed of. The
10069 @code{kill-ring} is spoken of as if it were the list rather than that it
10070 points to the list. Conversely, the @code{kill-ring-yank-pointer} is
10071 spoken of as pointing to a list.
10073 These two ways of talking about the same thing sound confusing at first but
10074 make sense on reflection. The kill ring is generally thought of as the
10075 complete structure of data that holds the information of what has recently
10076 been cut out of the Emacs buffers. The @code{kill-ring-yank-pointer}
10077 on the other hand, serves to indicate---that is, to `point to'---that part
10078 of the kill ring of which the first element (the @sc{car}) will be
10082 In GNU Emacs 22, the @code{kill-new} function calls
10084 @code{(setq kill-ring-yank-pointer kill-ring)}
10086 (defun rotate-yank-pointer (arg)
10087 "Rotate the yanking point in the kill ring.
10088 With argument, rotate that many kills forward (or backward, if negative)."
10090 (current-kill arg))
10092 (defun current-kill (n &optional do-not-move)
10093 "Rotate the yanking point by N places, and then return that kill.
10094 If N is zero, `interprogram-paste-function' is set, and calling it
10095 returns a string, then that string is added to the front of the
10096 kill ring and returned as the latest kill.
10097 If optional arg DO-NOT-MOVE is non-nil, then don't actually move the
10098 yanking point; just return the Nth kill forward."
10099 (let ((interprogram-paste (and (= n 0)
10100 interprogram-paste-function
10101 (funcall interprogram-paste-function))))
10102 (if interprogram-paste
10104 ;; Disable the interprogram cut function when we add the new
10105 ;; text to the kill ring, so Emacs doesn't try to own the
10106 ;; selection, with identical text.
10107 (let ((interprogram-cut-function nil))
10108 (kill-new interprogram-paste))
10109 interprogram-paste)
10110 (or kill-ring (error "Kill ring is empty"))
10111 (let ((ARGth-kill-element
10112 (nthcdr (mod (- n (length kill-ring-yank-pointer))
10113 (length kill-ring))
10116 (setq kill-ring-yank-pointer ARGth-kill-element))
10117 (car ARGth-kill-element)))))
10122 @node yank nthcdr Exercises
10123 @section Exercises with @code{yank} and @code{nthcdr}
10127 Using @kbd{C-h v} (@code{describe-variable}), look at the value of
10128 your kill ring. Add several items to your kill ring; look at its
10129 value again. Using @kbd{M-y} (@code{yank-pop)}, move all the way
10130 around the kill ring. How many items were in your kill ring? Find
10131 the value of @code{kill-ring-max}. Was your kill ring full, or could
10132 you have kept more blocks of text within it?
10135 Using @code{nthcdr} and @code{car}, construct a series of expressions
10136 to return the first, second, third, and fourth elements of a list.
10139 @node Loops & Recursion
10140 @chapter Loops and Recursion
10141 @cindex Loops and recursion
10142 @cindex Recursion and loops
10143 @cindex Repetition (loops)
10145 Emacs Lisp has two primary ways to cause an expression, or a series of
10146 expressions, to be evaluated repeatedly: one uses a @code{while}
10147 loop, and the other uses @dfn{recursion}.
10149 Repetition can be very valuable. For example, to move forward four
10150 sentences, you need only write a program that will move forward one
10151 sentence and then repeat the process four times. Since a computer does
10152 not get bored or tired, such repetitive action does not have the
10153 deleterious effects that excessive or the wrong kinds of repetition can
10156 People mostly write Emacs Lisp functions using @code{while} loops and
10157 their kin; but you can use recursion, which provides a very powerful
10158 way to think about and then to solve problems@footnote{You can write
10159 recursive functions to be frugal or wasteful of mental or computer
10160 resources; as it happens, methods that people find easy---that are
10161 frugal of `mental resources'---sometimes use considerable computer
10162 resources. Emacs was designed to run on machines that we now consider
10163 limited and its default settings are conservative. You may want to
10164 increase the values of @code{max-specpdl-size} and
10165 @code{max-lisp-eval-depth}. In my @file{.emacs} file, I set them to
10166 15 and 30 times their default value.}.
10169 * while:: Causing a stretch of code to repeat.
10171 * Recursion:: Causing a function to call itself.
10172 * Looping exercise::
10176 @section @code{while}
10180 The @code{while} special form tests whether the value returned by
10181 evaluating its first argument is true or false. This is similar to what
10182 the Lisp interpreter does with an @code{if}; what the interpreter does
10183 next, however, is different.
10185 In a @code{while} expression, if the value returned by evaluating the
10186 first argument is false, the Lisp interpreter skips the rest of the
10187 expression (the @dfn{body} of the expression) and does not evaluate it.
10188 However, if the value is true, the Lisp interpreter evaluates the body
10189 of the expression and then again tests whether the first argument to
10190 @code{while} is true or false. If the value returned by evaluating the
10191 first argument is again true, the Lisp interpreter again evaluates the
10192 body of the expression.
10195 The template for a @code{while} expression looks like this:
10199 (while @var{true-or-false-test}
10205 * Looping with while:: Repeat so long as test returns true.
10206 * Loop Example:: A @code{while} loop that uses a list.
10207 * print-elements-of-list:: Uses @code{while}, @code{car}, @code{cdr}.
10208 * Incrementing Loop:: A loop with an incrementing counter.
10209 * Incrementing Loop Details::
10210 * Decrementing Loop:: A loop with a decrementing counter.
10214 @node Looping with while
10215 @unnumberedsubsec Looping with @code{while}
10218 So long as the true-or-false-test of the @code{while} expression
10219 returns a true value when it is evaluated, the body is repeatedly
10220 evaluated. This process is called a loop since the Lisp interpreter
10221 repeats the same thing again and again, like an airplane doing a loop.
10222 When the result of evaluating the true-or-false-test is false, the
10223 Lisp interpreter does not evaluate the rest of the @code{while}
10224 expression and `exits the loop'.
10226 Clearly, if the value returned by evaluating the first argument to
10227 @code{while} is always true, the body following will be evaluated
10228 again and again @dots{} and again @dots{} forever. Conversely, if the
10229 value returned is never true, the expressions in the body will never
10230 be evaluated. The craft of writing a @code{while} loop consists of
10231 choosing a mechanism such that the true-or-false-test returns true
10232 just the number of times that you want the subsequent expressions to
10233 be evaluated, and then have the test return false.
10235 The value returned by evaluating a @code{while} is the value of the
10236 true-or-false-test. An interesting consequence of this is that a
10237 @code{while} loop that evaluates without error will return @code{nil}
10238 or false regardless of whether it has looped 1 or 100 times or none at
10239 all. A @code{while} expression that evaluates successfully never
10240 returns a true value! What this means is that @code{while} is always
10241 evaluated for its side effects, which is to say, the consequences of
10242 evaluating the expressions within the body of the @code{while} loop.
10243 This makes sense. It is not the mere act of looping that is desired,
10244 but the consequences of what happens when the expressions in the loop
10245 are repeatedly evaluated.
10248 @subsection A @code{while} Loop and a List
10250 A common way to control a @code{while} loop is to test whether a list
10251 has any elements. If it does, the loop is repeated; but if it does not,
10252 the repetition is ended. Since this is an important technique, we will
10253 create a short example to illustrate it.
10255 A simple way to test whether a list has elements is to evaluate the
10256 list: if it has no elements, it is an empty list and will return the
10257 empty list, @code{()}, which is a synonym for @code{nil} or false. On
10258 the other hand, a list with elements will return those elements when it
10259 is evaluated. Since Emacs Lisp considers as true any value that is not
10260 @code{nil}, a list that returns elements will test true in a
10264 For example, you can set the variable @code{empty-list} to @code{nil} by
10265 evaluating the following @code{setq} expression:
10268 (setq empty-list ())
10272 After evaluating the @code{setq} expression, you can evaluate the
10273 variable @code{empty-list} in the usual way, by placing the cursor after
10274 the symbol and typing @kbd{C-x C-e}; @code{nil} will appear in your
10281 On the other hand, if you set a variable to be a list with elements, the
10282 list will appear when you evaluate the variable, as you can see by
10283 evaluating the following two expressions:
10287 (setq animals '(gazelle giraffe lion tiger))
10293 Thus, to create a @code{while} loop that tests whether there are any
10294 items in the list @code{animals}, the first part of the loop will be
10305 When the @code{while} tests its first argument, the variable
10306 @code{animals} is evaluated. It returns a list. So long as the list
10307 has elements, the @code{while} considers the results of the test to be
10308 true; but when the list is empty, it considers the results of the test
10311 To prevent the @code{while} loop from running forever, some mechanism
10312 needs to be provided to empty the list eventually. An oft-used
10313 technique is to have one of the subsequent forms in the @code{while}
10314 expression set the value of the list to be the @sc{cdr} of the list.
10315 Each time the @code{cdr} function is evaluated, the list will be made
10316 shorter, until eventually only the empty list will be left. At this
10317 point, the test of the @code{while} loop will return false, and the
10318 arguments to the @code{while} will no longer be evaluated.
10320 For example, the list of animals bound to the variable @code{animals}
10321 can be set to be the @sc{cdr} of the original list with the
10322 following expression:
10325 (setq animals (cdr animals))
10329 If you have evaluated the previous expressions and then evaluate this
10330 expression, you will see @code{(giraffe lion tiger)} appear in the echo
10331 area. If you evaluate the expression again, @code{(lion tiger)} will
10332 appear in the echo area. If you evaluate it again and yet again,
10333 @code{(tiger)} appears and then the empty list, shown by @code{nil}.
10335 A template for a @code{while} loop that uses the @code{cdr} function
10336 repeatedly to cause the true-or-false-test eventually to test false
10341 (while @var{test-whether-list-is-empty}
10343 @var{set-list-to-cdr-of-list})
10347 This test and use of @code{cdr} can be put together in a function that
10348 goes through a list and prints each element of the list on a line of its
10351 @node print-elements-of-list
10352 @subsection An Example: @code{print-elements-of-list}
10353 @findex print-elements-of-list
10355 The @code{print-elements-of-list} function illustrates a @code{while}
10358 @cindex @file{*scratch*} buffer
10359 The function requires several lines for its output. If you are
10360 reading this in a recent instance of GNU Emacs,
10361 @c GNU Emacs 21, GNU Emacs 22, or a later version,
10362 you can evaluate the following expression inside of Info, as usual.
10364 If you are using an earlier version of Emacs, you need to copy the
10365 necessary expressions to your @file{*scratch*} buffer and evaluate
10366 them there. This is because the echo area had only one line in the
10369 You can copy the expressions by marking the beginning of the region
10370 with @kbd{C-@key{SPC}} (@code{set-mark-command}), moving the cursor to
10371 the end of the region and then copying the region using @kbd{M-w}
10372 (@code{kill-ring-save}, which calls @code{copy-region-as-kill} and
10373 then provides visual feedback). In the @file{*scratch*}
10374 buffer, you can yank the expressions back by typing @kbd{C-y}
10377 After you have copied the expressions to the @file{*scratch*} buffer,
10378 evaluate each expression in turn. Be sure to evaluate the last
10379 expression, @code{(print-elements-of-list animals)}, by typing
10380 @kbd{C-u C-x C-e}, that is, by giving an argument to
10381 @code{eval-last-sexp}. This will cause the result of the evaluation
10382 to be printed in the @file{*scratch*} buffer instead of being printed
10383 in the echo area. (Otherwise you will see something like this in your
10384 echo area: @code{^Jgazelle^J^Jgiraffe^J^Jlion^J^Jtiger^Jnil}, in which
10385 each @samp{^J} stands for a `newline'.)
10388 In a recent instance of GNU Emacs, you can evaluate these expressions
10389 directly in the Info buffer, and the echo area will grow to show the
10394 (setq animals '(gazelle giraffe lion tiger))
10396 (defun print-elements-of-list (list)
10397 "Print each element of LIST on a line of its own."
10400 (setq list (cdr list))))
10402 (print-elements-of-list animals)
10408 When you evaluate the three expressions in sequence, you will see
10424 Each element of the list is printed on a line of its own (that is what
10425 the function @code{print} does) and then the value returned by the
10426 function is printed. Since the last expression in the function is the
10427 @code{while} loop, and since @code{while} loops always return
10428 @code{nil}, a @code{nil} is printed after the last element of the list.
10430 @node Incrementing Loop
10431 @subsection A Loop with an Incrementing Counter
10433 A loop is not useful unless it stops when it ought. Besides
10434 controlling a loop with a list, a common way of stopping a loop is to
10435 write the first argument as a test that returns false when the correct
10436 number of repetitions are complete. This means that the loop must
10437 have a counter---an expression that counts how many times the loop
10441 @node Incrementing Loop Details
10442 @unnumberedsubsec Details of an Incrementing Loop
10445 The test for a loop with an incrementing counter can be an expression
10446 such as @code{(< count desired-number)} which returns @code{t} for
10447 true if the value of @code{count} is less than the
10448 @code{desired-number} of repetitions and @code{nil} for false if the
10449 value of @code{count} is equal to or is greater than the
10450 @code{desired-number}. The expression that increments the count can
10451 be a simple @code{setq} such as @code{(setq count (1+ count))}, where
10452 @code{1+} is a built-in function in Emacs Lisp that adds 1 to its
10453 argument. (The expression @w{@code{(1+ count)}} has the same result
10454 as @w{@code{(+ count 1)}}, but is easier for a human to read.)
10457 The template for a @code{while} loop controlled by an incrementing
10458 counter looks like this:
10462 @var{set-count-to-initial-value}
10463 (while (< count desired-number) ; @r{true-or-false-test}
10465 (setq count (1+ count))) ; @r{incrementer}
10470 Note that you need to set the initial value of @code{count}; usually it
10474 * Incrementing Example:: Counting pebbles in a triangle.
10475 * Inc Example parts:: The parts of the function definition.
10476 * Inc Example altogether:: Putting the function definition together.
10479 @node Incrementing Example
10480 @unnumberedsubsubsec Example with incrementing counter
10482 Suppose you are playing on the beach and decide to make a triangle of
10483 pebbles, putting one pebble in the first row, two in the second row,
10484 three in the third row and so on, like this:
10502 @bullet{} @bullet{}
10503 @bullet{} @bullet{} @bullet{}
10504 @bullet{} @bullet{} @bullet{} @bullet{}
10511 (About 2500 years ago, Pythagoras and others developed the beginnings of
10512 number theory by considering questions such as this.)
10514 Suppose you want to know how many pebbles you will need to make a
10515 triangle with 7 rows?
10517 Clearly, what you need to do is add up the numbers from 1 to 7. There
10518 are two ways to do this; start with the smallest number, one, and add up
10519 the list in sequence, 1, 2, 3, 4 and so on; or start with the largest
10520 number and add the list going down: 7, 6, 5, 4 and so on. Because both
10521 mechanisms illustrate common ways of writing @code{while} loops, we will
10522 create two examples, one counting up and the other counting down. In
10523 this first example, we will start with 1 and add 2, 3, 4 and so on.
10525 If you are just adding up a short list of numbers, the easiest way to do
10526 it is to add up all the numbers at once. However, if you do not know
10527 ahead of time how many numbers your list will have, or if you want to be
10528 prepared for a very long list, then you need to design your addition so
10529 that what you do is repeat a simple process many times instead of doing
10530 a more complex process once.
10532 For example, instead of adding up all the pebbles all at once, what you
10533 can do is add the number of pebbles in the first row, 1, to the number
10534 in the second row, 2, and then add the total of those two rows to the
10535 third row, 3. Then you can add the number in the fourth row, 4, to the
10536 total of the first three rows; and so on.
10538 The critical characteristic of the process is that each repetitive
10539 action is simple. In this case, at each step we add only two numbers,
10540 the number of pebbles in the row and the total already found. This
10541 process of adding two numbers is repeated again and again until the last
10542 row has been added to the total of all the preceding rows. In a more
10543 complex loop the repetitive action might not be so simple, but it will
10544 be simpler than doing everything all at once.
10546 @node Inc Example parts
10547 @unnumberedsubsubsec The parts of the function definition
10549 The preceding analysis gives us the bones of our function definition:
10550 first, we will need a variable that we can call @code{total} that will
10551 be the total number of pebbles. This will be the value returned by
10554 Second, we know that the function will require an argument: this
10555 argument will be the total number of rows in the triangle. It can be
10556 called @code{number-of-rows}.
10558 Finally, we need a variable to use as a counter. We could call this
10559 variable @code{counter}, but a better name is @code{row-number}. That
10560 is because what the counter does in this function is count rows, and a
10561 program should be written to be as understandable as possible.
10563 When the Lisp interpreter first starts evaluating the expressions in the
10564 function, the value of @code{total} should be set to zero, since we have
10565 not added anything to it. Then the function should add the number of
10566 pebbles in the first row to the total, and then add the number of
10567 pebbles in the second to the total, and then add the number of
10568 pebbles in the third row to the total, and so on, until there are no
10569 more rows left to add.
10571 Both @code{total} and @code{row-number} are used only inside the
10572 function, so they can be declared as local variables with @code{let}
10573 and given initial values. Clearly, the initial value for @code{total}
10574 should be 0. The initial value of @code{row-number} should be 1,
10575 since we start with the first row. This means that the @code{let}
10576 statement will look like this:
10586 After the internal variables are declared and bound to their initial
10587 values, we can begin the @code{while} loop. The expression that serves
10588 as the test should return a value of @code{t} for true so long as the
10589 @code{row-number} is less than or equal to the @code{number-of-rows}.
10590 (If the expression tests true only so long as the row number is less
10591 than the number of rows in the triangle, the last row will never be
10592 added to the total; hence the row number has to be either less than or
10593 equal to the number of rows.)
10596 @findex <= @r{(less than or equal)}
10597 Lisp provides the @code{<=} function that returns true if the value of
10598 its first argument is less than or equal to the value of its second
10599 argument and false otherwise. So the expression that the @code{while}
10600 will evaluate as its test should look like this:
10603 (<= row-number number-of-rows)
10606 The total number of pebbles can be found by repeatedly adding the number
10607 of pebbles in a row to the total already found. Since the number of
10608 pebbles in the row is equal to the row number, the total can be found by
10609 adding the row number to the total. (Clearly, in a more complex
10610 situation, the number of pebbles in the row might be related to the row
10611 number in a more complicated way; if this were the case, the row number
10612 would be replaced by the appropriate expression.)
10615 (setq total (+ total row-number))
10619 What this does is set the new value of @code{total} to be equal to the
10620 sum of adding the number of pebbles in the row to the previous total.
10622 After setting the value of @code{total}, the conditions need to be
10623 established for the next repetition of the loop, if there is one. This
10624 is done by incrementing the value of the @code{row-number} variable,
10625 which serves as a counter. After the @code{row-number} variable has
10626 been incremented, the true-or-false-test at the beginning of the
10627 @code{while} loop tests whether its value is still less than or equal to
10628 the value of the @code{number-of-rows} and if it is, adds the new value
10629 of the @code{row-number} variable to the @code{total} of the previous
10630 repetition of the loop.
10633 The built-in Emacs Lisp function @code{1+} adds 1 to a number, so the
10634 @code{row-number} variable can be incremented with this expression:
10637 (setq row-number (1+ row-number))
10640 @node Inc Example altogether
10641 @unnumberedsubsubsec Putting the function definition together
10643 We have created the parts for the function definition; now we need to
10647 First, the contents of the @code{while} expression:
10651 (while (<= row-number number-of-rows) ; @r{true-or-false-test}
10652 (setq total (+ total row-number))
10653 (setq row-number (1+ row-number))) ; @r{incrementer}
10657 Along with the @code{let} expression varlist, this very nearly
10658 completes the body of the function definition. However, it requires
10659 one final element, the need for which is somewhat subtle.
10661 The final touch is to place the variable @code{total} on a line by
10662 itself after the @code{while} expression. Otherwise, the value returned
10663 by the whole function is the value of the last expression that is
10664 evaluated in the body of the @code{let}, and this is the value
10665 returned by the @code{while}, which is always @code{nil}.
10667 This may not be evident at first sight. It almost looks as if the
10668 incrementing expression is the last expression of the whole function.
10669 But that expression is part of the body of the @code{while}; it is the
10670 last element of the list that starts with the symbol @code{while}.
10671 Moreover, the whole of the @code{while} loop is a list within the body
10675 In outline, the function will look like this:
10679 (defun @var{name-of-function} (@var{argument-list})
10680 "@var{documentation}@dots{}"
10681 (let (@var{varlist})
10682 (while (@var{true-or-false-test})
10683 @var{body-of-while}@dots{} )
10684 @dots{} )) ; @r{Need final expression here.}
10688 The result of evaluating the @code{let} is what is going to be returned
10689 by the @code{defun} since the @code{let} is not embedded within any
10690 containing list, except for the @code{defun} as a whole. However, if
10691 the @code{while} is the last element of the @code{let} expression, the
10692 function will always return @code{nil}. This is not what we want!
10693 Instead, what we want is the value of the variable @code{total}. This
10694 is returned by simply placing the symbol as the last element of the list
10695 starting with @code{let}. It gets evaluated after the preceding
10696 elements of the list are evaluated, which means it gets evaluated after
10697 it has been assigned the correct value for the total.
10699 It may be easier to see this by printing the list starting with
10700 @code{let} all on one line. This format makes it evident that the
10701 @var{varlist} and @code{while} expressions are the second and third
10702 elements of the list starting with @code{let}, and the @code{total} is
10707 (let (@var{varlist}) (while (@var{true-or-false-test}) @var{body-of-while}@dots{} ) total)
10712 Putting everything together, the @code{triangle} function definition
10717 (defun triangle (number-of-rows) ; @r{Version with}
10718 ; @r{ incrementing counter.}
10719 "Add up the number of pebbles in a triangle.
10720 The first row has one pebble, the second row two pebbles,
10721 the third row three pebbles, and so on.
10722 The argument is NUMBER-OF-ROWS."
10727 (while (<= row-number number-of-rows)
10728 (setq total (+ total row-number))
10729 (setq row-number (1+ row-number)))
10735 After you have installed @code{triangle} by evaluating the function, you
10736 can try it out. Here are two examples:
10747 The sum of the first four numbers is 10 and the sum of the first seven
10750 @node Decrementing Loop
10751 @subsection Loop with a Decrementing Counter
10753 Another common way to write a @code{while} loop is to write the test
10754 so that it determines whether a counter is greater than zero. So long
10755 as the counter is greater than zero, the loop is repeated. But when
10756 the counter is equal to or less than zero, the loop is stopped. For
10757 this to work, the counter has to start out greater than zero and then
10758 be made smaller and smaller by a form that is evaluated
10761 The test will be an expression such as @code{(> counter 0)} which
10762 returns @code{t} for true if the value of @code{counter} is greater
10763 than zero, and @code{nil} for false if the value of @code{counter} is
10764 equal to or less than zero. The expression that makes the number
10765 smaller and smaller can be a simple @code{setq} such as @code{(setq
10766 counter (1- counter))}, where @code{1-} is a built-in function in
10767 Emacs Lisp that subtracts 1 from its argument.
10770 The template for a decrementing @code{while} loop looks like this:
10774 (while (> counter 0) ; @r{true-or-false-test}
10776 (setq counter (1- counter))) ; @r{decrementer}
10781 * Decrementing Example:: More pebbles on the beach.
10782 * Dec Example parts:: The parts of the function definition.
10783 * Dec Example altogether:: Putting the function definition together.
10786 @node Decrementing Example
10787 @unnumberedsubsubsec Example with decrementing counter
10789 To illustrate a loop with a decrementing counter, we will rewrite the
10790 @code{triangle} function so the counter decreases to zero.
10792 This is the reverse of the earlier version of the function. In this
10793 case, to find out how many pebbles are needed to make a triangle with
10794 3 rows, add the number of pebbles in the third row, 3, to the number
10795 in the preceding row, 2, and then add the total of those two rows to
10796 the row that precedes them, which is 1.
10798 Likewise, to find the number of pebbles in a triangle with 7 rows, add
10799 the number of pebbles in the seventh row, 7, to the number in the
10800 preceding row, which is 6, and then add the total of those two rows to
10801 the row that precedes them, which is 5, and so on. As in the previous
10802 example, each addition only involves adding two numbers, the total of
10803 the rows already added up and the number of pebbles in the row that is
10804 being added to the total. This process of adding two numbers is
10805 repeated again and again until there are no more pebbles to add.
10807 We know how many pebbles to start with: the number of pebbles in the
10808 last row is equal to the number of rows. If the triangle has seven
10809 rows, the number of pebbles in the last row is 7. Likewise, we know how
10810 many pebbles are in the preceding row: it is one less than the number in
10813 @node Dec Example parts
10814 @unnumberedsubsubsec The parts of the function definition
10816 We start with three variables: the total number of rows in the
10817 triangle; the number of pebbles in a row; and the total number of
10818 pebbles, which is what we want to calculate. These variables can be
10819 named @code{number-of-rows}, @code{number-of-pebbles-in-row}, and
10820 @code{total}, respectively.
10822 Both @code{total} and @code{number-of-pebbles-in-row} are used only
10823 inside the function and are declared with @code{let}. The initial
10824 value of @code{total} should, of course, be zero. However, the
10825 initial value of @code{number-of-pebbles-in-row} should be equal to
10826 the number of rows in the triangle, since the addition will start with
10830 This means that the beginning of the @code{let} expression will look
10836 (number-of-pebbles-in-row number-of-rows))
10841 The total number of pebbles can be found by repeatedly adding the number
10842 of pebbles in a row to the total already found, that is, by repeatedly
10843 evaluating the following expression:
10846 (setq total (+ total number-of-pebbles-in-row))
10850 After the @code{number-of-pebbles-in-row} is added to the @code{total},
10851 the @code{number-of-pebbles-in-row} should be decremented by one, since
10852 the next time the loop repeats, the preceding row will be
10853 added to the total.
10855 The number of pebbles in a preceding row is one less than the number of
10856 pebbles in a row, so the built-in Emacs Lisp function @code{1-} can be
10857 used to compute the number of pebbles in the preceding row. This can be
10858 done with the following expression:
10862 (setq number-of-pebbles-in-row
10863 (1- number-of-pebbles-in-row))
10867 Finally, we know that the @code{while} loop should stop making repeated
10868 additions when there are no pebbles in a row. So the test for
10869 the @code{while} loop is simply:
10872 (while (> number-of-pebbles-in-row 0)
10875 @node Dec Example altogether
10876 @unnumberedsubsubsec Putting the function definition together
10878 We can put these expressions together to create a function definition
10879 that works. However, on examination, we find that one of the local
10880 variables is unneeded!
10883 The function definition looks like this:
10887 ;;; @r{First subtractive version.}
10888 (defun triangle (number-of-rows)
10889 "Add up the number of pebbles in a triangle."
10891 (number-of-pebbles-in-row number-of-rows))
10892 (while (> number-of-pebbles-in-row 0)
10893 (setq total (+ total number-of-pebbles-in-row))
10894 (setq number-of-pebbles-in-row
10895 (1- number-of-pebbles-in-row)))
10900 As written, this function works.
10902 However, we do not need @code{number-of-pebbles-in-row}.
10904 @cindex Argument as local variable
10905 When the @code{triangle} function is evaluated, the symbol
10906 @code{number-of-rows} will be bound to a number, giving it an initial
10907 value. That number can be changed in the body of the function as if
10908 it were a local variable, without any fear that such a change will
10909 effect the value of the variable outside of the function. This is a
10910 very useful characteristic of Lisp; it means that the variable
10911 @code{number-of-rows} can be used anywhere in the function where
10912 @code{number-of-pebbles-in-row} is used.
10915 Here is a second version of the function written a bit more cleanly:
10919 (defun triangle (number) ; @r{Second version.}
10920 "Return sum of numbers 1 through NUMBER inclusive."
10922 (while (> number 0)
10923 (setq total (+ total number))
10924 (setq number (1- number)))
10929 In brief, a properly written @code{while} loop will consist of three parts:
10933 A test that will return false after the loop has repeated itself the
10934 correct number of times.
10937 An expression the evaluation of which will return the value desired
10938 after being repeatedly evaluated.
10941 An expression to change the value passed to the true-or-false-test so
10942 that the test returns false after the loop has repeated itself the right
10946 @node dolist dotimes
10947 @section Save your time: @code{dolist} and @code{dotimes}
10949 In addition to @code{while}, both @code{dolist} and @code{dotimes}
10950 provide for looping. Sometimes these are quicker to write than the
10951 equivalent @code{while} loop. Both are Lisp macros. (@xref{Macros, ,
10952 Macros, elisp, The GNU Emacs Lisp Reference Manual}. )
10954 @code{dolist} works like a @code{while} loop that `@sc{cdr}s down a
10955 list': @code{dolist} automatically shortens the list each time it
10956 loops---takes the @sc{cdr} of the list---and binds the @sc{car} of
10957 each shorter version of the list to the first of its arguments.
10959 @code{dotimes} loops a specific number of times: you specify the number.
10967 @unnumberedsubsec The @code{dolist} Macro
10970 Suppose, for example, you want to reverse a list, so that
10971 ``first'' ``second'' ``third'' becomes ``third'' ``second'' ``first''.
10974 In practice, you would use the @code{reverse} function, like this:
10978 (setq animals '(gazelle giraffe lion tiger))
10986 Here is how you could reverse the list using a @code{while} loop:
10990 (setq animals '(gazelle giraffe lion tiger))
10992 (defun reverse-list-with-while (list)
10993 "Using while, reverse the order of LIST."
10994 (let (value) ; make sure list starts empty
10996 (setq value (cons (car list) value))
10997 (setq list (cdr list)))
11000 (reverse-list-with-while animals)
11006 And here is how you could use the @code{dolist} macro:
11010 (setq animals '(gazelle giraffe lion tiger))
11012 (defun reverse-list-with-dolist (list)
11013 "Using dolist, reverse the order of LIST."
11014 (let (value) ; make sure list starts empty
11015 (dolist (element list value)
11016 (setq value (cons element value)))))
11018 (reverse-list-with-dolist animals)
11024 In Info, you can place your cursor after the closing parenthesis of
11025 each expression and type @kbd{C-x C-e}; in each case, you should see
11028 (tiger lion giraffe gazelle)
11034 For this example, the existing @code{reverse} function is obviously best.
11035 The @code{while} loop is just like our first example (@pxref{Loop
11036 Example, , A @code{while} Loop and a List}). The @code{while} first
11037 checks whether the list has elements; if so, it constructs a new list
11038 by adding the first element of the list to the existing list (which in
11039 the first iteration of the loop is @code{nil}). Since the second
11040 element is prepended in front of the first element, and the third
11041 element is prepended in front of the second element, the list is reversed.
11043 In the expression using a @code{while} loop,
11044 the @w{@code{(setq list (cdr list))}}
11045 expression shortens the list, so the @code{while} loop eventually
11046 stops. In addition, it provides the @code{cons} expression with a new
11047 first element by creating a new and shorter list at each repetition of
11050 The @code{dolist} expression does very much the same as the
11051 @code{while} expression, except that the @code{dolist} macro does some
11052 of the work you have to do when writing a @code{while} expression.
11054 Like a @code{while} loop, a @code{dolist} loops. What is different is
11055 that it automatically shortens the list each time it loops---it
11056 `@sc{cdr}s down the list' on its own---and it automatically binds
11057 the @sc{car} of each shorter version of the list to the first of its
11060 In the example, the @sc{car} of each shorter version of the list is
11061 referred to using the symbol @samp{element}, the list itself is called
11062 @samp{list}, and the value returned is called @samp{value}. The
11063 remainder of the @code{dolist} expression is the body.
11065 The @code{dolist} expression binds the @sc{car} of each shorter
11066 version of the list to @code{element} and then evaluates the body of
11067 the expression; and repeats the loop. The result is returned in
11071 @unnumberedsubsec The @code{dotimes} Macro
11074 The @code{dotimes} macro is similar to @code{dolist}, except that it
11075 loops a specific number of times.
11077 The first argument to @code{dotimes} is assigned the numbers 0, 1, 2
11078 and so forth each time around the loop, and the value of the third
11079 argument is returned. You need to provide the value of the second
11080 argument, which is how many times the macro loops.
11083 For example, the following binds the numbers from 0 up to, but not
11084 including, the number 3 to the first argument, @var{number}, and then
11085 constructs a list of the three numbers. (The first number is 0, the
11086 second number is 1, and the third number is 2; this makes a total of
11087 three numbers in all, starting with zero as the first number.)
11091 (let (value) ; otherwise a value is a void variable
11092 (dotimes (number 3 value)
11093 (setq value (cons number value))))
11100 @code{dotimes} returns @code{value}, so the way to use
11101 @code{dotimes} is to operate on some expression @var{number} number of
11102 times and then return the result, either as a list or an atom.
11105 Here is an example of a @code{defun} that uses @code{dotimes} to add
11106 up the number of pebbles in a triangle.
11110 (defun triangle-using-dotimes (number-of-rows)
11111 "Using dotimes, add up the number of pebbles in a triangle."
11112 (let ((total 0)) ; otherwise a total is a void variable
11113 (dotimes (number number-of-rows total)
11114 (setq total (+ total (1+ number))))))
11116 (triangle-using-dotimes 4)
11124 A recursive function contains code that tells the Lisp interpreter to
11125 call a program that runs exactly like itself, but with slightly
11126 different arguments. The code runs exactly the same because it has
11127 the same name. However, even though the program has the same name, it
11128 is not the same entity. It is different. In the jargon, it is a
11129 different `instance'.
11131 Eventually, if the program is written correctly, the `slightly
11132 different arguments' will become sufficiently different from the first
11133 arguments that the final instance will stop.
11136 * Building Robots:: Same model, different serial number ...
11137 * Recursive Definition Parts:: Walk until you stop ...
11138 * Recursion with list:: Using a list as the test whether to recurse.
11139 * Recursive triangle function::
11140 * Recursion with cond::
11141 * Recursive Patterns:: Often used templates.
11142 * No Deferment:: Don't store up work ...
11143 * No deferment solution::
11146 @node Building Robots
11147 @subsection Building Robots: Extending the Metaphor
11148 @cindex Building robots
11149 @cindex Robots, building
11151 It is sometimes helpful to think of a running program as a robot that
11152 does a job. In doing its job, a recursive function calls on a second
11153 robot to help it. The second robot is identical to the first in every
11154 way, except that the second robot helps the first and has been
11155 passed different arguments than the first.
11157 In a recursive function, the second robot may call a third; and the
11158 third may call a fourth, and so on. Each of these is a different
11159 entity; but all are clones.
11161 Since each robot has slightly different instructions---the arguments
11162 will differ from one robot to the next---the last robot should know
11165 Let's expand on the metaphor in which a computer program is a robot.
11167 A function definition provides the blueprints for a robot. When you
11168 install a function definition, that is, when you evaluate a
11169 @code{defun} macro, you install the necessary equipment to build
11170 robots. It is as if you were in a factory, setting up an assembly
11171 line. Robots with the same name are built according to the same
11172 blueprints. So they have, as it were, the same `model number', but a
11173 different `serial number'.
11175 We often say that a recursive function `calls itself'. What we mean
11176 is that the instructions in a recursive function cause the Lisp
11177 interpreter to run a different function that has the same name and
11178 does the same job as the first, but with different arguments.
11180 It is important that the arguments differ from one instance to the
11181 next; otherwise, the process will never stop.
11183 @node Recursive Definition Parts
11184 @subsection The Parts of a Recursive Definition
11185 @cindex Parts of a Recursive Definition
11186 @cindex Recursive Definition Parts
11188 A recursive function typically contains a conditional expression which
11193 A true-or-false-test that determines whether the function is called
11194 again, here called the @dfn{do-again-test}.
11197 The name of the function. When this name is called, a new instance of
11198 the function---a new robot, as it were---is created and told what to do.
11201 An expression that returns a different value each time the function is
11202 called, here called the @dfn{next-step-expression}. Consequently, the
11203 argument (or arguments) passed to the new instance of the function
11204 will be different from that passed to the previous instance. This
11205 causes the conditional expression, the @dfn{do-again-test}, to test
11206 false after the correct number of repetitions.
11209 Recursive functions can be much simpler than any other kind of
11210 function. Indeed, when people first start to use them, they often look
11211 so mysteriously simple as to be incomprehensible. Like riding a
11212 bicycle, reading a recursive function definition takes a certain knack
11213 which is hard at first but then seems simple.
11216 There are several different common recursive patterns. A very simple
11217 pattern looks like this:
11221 (defun @var{name-of-recursive-function} (@var{argument-list})
11222 "@var{documentation}@dots{}"
11223 (if @var{do-again-test}
11225 (@var{name-of-recursive-function}
11226 @var{next-step-expression})))
11230 Each time a recursive function is evaluated, a new instance of it is
11231 created and told what to do. The arguments tell the instance what to do.
11233 An argument is bound to the value of the next-step-expression. Each
11234 instance runs with a different value of the next-step-expression.
11236 The value in the next-step-expression is used in the do-again-test.
11238 The value returned by the next-step-expression is passed to the new
11239 instance of the function, which evaluates it (or some
11240 transmogrification of it) to determine whether to continue or stop.
11241 The next-step-expression is designed so that the do-again-test returns
11242 false when the function should no longer be repeated.
11244 The do-again-test is sometimes called the @dfn{stop condition},
11245 since it stops the repetitions when it tests false.
11247 @node Recursion with list
11248 @subsection Recursion with a List
11250 The example of a @code{while} loop that printed the elements of a list
11251 of numbers can be written recursively. Here is the code, including
11252 an expression to set the value of the variable @code{animals} to a list.
11254 If you are reading this in Info in Emacs, you can evaluate this
11255 expression directly in Info. Otherwise, you must copy the example
11256 to the @file{*scratch*} buffer and evaluate each expression there.
11257 Use @kbd{C-u C-x C-e} to evaluate the
11258 @code{(print-elements-recursively animals)} expression so that the
11259 results are printed in the buffer; otherwise the Lisp interpreter will
11260 try to squeeze the results into the one line of the echo area.
11262 Also, place your cursor immediately after the last closing parenthesis
11263 of the @code{print-elements-recursively} function, before the comment.
11264 Otherwise, the Lisp interpreter will try to evaluate the comment.
11266 @findex print-elements-recursively
11269 (setq animals '(gazelle giraffe lion tiger))
11271 (defun print-elements-recursively (list)
11272 "Print each element of LIST on a line of its own.
11274 (when list ; @r{do-again-test}
11275 (print (car list)) ; @r{body}
11276 (print-elements-recursively ; @r{recursive call}
11277 (cdr list)))) ; @r{next-step-expression}
11279 (print-elements-recursively animals)
11283 The @code{print-elements-recursively} function first tests whether
11284 there is any content in the list; if there is, the function prints the
11285 first element of the list, the @sc{car} of the list. Then the
11286 function `invokes itself', but gives itself as its argument, not the
11287 whole list, but the second and subsequent elements of the list, the
11288 @sc{cdr} of the list.
11290 Put another way, if the list is not empty, the function invokes
11291 another instance of code that is similar to the initial code, but is a
11292 different thread of execution, with different arguments than the first
11295 Put in yet another way, if the list is not empty, the first robot
11296 assembles a second robot and tells it what to do; the second robot is
11297 a different individual from the first, but is the same model.
11299 When the second evaluation occurs, the @code{when} expression is
11300 evaluated and if true, prints the first element of the list it
11301 receives as its argument (which is the second element of the original
11302 list). Then the function `calls itself' with the @sc{cdr} of the list
11303 it is invoked with, which (the second time around) is the @sc{cdr} of
11304 the @sc{cdr} of the original list.
11306 Note that although we say that the function `calls itself', what we
11307 mean is that the Lisp interpreter assembles and instructs a new
11308 instance of the program. The new instance is a clone of the first,
11309 but is a separate individual.
11311 Each time the function `invokes itself', it invokes itself on a
11312 shorter version of the original list. It creates a new instance that
11313 works on a shorter list.
11315 Eventually, the function invokes itself on an empty list. It creates
11316 a new instance whose argument is @code{nil}. The conditional expression
11317 tests the value of @code{list}. Since the value of @code{list} is
11318 @code{nil}, the @code{when} expression tests false so the then-part is
11319 not evaluated. The function as a whole then returns @code{nil}.
11322 When you evaluate the expression @code{(print-elements-recursively
11323 animals)} in the @file{*scratch*} buffer, you see this result:
11339 @node Recursive triangle function
11340 @subsection Recursion in Place of a Counter
11341 @findex triangle-recursively
11344 The @code{triangle} function described in a previous section can also
11345 be written recursively. It looks like this:
11349 (defun triangle-recursively (number)
11350 "Return the sum of the numbers 1 through NUMBER inclusive.
11352 (if (= number 1) ; @r{do-again-test}
11354 (+ number ; @r{else-part}
11355 (triangle-recursively ; @r{recursive call}
11356 (1- number))))) ; @r{next-step-expression}
11358 (triangle-recursively 7)
11363 You can install this function by evaluating it and then try it by
11364 evaluating @code{(triangle-recursively 7)}. (Remember to put your
11365 cursor immediately after the last parenthesis of the function
11366 definition, before the comment.) The function evaluates to 28.
11368 To understand how this function works, let's consider what happens in the
11369 various cases when the function is passed 1, 2, 3, or 4 as the value of
11373 * Recursive Example arg of 1 or 2::
11374 * Recursive Example arg of 3 or 4::
11378 @node Recursive Example arg of 1 or 2
11379 @unnumberedsubsubsec An argument of 1 or 2
11382 First, what happens if the value of the argument is 1?
11384 The function has an @code{if} expression after the documentation
11385 string. It tests whether the value of @code{number} is equal to 1; if
11386 so, Emacs evaluates the then-part of the @code{if} expression, which
11387 returns the number 1 as the value of the function. (A triangle with
11388 one row has one pebble in it.)
11390 Suppose, however, that the value of the argument is 2. In this case,
11391 Emacs evaluates the else-part of the @code{if} expression.
11394 The else-part consists of an addition, the recursive call to
11395 @code{triangle-recursively} and a decrementing action; and it looks like
11399 (+ number (triangle-recursively (1- number)))
11402 When Emacs evaluates this expression, the innermost expression is
11403 evaluated first; then the other parts in sequence. Here are the steps
11407 @item Step 1 @w{ } Evaluate the innermost expression.
11409 The innermost expression is @code{(1- number)} so Emacs decrements the
11410 value of @code{number} from 2 to 1.
11412 @item Step 2 @w{ } Evaluate the @code{triangle-recursively} function.
11414 The Lisp interpreter creates an individual instance of
11415 @code{triangle-recursively}. It does not matter that this function is
11416 contained within itself. Emacs passes the result Step 1 as the
11417 argument used by this instance of the @code{triangle-recursively}
11420 In this case, Emacs evaluates @code{triangle-recursively} with an
11421 argument of 1. This means that this evaluation of
11422 @code{triangle-recursively} returns 1.
11424 @item Step 3 @w{ } Evaluate the value of @code{number}.
11426 The variable @code{number} is the second element of the list that
11427 starts with @code{+}; its value is 2.
11429 @item Step 4 @w{ } Evaluate the @code{+} expression.
11431 The @code{+} expression receives two arguments, the first
11432 from the evaluation of @code{number} (Step 3) and the second from the
11433 evaluation of @code{triangle-recursively} (Step 2).
11435 The result of the addition is the sum of 2 plus 1, and the number 3 is
11436 returned, which is correct. A triangle with two rows has three
11440 @node Recursive Example arg of 3 or 4
11441 @unnumberedsubsubsec An argument of 3 or 4
11443 Suppose that @code{triangle-recursively} is called with an argument of
11447 @item Step 1 @w{ } Evaluate the do-again-test.
11449 The @code{if} expression is evaluated first. This is the do-again
11450 test and returns false, so the else-part of the @code{if} expression
11451 is evaluated. (Note that in this example, the do-again-test causes
11452 the function to call itself when it tests false, not when it tests
11455 @item Step 2 @w{ } Evaluate the innermost expression of the else-part.
11457 The innermost expression of the else-part is evaluated, which decrements
11458 3 to 2. This is the next-step-expression.
11460 @item Step 3 @w{ } Evaluate the @code{triangle-recursively} function.
11462 The number 2 is passed to the @code{triangle-recursively} function.
11464 We already know what happens when Emacs evaluates @code{triangle-recursively} with
11465 an argument of 2. After going through the sequence of actions described
11466 earlier, it returns a value of 3. So that is what will happen here.
11468 @item Step 4 @w{ } Evaluate the addition.
11470 3 will be passed as an argument to the addition and will be added to the
11471 number with which the function was called, which is 3.
11475 The value returned by the function as a whole will be 6.
11477 Now that we know what will happen when @code{triangle-recursively} is
11478 called with an argument of 3, it is evident what will happen if it is
11479 called with an argument of 4:
11483 In the recursive call, the evaluation of
11486 (triangle-recursively (1- 4))
11491 will return the value of evaluating
11494 (triangle-recursively 3)
11498 which is 6 and this value will be added to 4 by the addition in the
11503 The value returned by the function as a whole will be 10.
11505 Each time @code{triangle-recursively} is evaluated, it evaluates a
11506 version of itself---a different instance of itself---with a smaller
11507 argument, until the argument is small enough so that it does not
11510 Note that this particular design for a recursive function
11511 requires that operations be deferred.
11513 Before @code{(triangle-recursively 7)} can calculate its answer, it
11514 must call @code{(triangle-recursively 6)}; and before
11515 @code{(triangle-recursively 6)} can calculate its answer, it must call
11516 @code{(triangle-recursively 5)}; and so on. That is to say, the
11517 calculation that @code{(triangle-recursively 7)} makes must be
11518 deferred until @code{(triangle-recursively 6)} makes its calculation;
11519 and @code{(triangle-recursively 6)} must defer until
11520 @code{(triangle-recursively 5)} completes; and so on.
11522 If each of these instances of @code{triangle-recursively} are thought
11523 of as different robots, the first robot must wait for the second to
11524 complete its job, which must wait until the third completes, and so
11527 There is a way around this kind of waiting, which we will discuss in
11528 @ref{No Deferment, , Recursion without Deferments}.
11530 @node Recursion with cond
11531 @subsection Recursion Example Using @code{cond}
11534 The version of @code{triangle-recursively} described earlier is written
11535 with the @code{if} special form. It can also be written using another
11536 special form called @code{cond}. The name of the special form
11537 @code{cond} is an abbreviation of the word @samp{conditional}.
11539 Although the @code{cond} special form is not used as often in the
11540 Emacs Lisp sources as @code{if}, it is used often enough to justify
11544 The template for a @code{cond} expression looks like this:
11554 where the @var{body} is a series of lists.
11557 Written out more fully, the template looks like this:
11562 (@var{first-true-or-false-test} @var{first-consequent})
11563 (@var{second-true-or-false-test} @var{second-consequent})
11564 (@var{third-true-or-false-test} @var{third-consequent})
11569 When the Lisp interpreter evaluates the @code{cond} expression, it
11570 evaluates the first element (the @sc{car} or true-or-false-test) of
11571 the first expression in a series of expressions within the body of the
11574 If the true-or-false-test returns @code{nil} the rest of that
11575 expression, the consequent, is skipped and the true-or-false-test of the
11576 next expression is evaluated. When an expression is found whose
11577 true-or-false-test returns a value that is not @code{nil}, the
11578 consequent of that expression is evaluated. The consequent can be one
11579 or more expressions. If the consequent consists of more than one
11580 expression, the expressions are evaluated in sequence and the value of
11581 the last one is returned. If the expression does not have a consequent,
11582 the value of the true-or-false-test is returned.
11584 If none of the true-or-false-tests test true, the @code{cond} expression
11585 returns @code{nil}.
11588 Written using @code{cond}, the @code{triangle} function looks like this:
11592 (defun triangle-using-cond (number)
11593 (cond ((<= number 0) 0)
11596 (+ number (triangle-using-cond (1- number))))))
11601 In this example, the @code{cond} returns 0 if the number is less than or
11602 equal to 0, it returns 1 if the number is 1 and it evaluates @code{(+
11603 number (triangle-using-cond (1- number)))} if the number is greater than
11606 @node Recursive Patterns
11607 @subsection Recursive Patterns
11608 @cindex Recursive Patterns
11610 Here are three common recursive patterns. Each involves a list.
11611 Recursion does not need to involve lists, but Lisp is designed for lists
11612 and this provides a sense of its primal capabilities.
11621 @unnumberedsubsubsec Recursive Pattern: @emph{every}
11622 @cindex Every, type of recursive pattern
11623 @cindex Recursive pattern: every
11625 In the @code{every} recursive pattern, an action is performed on every
11629 The basic pattern is:
11633 If a list be empty, return @code{nil}.
11635 Else, act on the beginning of the list (the @sc{car} of the list)
11638 through a recursive call by the function on the rest (the
11639 @sc{cdr}) of the list,
11641 and, optionally, combine the acted-on element, using @code{cons},
11642 with the results of acting on the rest.
11651 (defun square-each (numbers-list)
11652 "Square each of a NUMBERS LIST, recursively."
11653 (if (not numbers-list) ; do-again-test
11656 (* (car numbers-list) (car numbers-list))
11657 (square-each (cdr numbers-list))))) ; next-step-expression
11661 (square-each '(1 2 3))
11668 If @code{numbers-list} is empty, do nothing. But if it has content,
11669 construct a list combining the square of the first number in the list
11670 with the result of the recursive call.
11672 (The example follows the pattern exactly: @code{nil} is returned if
11673 the numbers' list is empty. In practice, you would write the
11674 conditional so it carries out the action when the numbers' list is not
11677 The @code{print-elements-recursively} function (@pxref{Recursion with
11678 list, , Recursion with a List}) is another example of an @code{every}
11679 pattern, except in this case, rather than bring the results together
11680 using @code{cons}, we print each element of output.
11683 The @code{print-elements-recursively} function looks like this:
11687 (setq animals '(gazelle giraffe lion tiger))
11691 (defun print-elements-recursively (list)
11692 "Print each element of LIST on a line of its own.
11694 (when list ; @r{do-again-test}
11695 (print (car list)) ; @r{body}
11696 (print-elements-recursively ; @r{recursive call}
11697 (cdr list)))) ; @r{next-step-expression}
11699 (print-elements-recursively animals)
11704 The pattern for @code{print-elements-recursively} is:
11708 When the list is empty, do nothing.
11710 But when the list has at least one element,
11713 act on the beginning of the list (the @sc{car} of the list),
11715 and make a recursive call on the rest (the @sc{cdr}) of the list.
11720 @unnumberedsubsubsec Recursive Pattern: @emph{accumulate}
11721 @cindex Accumulate, type of recursive pattern
11722 @cindex Recursive pattern: accumulate
11724 Another recursive pattern is called the @code{accumulate} pattern. In
11725 the @code{accumulate} recursive pattern, an action is performed on
11726 every element of a list and the result of that action is accumulated
11727 with the results of performing the action on the other elements.
11729 This is very like the `every' pattern using @code{cons}, except that
11730 @code{cons} is not used, but some other combiner.
11737 If a list be empty, return zero or some other constant.
11739 Else, act on the beginning of the list (the @sc{car} of the list),
11742 and combine that acted-on element, using @code{+} or
11743 some other combining function, with
11745 a recursive call by the function on the rest (the @sc{cdr}) of the list.
11750 Here is an example:
11754 (defun add-elements (numbers-list)
11755 "Add the elements of NUMBERS-LIST together."
11756 (if (not numbers-list)
11758 (+ (car numbers-list) (add-elements (cdr numbers-list)))))
11762 (add-elements '(1 2 3 4))
11767 @xref{Files List, , Making a List of Files}, for an example of the
11768 accumulate pattern.
11771 @unnumberedsubsubsec Recursive Pattern: @emph{keep}
11772 @cindex Keep, type of recursive pattern
11773 @cindex Recursive pattern: keep
11775 A third recursive pattern is called the @code{keep} pattern.
11776 In the @code{keep} recursive pattern, each element of a list is tested;
11777 the element is acted on and the results are kept only if the element
11780 Again, this is very like the `every' pattern, except the element is
11781 skipped unless it meets a criterion.
11784 The pattern has three parts:
11788 If a list be empty, return @code{nil}.
11790 Else, if the beginning of the list (the @sc{car} of the list) passes
11794 act on that element and combine it, using @code{cons} with
11796 a recursive call by the function on the rest (the @sc{cdr}) of the list.
11799 Otherwise, if the beginning of the list (the @sc{car} of the list) fails
11803 skip on that element,
11805 and, recursively call the function on the rest (the @sc{cdr}) of the list.
11810 Here is an example that uses @code{cond}:
11814 (defun keep-three-letter-words (word-list)
11815 "Keep three letter words in WORD-LIST."
11817 ;; First do-again-test: stop-condition
11818 ((not word-list) nil)
11820 ;; Second do-again-test: when to act
11821 ((eq 3 (length (symbol-name (car word-list))))
11822 ;; combine acted-on element with recursive call on shorter list
11823 (cons (car word-list) (keep-three-letter-words (cdr word-list))))
11825 ;; Third do-again-test: when to skip element;
11826 ;; recursively call shorter list with next-step expression
11827 (t (keep-three-letter-words (cdr word-list)))))
11831 (keep-three-letter-words '(one two three four five six))
11832 @result{} (one two six)
11836 It goes without saying that you need not use @code{nil} as the test for
11837 when to stop; and you can, of course, combine these patterns.
11840 @subsection Recursion without Deferments
11841 @cindex Deferment in recursion
11842 @cindex Recursion without Deferments
11844 Let's consider again what happens with the @code{triangle-recursively}
11845 function. We will find that the intermediate calculations are
11846 deferred until all can be done.
11849 Here is the function definition:
11853 (defun triangle-recursively (number)
11854 "Return the sum of the numbers 1 through NUMBER inclusive.
11856 (if (= number 1) ; @r{do-again-test}
11858 (+ number ; @r{else-part}
11859 (triangle-recursively ; @r{recursive call}
11860 (1- number))))) ; @r{next-step-expression}
11864 What happens when we call this function with a argument of 7?
11866 The first instance of the @code{triangle-recursively} function adds
11867 the number 7 to the value returned by a second instance of
11868 @code{triangle-recursively}, an instance that has been passed an
11869 argument of 6. That is to say, the first calculation is:
11872 (+ 7 (triangle-recursively 6))
11876 The first instance of @code{triangle-recursively}---you may want to
11877 think of it as a little robot---cannot complete its job. It must hand
11878 off the calculation for @code{(triangle-recursively 6)} to a second
11879 instance of the program, to a second robot. This second individual is
11880 completely different from the first one; it is, in the jargon, a
11881 `different instantiation'. Or, put another way, it is a different
11882 robot. It is the same model as the first; it calculates triangle
11883 numbers recursively; but it has a different serial number.
11885 And what does @code{(triangle-recursively 6)} return? It returns the
11886 number 6 added to the value returned by evaluating
11887 @code{triangle-recursively} with an argument of 5. Using the robot
11888 metaphor, it asks yet another robot to help it.
11894 (+ 7 6 (triangle-recursively 5))
11898 And what happens next?
11901 (+ 7 6 5 (triangle-recursively 4))
11904 Each time @code{triangle-recursively} is called, except for the last
11905 time, it creates another instance of the program---another robot---and
11906 asks it to make a calculation.
11909 Eventually, the full addition is set up and performed:
11915 This design for the function defers the calculation of the first step
11916 until the second can be done, and defers that until the third can be
11917 done, and so on. Each deferment means the computer must remember what
11918 is being waited on. This is not a problem when there are only a few
11919 steps, as in this example. But it can be a problem when there are
11922 @node No deferment solution
11923 @subsection No Deferment Solution
11924 @cindex No deferment solution
11925 @cindex Defermentless solution
11926 @cindex Solution without deferment
11928 The solution to the problem of deferred operations is to write in a
11929 manner that does not defer operations@footnote{The phrase @dfn{tail
11930 recursive} is used to describe such a process, one that uses
11931 `constant space'.}. This requires
11932 writing to a different pattern, often one that involves writing two
11933 function definitions, an `initialization' function and a `helper'
11936 The `initialization' function sets up the job; the `helper' function
11940 Here are the two function definitions for adding up numbers. They are
11941 so simple, I find them hard to understand.
11945 (defun triangle-initialization (number)
11946 "Return the sum of the numbers 1 through NUMBER inclusive.
11947 This is the `initialization' component of a two function
11948 duo that uses recursion."
11949 (triangle-recursive-helper 0 0 number))
11955 (defun triangle-recursive-helper (sum counter number)
11956 "Return SUM, using COUNTER, through NUMBER inclusive.
11957 This is the `helper' component of a two function duo
11958 that uses recursion."
11959 (if (> counter number)
11961 (triangle-recursive-helper (+ sum counter) ; @r{sum}
11962 (1+ counter) ; @r{counter}
11963 number))) ; @r{number}
11968 Install both function definitions by evaluating them, then call
11969 @code{triangle-initialization} with 2 rows:
11973 (triangle-initialization 2)
11978 The `initialization' function calls the first instance of the `helper'
11979 function with three arguments: zero, zero, and a number which is the
11980 number of rows in the triangle.
11982 The first two arguments passed to the `helper' function are
11983 initialization values. These values are changed when
11984 @code{triangle-recursive-helper} invokes new instances.@footnote{The
11985 jargon is mildly confusing: @code{triangle-recursive-helper} uses a
11986 process that is iterative in a procedure that is recursive. The
11987 process is called iterative because the computer need only record the
11988 three values, @code{sum}, @code{counter}, and @code{number}; the
11989 procedure is recursive because the function `calls itself'. On the
11990 other hand, both the process and the procedure used by
11991 @code{triangle-recursively} are called recursive. The word
11992 `recursive' has different meanings in the two contexts.}
11994 Let's see what happens when we have a triangle that has one row. (This
11995 triangle will have one pebble in it!)
11998 @code{triangle-initialization} will call its helper with
11999 the arguments @w{@code{0 0 1}}. That function will run the conditional
12000 test whether @code{(> counter number)}:
12008 and find that the result is false, so it will invoke
12009 the else-part of the @code{if} clause:
12013 (triangle-recursive-helper
12014 (+ sum counter) ; @r{sum plus counter} @result{} @r{sum}
12015 (1+ counter) ; @r{increment counter} @result{} @r{counter}
12016 number) ; @r{number stays the same}
12022 which will first compute:
12026 (triangle-recursive-helper (+ 0 0) ; @r{sum}
12027 (1+ 0) ; @r{counter}
12031 (triangle-recursive-helper 0 1 1)
12035 Again, @code{(> counter number)} will be false, so again, the Lisp
12036 interpreter will evaluate @code{triangle-recursive-helper}, creating a
12037 new instance with new arguments.
12040 This new instance will be;
12044 (triangle-recursive-helper
12045 (+ sum counter) ; @r{sum plus counter} @result{} @r{sum}
12046 (1+ counter) ; @r{increment counter} @result{} @r{counter}
12047 number) ; @r{number stays the same}
12051 (triangle-recursive-helper 1 2 1)
12055 In this case, the @code{(> counter number)} test will be true! So the
12056 instance will return the value of the sum, which will be 1, as
12059 Now, let's pass @code{triangle-initialization} an argument
12060 of 2, to find out how many pebbles there are in a triangle with two rows.
12062 That function calls @code{(triangle-recursive-helper 0 0 2)}.
12065 In stages, the instances called will be:
12069 @r{sum counter number}
12070 (triangle-recursive-helper 0 1 2)
12072 (triangle-recursive-helper 1 2 2)
12074 (triangle-recursive-helper 3 3 2)
12078 When the last instance is called, the @code{(> counter number)} test
12079 will be true, so the instance will return the value of @code{sum},
12082 This kind of pattern helps when you are writing functions that can use
12083 many resources in a computer.
12086 @node Looping exercise
12087 @section Looping Exercise
12091 Write a function similar to @code{triangle} in which each row has a
12092 value which is the square of the row number. Use a @code{while} loop.
12095 Write a function similar to @code{triangle} that multiplies instead of
12099 Rewrite these two functions recursively. Rewrite these functions
12102 @c comma in printed title causes problem in Info cross reference
12104 Write a function for Texinfo mode that creates an index entry at the
12105 beginning of a paragraph for every @samp{@@dfn} within the paragraph.
12106 (In a Texinfo file, @samp{@@dfn} marks a definition. This book is
12107 written in Texinfo.)
12109 Many of the functions you will need are described in two of the
12110 previous chapters, @ref{Cutting & Storing Text, , Cutting and Storing
12111 Text}, and @ref{Yanking, , Yanking Text Back}. If you use
12112 @code{forward-paragraph} to put the index entry at the beginning of
12113 the paragraph, you will have to use @w{@kbd{C-h f}}
12114 (@code{describe-function}) to find out how to make the command go
12117 For more information, see
12119 @ref{Indicating, , Indicating Definitions, texinfo}.
12122 @ref{Indicating, , Indicating, texinfo, Texinfo Manual}, which goes to
12123 a Texinfo manual in the current directory. Or, if you are on the
12125 @uref{http://www.gnu.org/software/texinfo/manual/texinfo/}
12128 ``Indicating Definitions, Commands, etc.'' in @cite{Texinfo, The GNU
12129 Documentation Format}.
12133 @node Regexp Search
12134 @chapter Regular Expression Searches
12135 @cindex Searches, illustrating
12136 @cindex Regular expression searches
12137 @cindex Patterns, searching for
12138 @cindex Motion by sentence and paragraph
12139 @cindex Sentences, movement by
12140 @cindex Paragraphs, movement by
12142 Regular expression searches are used extensively in GNU Emacs. The
12143 two functions, @code{forward-sentence} and @code{forward-paragraph},
12144 illustrate these searches well. They use regular expressions to find
12145 where to move point. The phrase `regular expression' is often written
12148 Regular expression searches are described in @ref{Regexp Search, ,
12149 Regular Expression Search, emacs, The GNU Emacs Manual}, as well as in
12150 @ref{Regular Expressions, , , elisp, The GNU Emacs Lisp Reference
12151 Manual}. In writing this chapter, I am presuming that you have at
12152 least a mild acquaintance with them. The major point to remember is
12153 that regular expressions permit you to search for patterns as well as
12154 for literal strings of characters. For example, the code in
12155 @code{forward-sentence} searches for the pattern of possible
12156 characters that could mark the end of a sentence, and moves point to
12159 Before looking at the code for the @code{forward-sentence} function, it
12160 is worth considering what the pattern that marks the end of a sentence
12161 must be. The pattern is discussed in the next section; following that
12162 is a description of the regular expression search function,
12163 @code{re-search-forward}. The @code{forward-sentence} function
12164 is described in the section following. Finally, the
12165 @code{forward-paragraph} function is described in the last section of
12166 this chapter. @code{forward-paragraph} is a complex function that
12167 introduces several new features.
12170 * sentence-end:: The regular expression for @code{sentence-end}.
12171 * re-search-forward:: Very similar to @code{search-forward}.
12172 * forward-sentence:: A straightforward example of regexp search.
12173 * forward-paragraph:: A somewhat complex example.
12174 * etags:: How to create your own @file{TAGS} table.
12176 * re-search Exercises::
12180 @section The Regular Expression for @code{sentence-end}
12181 @findex sentence-end
12183 The symbol @code{sentence-end} is bound to the pattern that marks the
12184 end of a sentence. What should this regular expression be?
12186 Clearly, a sentence may be ended by a period, a question mark, or an
12187 exclamation mark. Indeed, in English, only clauses that end with one
12188 of those three characters should be considered the end of a sentence.
12189 This means that the pattern should include the character set:
12195 However, we do not want @code{forward-sentence} merely to jump to a
12196 period, a question mark, or an exclamation mark, because such a character
12197 might be used in the middle of a sentence. A period, for example, is
12198 used after abbreviations. So other information is needed.
12200 According to convention, you type two spaces after every sentence, but
12201 only one space after a period, a question mark, or an exclamation mark in
12202 the body of a sentence. So a period, a question mark, or an exclamation
12203 mark followed by two spaces is a good indicator of an end of sentence.
12204 However, in a file, the two spaces may instead be a tab or the end of a
12205 line. This means that the regular expression should include these three
12206 items as alternatives.
12209 This group of alternatives will look like this:
12220 Here, @samp{$} indicates the end of the line, and I have pointed out
12221 where the tab and two spaces are inserted in the expression. Both are
12222 inserted by putting the actual characters into the expression.
12224 Two backslashes, @samp{\\}, are required before the parentheses and
12225 vertical bars: the first backslash quotes the following backslash in
12226 Emacs; and the second indicates that the following character, the
12227 parenthesis or the vertical bar, is special.
12230 Also, a sentence may be followed by one or more carriage returns, like
12241 Like tabs and spaces, a carriage return is inserted into a regular
12242 expression by inserting it literally. The asterisk indicates that the
12243 @key{RET} is repeated zero or more times.
12245 But a sentence end does not consist only of a period, a question mark or
12246 an exclamation mark followed by appropriate space: a closing quotation
12247 mark or a closing brace of some kind may precede the space. Indeed more
12248 than one such mark or brace may precede the space. These require a
12249 expression that looks like this:
12255 In this expression, the first @samp{]} is the first character in the
12256 expression; the second character is @samp{"}, which is preceded by a
12257 @samp{\} to tell Emacs the @samp{"} is @emph{not} special. The last
12258 three characters are @samp{'}, @samp{)}, and @samp{@}}.
12260 All this suggests what the regular expression pattern for matching the
12261 end of a sentence should be; and, indeed, if we evaluate
12262 @code{sentence-end} we find that it returns the following value:
12267 @result{} "[.?!][]\"')@}]*\\($\\| \\| \\)[
12273 (Well, not in GNU Emacs 22; that is because of an effort to make the
12274 process simpler and to handle more glyphs and languages. When the
12275 value of @code{sentence-end} is @code{nil}, then use the value defined
12276 by the function @code{sentence-end}. (Here is a use of the difference
12277 between a value and a function in Emacs Lisp.) The function returns a
12278 value constructed from the variables @code{sentence-end-base},
12279 @code{sentence-end-double-space}, @code{sentence-end-without-period},
12280 and @code{sentence-end-without-space}. The critical variable is
12281 @code{sentence-end-base}; its global value is similar to the one
12282 described above but it also contains two additional quotation marks.
12283 These have differing degrees of curliness. The
12284 @code{sentence-end-without-period} variable, when true, tells Emacs
12285 that a sentence may end without a period, such as text in Thai.)
12289 (Note that here the @key{TAB}, two spaces, and @key{RET} are shown
12290 literally in the pattern.)
12292 This regular expression can be deciphered as follows:
12296 The first part of the pattern is the three characters, a period, a question
12297 mark and an exclamation mark, within square brackets. The pattern must
12298 begin with one or other of these characters.
12301 The second part of the pattern is the group of closing braces and
12302 quotation marks, which can appear zero or more times. These may follow
12303 the period, question mark or exclamation mark. In a regular expression,
12304 the backslash, @samp{\}, followed by the double quotation mark,
12305 @samp{"}, indicates the class of string-quote characters. Usually, the
12306 double quotation mark is the only character in this class. The
12307 asterisk, @samp{*}, indicates that the items in the previous group (the
12308 group surrounded by square brackets, @samp{[]}) may be repeated zero or
12311 @item \\($\\| \\| \\)
12312 The third part of the pattern is one or other of: either the end of a
12313 line, or two blank spaces, or a tab. The double back-slashes are used
12314 to prevent Emacs from reading the parentheses and vertical bars as part
12315 of the search pattern; the parentheses are used to mark the group and
12316 the vertical bars are used to indicated that the patterns to either side
12317 of them are alternatives. The dollar sign is used to indicate the end
12318 of a line and both the two spaces and the tab are each inserted as is to
12319 indicate what they are.
12322 Finally, the last part of the pattern indicates that the end of the line
12323 or the whitespace following the period, question mark or exclamation
12324 mark may, but need not, be followed by one or more carriage returns. In
12325 the pattern, the carriage return is inserted as an actual carriage
12326 return between square brackets but here it is shown as @key{RET}.
12330 @node re-search-forward
12331 @section The @code{re-search-forward} Function
12332 @findex re-search-forward
12334 The @code{re-search-forward} function is very like the
12335 @code{search-forward} function. (@xref{search-forward, , The
12336 @code{search-forward} Function}.)
12338 @code{re-search-forward} searches for a regular expression. If the
12339 search is successful, it leaves point immediately after the last
12340 character in the target. If the search is backwards, it leaves point
12341 just before the first character in the target. You may tell
12342 @code{re-search-forward} to return @code{t} for true. (Moving point
12343 is therefore a `side effect'.)
12345 Like @code{search-forward}, the @code{re-search-forward} function takes
12350 The first argument is the regular expression that the function searches
12351 for. The regular expression will be a string between quotation marks.
12354 The optional second argument limits how far the function will search; it is a
12355 bound, which is specified as a position in the buffer.
12358 The optional third argument specifies how the function responds to
12359 failure: @code{nil} as the third argument causes the function to
12360 signal an error (and print a message) when the search fails; any other
12361 value causes it to return @code{nil} if the search fails and @code{t}
12362 if the search succeeds.
12365 The optional fourth argument is the repeat count. A negative repeat
12366 count causes @code{re-search-forward} to search backwards.
12370 The template for @code{re-search-forward} looks like this:
12374 (re-search-forward "@var{regular-expression}"
12375 @var{limit-of-search}
12376 @var{what-to-do-if-search-fails}
12377 @var{repeat-count})
12381 The second, third, and fourth arguments are optional. However, if you
12382 want to pass a value to either or both of the last two arguments, you
12383 must also pass a value to all the preceding arguments. Otherwise, the
12384 Lisp interpreter will mistake which argument you are passing the value
12388 In the @code{forward-sentence} function, the regular expression will be
12389 the value of the variable @code{sentence-end}. In simple form, that is:
12393 "[.?!][]\"')@}]*\\($\\| \\| \\)[
12399 The limit of the search will be the end of the paragraph (since a
12400 sentence cannot go beyond a paragraph). If the search fails, the
12401 function will return @code{nil}; and the repeat count will be provided
12402 by the argument to the @code{forward-sentence} function.
12404 @node forward-sentence
12405 @section @code{forward-sentence}
12406 @findex forward-sentence
12408 The command to move the cursor forward a sentence is a straightforward
12409 illustration of how to use regular expression searches in Emacs Lisp.
12410 Indeed, the function looks longer and more complicated than it is; this
12411 is because the function is designed to go backwards as well as forwards;
12412 and, optionally, over more than one sentence. The function is usually
12413 bound to the key command @kbd{M-e}.
12416 * Complete forward-sentence::
12417 * fwd-sentence while loops:: Two @code{while} loops.
12418 * fwd-sentence re-search:: A regular expression search.
12422 @node Complete forward-sentence
12423 @unnumberedsubsec Complete @code{forward-sentence} function definition
12427 Here is the code for @code{forward-sentence}:
12432 (defun forward-sentence (&optional arg)
12433 "Move forward to next `sentence-end'. With argument, repeat.
12434 With negative argument, move backward repeatedly to `sentence-beginning'.
12436 The variable `sentence-end' is a regular expression that matches ends of
12437 sentences. Also, every paragraph boundary terminates sentences as well."
12441 (or arg (setq arg 1))
12442 (let ((opoint (point))
12443 (sentence-end (sentence-end)))
12445 (let ((pos (point))
12446 (par-beg (save-excursion (start-of-paragraph-text) (point))))
12447 (if (and (re-search-backward sentence-end par-beg t)
12448 (or (< (match-end 0) pos)
12449 (re-search-backward sentence-end par-beg t)))
12450 (goto-char (match-end 0))
12451 (goto-char par-beg)))
12452 (setq arg (1+ arg)))
12456 (let ((par-end (save-excursion (end-of-paragraph-text) (point))))
12457 (if (re-search-forward sentence-end par-end t)
12458 (skip-chars-backward " \t\n")
12459 (goto-char par-end)))
12460 (setq arg (1- arg)))
12461 (constrain-to-field nil opoint t)))
12469 (defun forward-sentence (&optional arg)
12470 "Move forward to next sentence-end. With argument, repeat.
12471 With negative argument, move backward repeatedly to sentence-beginning.
12472 Sentence ends are identified by the value of sentence-end
12473 treated as a regular expression. Also, every paragraph boundary
12474 terminates sentences as well."
12478 (or arg (setq arg 1))
12481 (save-excursion (start-of-paragraph-text) (point))))
12482 (if (re-search-backward
12483 (concat sentence-end "[^ \t\n]") par-beg t)
12484 (goto-char (1- (match-end 0)))
12485 (goto-char par-beg)))
12486 (setq arg (1+ arg)))
12489 (save-excursion (end-of-paragraph-text) (point))))
12490 (if (re-search-forward sentence-end par-end t)
12491 (skip-chars-backward " \t\n")
12492 (goto-char par-end)))
12493 (setq arg (1- arg))))
12498 The function looks long at first sight and it is best to look at its
12499 skeleton first, and then its muscle. The way to see the skeleton is to
12500 look at the expressions that start in the left-most columns:
12504 (defun forward-sentence (&optional arg)
12505 "@var{documentation}@dots{}"
12507 (or arg (setq arg 1))
12508 (let ((opoint (point)) (sentence-end (sentence-end)))
12510 (let ((pos (point))
12511 (par-beg (save-excursion (start-of-paragraph-text) (point))))
12512 @var{rest-of-body-of-while-loop-when-going-backwards}
12514 (let ((par-end (save-excursion (end-of-paragraph-text) (point))))
12515 @var{rest-of-body-of-while-loop-when-going-forwards}
12516 @var{handle-forms-and-equivalent}
12520 This looks much simpler! The function definition consists of
12521 documentation, an @code{interactive} expression, an @code{or}
12522 expression, a @code{let} expression, and @code{while} loops.
12524 Let's look at each of these parts in turn.
12526 We note that the documentation is thorough and understandable.
12528 The function has an @code{interactive "p"} declaration. This means
12529 that the processed prefix argument, if any, is passed to the
12530 function as its argument. (This will be a number.) If the function
12531 is not passed an argument (it is optional) then the argument
12532 @code{arg} will be bound to 1.
12534 When @code{forward-sentence} is called non-interactively without an
12535 argument, @code{arg} is bound to @code{nil}. The @code{or} expression
12536 handles this. What it does is either leave the value of @code{arg} as
12537 it is, but only if @code{arg} is bound to a value; or it sets the
12538 value of @code{arg} to 1, in the case when @code{arg} is bound to
12541 Next is a @code{let}. That specifies the values of two local
12542 variables, @code{point} and @code{sentence-end}. The local value of
12543 point, from before the search, is used in the
12544 @code{constrain-to-field} function which handles forms and
12545 equivalents. The @code{sentence-end} variable is set by the
12546 @code{sentence-end} function.
12548 @node fwd-sentence while loops
12549 @unnumberedsubsec The @code{while} loops
12551 Two @code{while} loops follow. The first @code{while} has a
12552 true-or-false-test that tests true if the prefix argument for
12553 @code{forward-sentence} is a negative number. This is for going
12554 backwards. The body of this loop is similar to the body of the second
12555 @code{while} clause, but it is not exactly the same. We will skip
12556 this @code{while} loop and concentrate on the second @code{while}
12560 The second @code{while} loop is for moving point forward. Its skeleton
12565 (while (> arg 0) ; @r{true-or-false-test}
12567 (if (@var{true-or-false-test})
12570 (setq arg (1- arg)))) ; @code{while} @r{loop decrementer}
12574 The @code{while} loop is of the decrementing kind.
12575 (@xref{Decrementing Loop, , A Loop with a Decrementing Counter}.) It
12576 has a true-or-false-test that tests true so long as the counter (in
12577 this case, the variable @code{arg}) is greater than zero; and it has a
12578 decrementer that subtracts 1 from the value of the counter every time
12581 If no prefix argument is given to @code{forward-sentence}, which is
12582 the most common way the command is used, this @code{while} loop will
12583 run once, since the value of @code{arg} will be 1.
12585 The body of the @code{while} loop consists of a @code{let} expression,
12586 which creates and binds a local variable, and has, as its body, an
12587 @code{if} expression.
12590 The body of the @code{while} loop looks like this:
12595 (save-excursion (end-of-paragraph-text) (point))))
12596 (if (re-search-forward sentence-end par-end t)
12597 (skip-chars-backward " \t\n")
12598 (goto-char par-end)))
12602 The @code{let} expression creates and binds the local variable
12603 @code{par-end}. As we shall see, this local variable is designed to
12604 provide a bound or limit to the regular expression search. If the
12605 search fails to find a proper sentence ending in the paragraph, it will
12606 stop on reaching the end of the paragraph.
12608 But first, let us examine how @code{par-end} is bound to the value of
12609 the end of the paragraph. What happens is that the @code{let} sets the
12610 value of @code{par-end} to the value returned when the Lisp interpreter
12611 evaluates the expression
12615 (save-excursion (end-of-paragraph-text) (point))
12620 In this expression, @code{(end-of-paragraph-text)} moves point to the
12621 end of the paragraph, @code{(point)} returns the value of point, and then
12622 @code{save-excursion} restores point to its original position. Thus,
12623 the @code{let} binds @code{par-end} to the value returned by the
12624 @code{save-excursion} expression, which is the position of the end of
12625 the paragraph. (The @code{end-of-paragraph-text} function uses
12626 @code{forward-paragraph}, which we will discuss shortly.)
12629 Emacs next evaluates the body of the @code{let}, which is an @code{if}
12630 expression that looks like this:
12634 (if (re-search-forward sentence-end par-end t) ; @r{if-part}
12635 (skip-chars-backward " \t\n") ; @r{then-part}
12636 (goto-char par-end))) ; @r{else-part}
12640 The @code{if} tests whether its first argument is true and if so,
12641 evaluates its then-part; otherwise, the Emacs Lisp interpreter
12642 evaluates the else-part. The true-or-false-test of the @code{if}
12643 expression is the regular expression search.
12645 It may seem odd to have what looks like the `real work' of
12646 the @code{forward-sentence} function buried here, but this is a common
12647 way this kind of operation is carried out in Lisp.
12649 @node fwd-sentence re-search
12650 @unnumberedsubsec The regular expression search
12652 The @code{re-search-forward} function searches for the end of the
12653 sentence, that is, for the pattern defined by the @code{sentence-end}
12654 regular expression. If the pattern is found---if the end of the sentence is
12655 found---then the @code{re-search-forward} function does two things:
12659 The @code{re-search-forward} function carries out a side effect, which
12660 is to move point to the end of the occurrence found.
12663 The @code{re-search-forward} function returns a value of true. This is
12664 the value received by the @code{if}, and means that the search was
12669 The side effect, the movement of point, is completed before the
12670 @code{if} function is handed the value returned by the successful
12671 conclusion of the search.
12673 When the @code{if} function receives the value of true from a successful
12674 call to @code{re-search-forward}, the @code{if} evaluates the then-part,
12675 which is the expression @code{(skip-chars-backward " \t\n")}. This
12676 expression moves backwards over any blank spaces, tabs or carriage
12677 returns until a printed character is found and then leaves point after
12678 the character. Since point has already been moved to the end of the
12679 pattern that marks the end of the sentence, this action leaves point
12680 right after the closing printed character of the sentence, which is
12683 On the other hand, if the @code{re-search-forward} function fails to
12684 find a pattern marking the end of the sentence, the function returns
12685 false. The false then causes the @code{if} to evaluate its third
12686 argument, which is @code{(goto-char par-end)}: it moves point to the
12687 end of the paragraph.
12689 (And if the text is in a form or equivalent, and point may not move
12690 fully, then the @code{constrain-to-field} function comes into play.)
12692 Regular expression searches are exceptionally useful and the pattern
12693 illustrated by @code{re-search-forward}, in which the search is the
12694 test of an @code{if} expression, is handy. You will see or write code
12695 incorporating this pattern often.
12697 @node forward-paragraph
12698 @section @code{forward-paragraph}: a Goldmine of Functions
12699 @findex forward-paragraph
12703 (defun forward-paragraph (&optional arg)
12704 "Move forward to end of paragraph.
12705 With argument ARG, do it ARG times;
12706 a negative argument ARG = -N means move backward N paragraphs.
12708 A line which `paragraph-start' matches either separates paragraphs
12709 \(if `paragraph-separate' matches it also) or is the first line of a paragraph.
12710 A paragraph end is the beginning of a line which is not part of the paragraph
12711 to which the end of the previous line belongs, or the end of the buffer.
12712 Returns the count of paragraphs left to move."
12714 (or arg (setq arg 1))
12715 (let* ((opoint (point))
12716 (fill-prefix-regexp
12717 (and fill-prefix (not (equal fill-prefix ""))
12718 (not paragraph-ignore-fill-prefix)
12719 (regexp-quote fill-prefix)))
12720 ;; Remove ^ from paragraph-start and paragraph-sep if they are there.
12721 ;; These regexps shouldn't be anchored, because we look for them
12722 ;; starting at the left-margin. This allows paragraph commands to
12723 ;; work normally with indented text.
12724 ;; This hack will not find problem cases like "whatever\\|^something".
12725 (parstart (if (and (not (equal "" paragraph-start))
12726 (equal ?^ (aref paragraph-start 0)))
12727 (substring paragraph-start 1)
12729 (parsep (if (and (not (equal "" paragraph-separate))
12730 (equal ?^ (aref paragraph-separate 0)))
12731 (substring paragraph-separate 1)
12732 paragraph-separate))
12734 (if fill-prefix-regexp
12735 (concat parsep "\\|"
12736 fill-prefix-regexp "[ \t]*$")
12738 ;; This is used for searching.
12739 (sp-parstart (concat "^[ \t]*\\(?:" parstart "\\|" parsep "\\)"))
12741 (while (and (< arg 0) (not (bobp)))
12742 (if (and (not (looking-at parsep))
12743 (re-search-backward "^\n" (max (1- (point)) (point-min)) t)
12744 (looking-at parsep))
12745 (setq arg (1+ arg))
12746 (setq start (point))
12747 ;; Move back over paragraph-separating lines.
12748 (forward-char -1) (beginning-of-line)
12749 (while (and (not (bobp))
12750 (progn (move-to-left-margin)
12751 (looking-at parsep)))
12755 (setq arg (1+ arg))
12756 ;; Go to end of the previous (non-separating) line.
12758 ;; Search back for line that starts or separates paragraphs.
12759 (if (if fill-prefix-regexp
12760 ;; There is a fill prefix; it overrides parstart.
12761 (let (multiple-lines)
12762 (while (and (progn (beginning-of-line) (not (bobp)))
12763 (progn (move-to-left-margin)
12764 (not (looking-at parsep)))
12765 (looking-at fill-prefix-regexp))
12766 (unless (= (point) start)
12767 (setq multiple-lines t))
12769 (move-to-left-margin)
12770 ;; This deleted code caused a long hanging-indent line
12771 ;; not to be filled together with the following lines.
12772 ;; ;; Don't move back over a line before the paragraph
12773 ;; ;; which doesn't start with fill-prefix
12774 ;; ;; unless that is the only line we've moved over.
12775 ;; (and (not (looking-at fill-prefix-regexp))
12777 ;; (forward-line 1))
12779 (while (and (re-search-backward sp-parstart nil 1)
12780 (setq found-start t)
12781 ;; Found a candidate, but need to check if it is a
12783 (progn (setq start (point))
12784 (move-to-left-margin)
12785 (not (looking-at parsep)))
12786 (not (and (looking-at parstart)
12787 (or (not use-hard-newlines)
12790 (1- start) 'hard)))))
12791 (setq found-start nil)
12796 ;; Move forward over paragraph separators.
12797 ;; We know this cannot reach the place we started
12798 ;; because we know we moved back over a non-separator.
12799 (while (and (not (eobp))
12800 (progn (move-to-left-margin)
12801 (looking-at parsep)))
12803 ;; If line before paragraph is just margin, back up to there.
12805 (if (> (current-column) (current-left-margin))
12807 (skip-chars-backward " \t")
12809 (forward-line 1))))
12810 ;; No starter or separator line => use buffer beg.
12811 (goto-char (point-min))))))
12813 (while (and (> arg 0) (not (eobp)))
12814 ;; Move forward over separator lines...
12815 (while (and (not (eobp))
12816 (progn (move-to-left-margin) (not (eobp)))
12817 (looking-at parsep))
12819 (unless (eobp) (setq arg (1- arg)))
12820 ;; ... and one more line.
12822 (if fill-prefix-regexp
12823 ;; There is a fill prefix; it overrides parstart.
12824 (while (and (not (eobp))
12825 (progn (move-to-left-margin) (not (eobp)))
12826 (not (looking-at parsep))
12827 (looking-at fill-prefix-regexp))
12829 (while (and (re-search-forward sp-parstart nil 1)
12830 (progn (setq start (match-beginning 0))
12833 (progn (move-to-left-margin)
12834 (not (looking-at parsep)))
12835 (or (not (looking-at parstart))
12836 (and use-hard-newlines
12837 (not (get-text-property (1- start) 'hard)))))
12839 (if (< (point) (point-max))
12840 (goto-char start))))
12841 (constrain-to-field nil opoint t)
12842 ;; Return the number of steps that could not be done.
12846 The @code{forward-paragraph} function moves point forward to the end
12847 of the paragraph. It is usually bound to @kbd{M-@}} and makes use of a
12848 number of functions that are important in themselves, including
12849 @code{let*}, @code{match-beginning}, and @code{looking-at}.
12851 The function definition for @code{forward-paragraph} is considerably
12852 longer than the function definition for @code{forward-sentence}
12853 because it works with a paragraph, each line of which may begin with a
12856 A fill prefix consists of a string of characters that are repeated at
12857 the beginning of each line. For example, in Lisp code, it is a
12858 convention to start each line of a paragraph-long comment with
12859 @samp{;;; }. In Text mode, four blank spaces make up another common
12860 fill prefix, creating an indented paragraph. (@xref{Fill Prefix, , ,
12861 emacs, The GNU Emacs Manual}, for more information about fill
12864 The existence of a fill prefix means that in addition to being able to
12865 find the end of a paragraph whose lines begin on the left-most
12866 column, the @code{forward-paragraph} function must be able to find the
12867 end of a paragraph when all or many of the lines in the buffer begin
12868 with the fill prefix.
12870 Moreover, it is sometimes practical to ignore a fill prefix that
12871 exists, especially when blank lines separate paragraphs.
12872 This is an added complication.
12875 * forward-paragraph in brief:: Key parts of the function definition.
12876 * fwd-para let:: The @code{let*} expression.
12877 * fwd-para while:: The forward motion @code{while} loop.
12881 @node forward-paragraph in brief
12882 @unnumberedsubsec Shortened @code{forward-paragraph} function definition
12885 Rather than print all of the @code{forward-paragraph} function, we
12886 will only print parts of it. Read without preparation, the function
12890 In outline, the function looks like this:
12894 (defun forward-paragraph (&optional arg)
12895 "@var{documentation}@dots{}"
12897 (or arg (setq arg 1))
12900 (while (and (< arg 0) (not (bobp))) ; @r{backward-moving-code}
12902 (while (and (> arg 0) (not (eobp))) ; @r{forward-moving-code}
12907 The first parts of the function are routine: the function's argument
12908 list consists of one optional argument. Documentation follows.
12910 The lower case @samp{p} in the @code{interactive} declaration means
12911 that the processed prefix argument, if any, is passed to the function.
12912 This will be a number, and is the repeat count of how many paragraphs
12913 point will move. The @code{or} expression in the next line handles
12914 the common case when no argument is passed to the function, which occurs
12915 if the function is called from other code rather than interactively.
12916 This case was described earlier. (@xref{forward-sentence, The
12917 @code{forward-sentence} function}.) Now we reach the end of the
12918 familiar part of this function.
12921 @unnumberedsubsec The @code{let*} expression
12923 The next line of the @code{forward-paragraph} function begins a
12924 @code{let*} expression. This is a different than @code{let}. The
12925 symbol is @code{let*} not @code{let}.
12927 The @code{let*} special form is like @code{let} except that Emacs sets
12928 each variable in sequence, one after another, and variables in the
12929 latter part of the varlist can make use of the values to which Emacs
12930 set variables in the earlier part of the varlist.
12933 ( refappend save-excursion, , code save-excursion in code append-to-buffer .)
12936 (@ref{append save-excursion, , @code{save-excursion} in @code{append-to-buffer}}.)
12938 In the @code{let*} expression in this function, Emacs binds a total of
12939 seven variables: @code{opoint}, @code{fill-prefix-regexp},
12940 @code{parstart}, @code{parsep}, @code{sp-parstart}, @code{start}, and
12941 @code{found-start}.
12943 The variable @code{parsep} appears twice, first, to remove instances
12944 of @samp{^}, and second, to handle fill prefixes.
12946 The variable @code{opoint} is just the value of @code{point}. As you
12947 can guess, it is used in a @code{constrain-to-field} expression, just
12948 as in @code{forward-sentence}.
12950 The variable @code{fill-prefix-regexp} is set to the value returned by
12951 evaluating the following list:
12956 (not (equal fill-prefix ""))
12957 (not paragraph-ignore-fill-prefix)
12958 (regexp-quote fill-prefix))
12963 This is an expression whose first element is the @code{and} special form.
12965 As we learned earlier (@pxref{kill-new function, , The @code{kill-new}
12966 function}), the @code{and} special form evaluates each of its
12967 arguments until one of the arguments returns a value of @code{nil}, in
12968 which case the @code{and} expression returns @code{nil}; however, if
12969 none of the arguments returns a value of @code{nil}, the value
12970 resulting from evaluating the last argument is returned. (Since such
12971 a value is not @code{nil}, it is considered true in Lisp.) In other
12972 words, an @code{and} expression returns a true value only if all its
12973 arguments are true.
12976 In this case, the variable @code{fill-prefix-regexp} is bound to a
12977 non-@code{nil} value only if the following four expressions produce a
12978 true (i.e., a non-@code{nil}) value when they are evaluated; otherwise,
12979 @code{fill-prefix-regexp} is bound to @code{nil}.
12983 When this variable is evaluated, the value of the fill prefix, if any,
12984 is returned. If there is no fill prefix, this variable returns
12987 @item (not (equal fill-prefix "")
12988 This expression checks whether an existing fill prefix is an empty
12989 string, that is, a string with no characters in it. An empty string is
12990 not a useful fill prefix.
12992 @item (not paragraph-ignore-fill-prefix)
12993 This expression returns @code{nil} if the variable
12994 @code{paragraph-ignore-fill-prefix} has been turned on by being set to a
12995 true value such as @code{t}.
12997 @item (regexp-quote fill-prefix)
12998 This is the last argument to the @code{and} special form. If all the
12999 arguments to the @code{and} are true, the value resulting from
13000 evaluating this expression will be returned by the @code{and} expression
13001 and bound to the variable @code{fill-prefix-regexp},
13004 @findex regexp-quote
13006 The result of evaluating this @code{and} expression successfully is that
13007 @code{fill-prefix-regexp} will be bound to the value of
13008 @code{fill-prefix} as modified by the @code{regexp-quote} function.
13009 What @code{regexp-quote} does is read a string and return a regular
13010 expression that will exactly match the string and match nothing else.
13011 This means that @code{fill-prefix-regexp} will be set to a value that
13012 will exactly match the fill prefix if the fill prefix exists.
13013 Otherwise, the variable will be set to @code{nil}.
13015 The next two local variables in the @code{let*} expression are
13016 designed to remove instances of @samp{^} from @code{parstart} and
13017 @code{parsep}, the local variables which indicate the paragraph start
13018 and the paragraph separator. The next expression sets @code{parsep}
13019 again. That is to handle fill prefixes.
13021 This is the setting that requires the definition call @code{let*}
13022 rather than @code{let}. The true-or-false-test for the @code{if}
13023 depends on whether the variable @code{fill-prefix-regexp} evaluates to
13024 @code{nil} or some other value.
13026 If @code{fill-prefix-regexp} does not have a value, Emacs evaluates
13027 the else-part of the @code{if} expression and binds @code{parsep} to
13028 its local value. (@code{parsep} is a regular expression that matches
13029 what separates paragraphs.)
13031 But if @code{fill-prefix-regexp} does have a value, Emacs evaluates
13032 the then-part of the @code{if} expression and binds @code{parsep} to a
13033 regular expression that includes the @code{fill-prefix-regexp} as part
13036 Specifically, @code{parsep} is set to the original value of the
13037 paragraph separate regular expression concatenated with an alternative
13038 expression that consists of the @code{fill-prefix-regexp} followed by
13039 optional whitespace to the end of the line. The whitespace is defined
13040 by @w{@code{"[ \t]*$"}}.) The @samp{\\|} defines this portion of the
13041 regexp as an alternative to @code{parsep}.
13043 According to a comment in the code, the next local variable,
13044 @code{sp-parstart}, is used for searching, and then the final two,
13045 @code{start} and @code{found-start}, are set to @code{nil}.
13047 Now we get into the body of the @code{let*}. The first part of the body
13048 of the @code{let*} deals with the case when the function is given a
13049 negative argument and is therefore moving backwards. We will skip this
13052 @node fwd-para while
13053 @unnumberedsubsec The forward motion @code{while} loop
13055 The second part of the body of the @code{let*} deals with forward
13056 motion. It is a @code{while} loop that repeats itself so long as the
13057 value of @code{arg} is greater than zero. In the most common use of
13058 the function, the value of the argument is 1, so the body of the
13059 @code{while} loop is evaluated exactly once, and the cursor moves
13060 forward one paragraph.
13063 (while (and (> arg 0) (not (eobp)))
13065 ;; Move forward over separator lines...
13066 (while (and (not (eobp))
13067 (progn (move-to-left-margin) (not (eobp)))
13068 (looking-at parsep))
13070 (unless (eobp) (setq arg (1- arg)))
13071 ;; ... and one more line.
13074 (if fill-prefix-regexp
13075 ;; There is a fill prefix; it overrides parstart.
13076 (while (and (not (eobp))
13077 (progn (move-to-left-margin) (not (eobp)))
13078 (not (looking-at parsep))
13079 (looking-at fill-prefix-regexp))
13082 (while (and (re-search-forward sp-parstart nil 1)
13083 (progn (setq start (match-beginning 0))
13086 (progn (move-to-left-margin)
13087 (not (looking-at parsep)))
13088 (or (not (looking-at parstart))
13089 (and use-hard-newlines
13090 (not (get-text-property (1- start) 'hard)))))
13093 (if (< (point) (point-max))
13094 (goto-char start))))
13097 This part handles three situations: when point is between paragraphs,
13098 when there is a fill prefix and when there is no fill prefix.
13101 The @code{while} loop looks like this:
13105 ;; @r{going forwards and not at the end of the buffer}
13106 (while (and (> arg 0) (not (eobp)))
13108 ;; @r{between paragraphs}
13109 ;; Move forward over separator lines...
13110 (while (and (not (eobp))
13111 (progn (move-to-left-margin) (not (eobp)))
13112 (looking-at parsep))
13114 ;; @r{This decrements the loop}
13115 (unless (eobp) (setq arg (1- arg)))
13116 ;; ... and one more line.
13121 (if fill-prefix-regexp
13122 ;; There is a fill prefix; it overrides parstart;
13123 ;; we go forward line by line
13124 (while (and (not (eobp))
13125 (progn (move-to-left-margin) (not (eobp)))
13126 (not (looking-at parsep))
13127 (looking-at fill-prefix-regexp))
13132 ;; There is no fill prefix;
13133 ;; we go forward character by character
13134 (while (and (re-search-forward sp-parstart nil 1)
13135 (progn (setq start (match-beginning 0))
13138 (progn (move-to-left-margin)
13139 (not (looking-at parsep)))
13140 (or (not (looking-at parstart))
13141 (and use-hard-newlines
13142 (not (get-text-property (1- start) 'hard)))))
13147 ;; and if there is no fill prefix and if we are not at the end,
13148 ;; go to whatever was found in the regular expression search
13150 (if (< (point) (point-max))
13151 (goto-char start))))
13156 We can see that this is a decrementing counter @code{while} loop,
13157 using the expression @code{(setq arg (1- arg))} as the decrementer.
13158 That expression is not far from the @code{while}, but is hidden in
13159 another Lisp macro, an @code{unless} macro. Unless we are at the end
13160 of the buffer---that is what the @code{eobp} function determines; it
13161 is an abbreviation of @samp{End Of Buffer P}---we decrease the value
13162 of @code{arg} by one.
13164 (If we are at the end of the buffer, we cannot go forward any more and
13165 the next loop of the @code{while} expression will test false since the
13166 test is an @code{and} with @code{(not (eobp))}. The @code{not}
13167 function means exactly as you expect; it is another name for
13168 @code{null}, a function that returns true when its argument is false.)
13170 Interestingly, the loop count is not decremented until we leave the
13171 space between paragraphs, unless we come to the end of buffer or stop
13172 seeing the local value of the paragraph separator.
13174 That second @code{while} also has a @code{(move-to-left-margin)}
13175 expression. The function is self-explanatory. It is inside a
13176 @code{progn} expression and not the last element of its body, so it is
13177 only invoked for its side effect, which is to move point to the left
13178 margin of the current line.
13181 The @code{looking-at} function is also self-explanatory; it returns
13182 true if the text after point matches the regular expression given as
13185 The rest of the body of the loop looks difficult at first, but makes
13186 sense as you come to understand it.
13189 First consider what happens if there is a fill prefix:
13193 (if fill-prefix-regexp
13194 ;; There is a fill prefix; it overrides parstart;
13195 ;; we go forward line by line
13196 (while (and (not (eobp))
13197 (progn (move-to-left-margin) (not (eobp)))
13198 (not (looking-at parsep))
13199 (looking-at fill-prefix-regexp))
13205 This expression moves point forward line by line so long
13206 as four conditions are true:
13210 Point is not at the end of the buffer.
13213 We can move to the left margin of the text and are
13214 not at the end of the buffer.
13217 The text following point does not separate paragraphs.
13220 The pattern following point is the fill prefix regular expression.
13223 The last condition may be puzzling, until you remember that point was
13224 moved to the beginning of the line early in the @code{forward-paragraph}
13225 function. This means that if the text has a fill prefix, the
13226 @code{looking-at} function will see it.
13229 Consider what happens when there is no fill prefix.
13233 (while (and (re-search-forward sp-parstart nil 1)
13234 (progn (setq start (match-beginning 0))
13237 (progn (move-to-left-margin)
13238 (not (looking-at parsep)))
13239 (or (not (looking-at parstart))
13240 (and use-hard-newlines
13241 (not (get-text-property (1- start) 'hard)))))
13247 This @code{while} loop has us searching forward for
13248 @code{sp-parstart}, which is the combination of possible whitespace
13249 with a the local value of the start of a paragraph or of a paragraph
13250 separator. (The latter two are within an expression starting
13251 @code{\(?:} so that they are not referenced by the
13252 @code{match-beginning} function.)
13255 The two expressions,
13259 (setq start (match-beginning 0))
13265 mean go to the start of the text matched by the regular expression
13268 The @code{(match-beginning 0)} expression is new. It returns a number
13269 specifying the location of the start of the text that was matched by
13272 The @code{match-beginning} function is used here because of a
13273 characteristic of a forward search: a successful forward search,
13274 regardless of whether it is a plain search or a regular expression
13275 search, moves point to the end of the text that is found. In this
13276 case, a successful search moves point to the end of the pattern for
13277 @code{sp-parstart}.
13279 However, we want to put point at the end of the current paragraph, not
13280 somewhere else. Indeed, since the search possibly includes the
13281 paragraph separator, point may end up at the beginning of the next one
13282 unless we use an expression that includes @code{match-beginning}.
13284 @findex match-beginning
13285 When given an argument of 0, @code{match-beginning} returns the
13286 position that is the start of the text matched by the most recent
13287 search. In this case, the most recent search looks for
13288 @code{sp-parstart}. The @code{(match-beginning 0)} expression returns
13289 the beginning position of that pattern, rather than the end position
13292 (Incidentally, when passed a positive number as an argument, the
13293 @code{match-beginning} function returns the location of point at that
13294 parenthesized expression in the last search unless that parenthesized
13295 expression begins with @code{\(?:}. I don't know why @code{\(?:}
13296 appears here since the argument is 0.)
13299 The last expression when there is no fill prefix is
13303 (if (< (point) (point-max))
13304 (goto-char start))))
13309 This says that if there is no fill prefix and if we are not at the
13310 end, point should move to the beginning of whatever was found by the
13311 regular expression search for @code{sp-parstart}.
13313 The full definition for the @code{forward-paragraph} function not only
13314 includes code for going forwards, but also code for going backwards.
13316 If you are reading this inside of GNU Emacs and you want to see the
13317 whole function, you can type @kbd{C-h f} (@code{describe-function})
13318 and the name of the function. This gives you the function
13319 documentation and the name of the library containing the function's
13320 source. Place point over the name of the library and press the RET
13321 key; you will be taken directly to the source. (Be sure to install
13322 your sources! Without them, you are like a person who tries to drive
13323 a car with his eyes shut!)
13326 @section Create Your Own @file{TAGS} File
13328 @cindex @file{TAGS} file, create own
13330 Besides @kbd{C-h f} (@code{describe-function}), another way to see the
13331 source of a function is to type @kbd{M-.} (@code{find-tag}) and the
13332 name of the function when prompted for it. This is a good habit to
13333 get into. The @kbd{M-.} (@code{find-tag}) command takes you directly
13334 to the source for a function, variable, or node. The function depends
13335 on tags tables to tell it where to go.
13337 If the @code{find-tag} function first asks you for the name of a
13338 @file{TAGS} table, give it the name of a @file{TAGS} file such as
13339 @file{/usr/local/src/emacs/src/TAGS}. (The exact path to your
13340 @file{TAGS} file depends on how your copy of Emacs was installed. I
13341 just told you the location that provides both my C and my Emacs Lisp
13344 You can also create your own @file{TAGS} file for directories that
13347 You often need to build and install tags tables yourself. They are
13348 not built automatically. A tags table is called a @file{TAGS} file;
13349 the name is in upper case letters.
13351 You can create a @file{TAGS} file by calling the @code{etags} program
13352 that comes as a part of the Emacs distribution. Usually, @code{etags}
13353 is compiled and installed when Emacs is built. (@code{etags} is not
13354 an Emacs Lisp function or a part of Emacs; it is a C program.)
13357 To create a @file{TAGS} file, first switch to the directory in which
13358 you want to create the file. In Emacs you can do this with the
13359 @kbd{M-x cd} command, or by visiting a file in the directory, or by
13360 listing the directory with @kbd{C-x d} (@code{dired}). Then run the
13361 compile command, with @w{@code{etags *.el}} as the command to execute
13364 M-x compile RET etags *.el RET
13368 to create a @file{TAGS} file for Emacs Lisp.
13370 For example, if you have a large number of files in your
13371 @file{~/emacs} directory, as I do---I have 137 @file{.el} files in it,
13372 of which I load 12---you can create a @file{TAGS} file for the Emacs
13373 Lisp files in that directory.
13376 The @code{etags} program takes all the usual shell `wildcards'. For
13377 example, if you have two directories for which you want a single
13378 @file{TAGS} file, type @w{@code{etags *.el ../elisp/*.el}}, where
13379 @file{../elisp/} is the second directory:
13382 M-x compile RET etags *.el ../elisp/*.el RET
13389 M-x compile RET etags --help RET
13393 to see a list of the options accepted by @code{etags} as well as a
13394 list of supported languages.
13396 The @code{etags} program handles more than 20 languages, including
13397 Emacs Lisp, Common Lisp, Scheme, C, C++, Ada, Fortran, HTML, Java,
13398 LaTeX, Pascal, Perl, PostScript, Python, TeX, Texinfo, makefiles, and
13399 most assemblers. The program has no switches for specifying the
13400 language; it recognizes the language in an input file according to its
13401 file name and contents.
13403 @file{etags} is very helpful when you are writing code yourself and
13404 want to refer back to functions you have already written. Just run
13405 @code{etags} again at intervals as you write new functions, so they
13406 become part of the @file{TAGS} file.
13408 If you think an appropriate @file{TAGS} file already exists for what
13409 you want, but do not know where it is, you can use the @code{locate}
13410 program to attempt to find it.
13412 Type @w{@kbd{M-x locate @key{RET} TAGS @key{RET}}} and Emacs will list
13413 for you the full path names of all your @file{TAGS} files. On my
13414 system, this command lists 34 @file{TAGS} files. On the other hand, a
13415 `plain vanilla' system I recently installed did not contain any
13418 If the tags table you want has been created, you can use the @code{M-x
13419 visit-tags-table} command to specify it. Otherwise, you will need to
13420 create the tag table yourself and then use @code{M-x
13423 @subsubheading Building Tags in the Emacs sources
13424 @cindex Building Tags in the Emacs sources
13425 @cindex Tags in the Emacs sources
13428 The GNU Emacs sources come with a @file{Makefile} that contains a
13429 sophisticated @code{etags} command that creates, collects, and merges
13430 tags tables from all over the Emacs sources and puts the information
13431 into one @file{TAGS} file in the @file{src/} directory. (The
13432 @file{src/} directory is below the top level of your Emacs directory.)
13435 To build this @file{TAGS} file, go to the top level of your Emacs
13436 source directory and run the compile command @code{make tags}:
13439 M-x compile RET make tags RET
13443 (The @code{make tags} command works well with the GNU Emacs sources,
13444 as well as with some other source packages.)
13446 For more information, see @ref{Tags, , Tag Tables, emacs, The GNU Emacs
13449 @node Regexp Review
13452 Here is a brief summary of some recently introduced functions.
13456 Repeatedly evaluate the body of the expression so long as the first
13457 element of the body tests true. Then return @code{nil}. (The
13458 expression is evaluated only for its side effects.)
13467 (insert (format "foo is %d.\n" foo))
13468 (setq foo (1- foo))))
13470 @result{} foo is 2.
13477 (The @code{insert} function inserts its arguments at point; the
13478 @code{format} function returns a string formatted from its arguments
13479 the way @code{message} formats its arguments; @code{\n} produces a new
13482 @item re-search-forward
13483 Search for a pattern, and if the pattern is found, move point to rest
13487 Takes four arguments, like @code{search-forward}:
13491 A regular expression that specifies the pattern to search for.
13492 (Remember to put quotation marks around this argument!)
13495 Optionally, the limit of the search.
13498 Optionally, what to do if the search fails, return @code{nil} or an
13502 Optionally, how many times to repeat the search; if negative, the
13503 search goes backwards.
13507 Bind some variables locally to particular values,
13508 and then evaluate the remaining arguments, returning the value of the
13509 last one. While binding the local variables, use the local values of
13510 variables bound earlier, if any.
13519 (message "`bar' is %d." bar))
13520 @result{} `bar' is 21.
13524 @item match-beginning
13525 Return the position of the start of the text found by the last regular
13529 Return @code{t} for true if the text after point matches the argument,
13530 which should be a regular expression.
13533 Return @code{t} for true if point is at the end of the accessible part
13534 of a buffer. The end of the accessible part is the end of the buffer
13535 if the buffer is not narrowed; it is the end of the narrowed part if
13536 the buffer is narrowed.
13540 @node re-search Exercises
13541 @section Exercises with @code{re-search-forward}
13545 Write a function to search for a regular expression that matches two
13546 or more blank lines in sequence.
13549 Write a function to search for duplicated words, such as `the the'.
13550 @xref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs
13551 Manual}, for information on how to write a regexp (a regular
13552 expression) to match a string that is composed of two identical
13553 halves. You can devise several regexps; some are better than others.
13554 The function I use is described in an appendix, along with several
13555 regexps. @xref{the-the, , @code{the-the} Duplicated Words Function}.
13558 @node Counting Words
13559 @chapter Counting: Repetition and Regexps
13560 @cindex Repetition for word counting
13561 @cindex Regular expressions for word counting
13563 Repetition and regular expression searches are powerful tools that you
13564 often use when you write code in Emacs Lisp. This chapter illustrates
13565 the use of regular expression searches through the construction of
13566 word count commands using @code{while} loops and recursion.
13569 * Why Count Words::
13570 * @value{COUNT-WORDS}:: Use a regexp, but find a problem.
13571 * recursive-count-words:: Start with case of no words in region.
13572 * Counting Exercise::
13576 @node Why Count Words
13577 @unnumberedsec Counting words
13580 The standard Emacs distribution contains functions for counting the
13581 number of lines and words within a region.
13583 Certain types of writing ask you to count words. Thus, if you write
13584 an essay, you may be limited to 800 words; if you write a novel, you
13585 may discipline yourself to write 1000 words a day. It seems odd, but
13586 for a long time, Emacs lacked a word count command. Perhaps people used
13587 Emacs mostly for code or types of documentation that did not require
13588 word counts; or perhaps they restricted themselves to the operating
13589 system word count command, @code{wc}. Alternatively, people may have
13590 followed the publishers' convention and computed a word count by
13591 dividing the number of characters in a document by five.
13593 There are many ways to implement a command to count words. Here are
13594 some examples, which you may wish to compare with the standard Emacs
13595 command, @code{count-words-region}.
13597 @node @value{COUNT-WORDS}
13598 @section The @code{@value{COUNT-WORDS}} Function
13599 @findex @value{COUNT-WORDS}
13601 A word count command could count words in a line, paragraph, region,
13602 or buffer. What should the command cover? You could design the
13603 command to count the number of words in a complete buffer. However,
13604 the Emacs tradition encourages flexibility---you may want to count
13605 words in just a section, rather than all of a buffer. So it makes
13606 more sense to design the command to count the number of words in a
13607 region. Once you have a command to count words in a region, you can,
13608 if you wish, count words in a whole buffer by marking it with
13609 @w{@kbd{C-x h}} (@code{mark-whole-buffer}).
13611 Clearly, counting words is a repetitive act: starting from the
13612 beginning of the region, you count the first word, then the second
13613 word, then the third word, and so on, until you reach the end of the
13614 region. This means that word counting is ideally suited to recursion
13615 or to a @code{while} loop.
13618 * Design @value{COUNT-WORDS}:: The definition using a @code{while} loop.
13619 * Whitespace Bug:: The Whitespace Bug in @code{@value{COUNT-WORDS}}.
13623 @node Design @value{COUNT-WORDS}
13624 @unnumberedsubsec Designing @code{@value{COUNT-WORDS}}
13627 First, we will implement the word count command with a @code{while}
13628 loop, then with recursion. The command will, of course, be
13632 The template for an interactive function definition is, as always:
13636 (defun @var{name-of-function} (@var{argument-list})
13637 "@var{documentation}@dots{}"
13638 (@var{interactive-expression}@dots{})
13643 What we need to do is fill in the slots.
13645 The name of the function should be self-explanatory and similar to the
13646 existing @code{count-lines-region} name. This makes the name easier
13647 to remember. @code{count-words-region} is the obvious choice. Since
13648 that name is now used for the standard Emacs command to count words, we
13649 will name our implementation @code{@value{COUNT-WORDS}}.
13651 The function counts words within a region. This means that the
13652 argument list must contain symbols that are bound to the two
13653 positions, the beginning and end of the region. These two positions
13654 can be called @samp{beginning} and @samp{end} respectively. The first
13655 line of the documentation should be a single sentence, since that is
13656 all that is printed as documentation by a command such as
13657 @code{apropos}. The interactive expression will be of the form
13658 @samp{(interactive "r")}, since that will cause Emacs to pass the
13659 beginning and end of the region to the function's argument list. All
13662 The body of the function needs to be written to do three tasks:
13663 first, to set up conditions under which the @code{while} loop can
13664 count words, second, to run the @code{while} loop, and third, to send
13665 a message to the user.
13667 When a user calls @code{@value{COUNT-WORDS}}, point may be at the
13668 beginning or the end of the region. However, the counting process
13669 must start at the beginning of the region. This means we will want
13670 to put point there if it is not already there. Executing
13671 @code{(goto-char beginning)} ensures this. Of course, we will want to
13672 return point to its expected position when the function finishes its
13673 work. For this reason, the body must be enclosed in a
13674 @code{save-excursion} expression.
13676 The central part of the body of the function consists of a
13677 @code{while} loop in which one expression jumps point forward word by
13678 word, and another expression counts those jumps. The true-or-false-test
13679 of the @code{while} loop should test true so long as point should jump
13680 forward, and false when point is at the end of the region.
13682 We could use @code{(forward-word 1)} as the expression for moving point
13683 forward word by word, but it is easier to see what Emacs identifies as a
13684 `word' if we use a regular expression search.
13686 A regular expression search that finds the pattern for which it is
13687 searching leaves point after the last character matched. This means
13688 that a succession of successful word searches will move point forward
13691 As a practical matter, we want the regular expression search to jump
13692 over whitespace and punctuation between words as well as over the
13693 words themselves. A regexp that refuses to jump over interword
13694 whitespace would never jump more than one word! This means that
13695 the regexp should include the whitespace and punctuation that follows
13696 a word, if any, as well as the word itself. (A word may end a buffer
13697 and not have any following whitespace or punctuation, so that part of
13698 the regexp must be optional.)
13700 Thus, what we want for the regexp is a pattern defining one or more
13701 word constituent characters followed, optionally, by one or more
13702 characters that are not word constituents. The regular expression for
13710 The buffer's syntax table determines which characters are and are not
13711 word constituents. For more information about syntax,
13712 @pxref{Syntax Tables, , Syntax Tables, elisp, The GNU Emacs Lisp
13716 The search expression looks like this:
13719 (re-search-forward "\\w+\\W*")
13723 (Note that paired backslashes precede the @samp{w} and @samp{W}. A
13724 single backslash has special meaning to the Emacs Lisp interpreter.
13725 It indicates that the following character is interpreted differently
13726 than usual. For example, the two characters, @samp{\n}, stand for
13727 @samp{newline}, rather than for a backslash followed by @samp{n}. Two
13728 backslashes in a row stand for an ordinary, `unspecial' backslash, so
13729 Emacs Lisp interpreter ends of seeing a single backslash followed by a
13730 letter. So it discovers the letter is special.)
13732 We need a counter to count how many words there are; this variable
13733 must first be set to 0 and then incremented each time Emacs goes
13734 around the @code{while} loop. The incrementing expression is simply:
13737 (setq count (1+ count))
13740 Finally, we want to tell the user how many words there are in the
13741 region. The @code{message} function is intended for presenting this
13742 kind of information to the user. The message has to be phrased so
13743 that it reads properly regardless of how many words there are in the
13744 region: we don't want to say that ``there are 1 words in the region''.
13745 The conflict between singular and plural is ungrammatical. We can
13746 solve this problem by using a conditional expression that evaluates
13747 different messages depending on the number of words in the region.
13748 There are three possibilities: no words in the region, one word in the
13749 region, and more than one word. This means that the @code{cond}
13750 special form is appropriate.
13753 All this leads to the following function definition:
13757 ;;; @r{First version; has bugs!}
13758 (defun @value{COUNT-WORDS} (beginning end)
13759 "Print number of words in the region.
13760 Words are defined as at least one word-constituent
13761 character followed by at least one character that
13762 is not a word-constituent. The buffer's syntax
13763 table determines which characters these are."
13765 (message "Counting words in region ... ")
13769 ;;; @r{1. Set up appropriate conditions.}
13771 (goto-char beginning)
13776 ;;; @r{2. Run the} while @r{loop.}
13777 (while (< (point) end)
13778 (re-search-forward "\\w+\\W*")
13779 (setq count (1+ count)))
13783 ;;; @r{3. Send a message to the user.}
13784 (cond ((zerop count)
13786 "The region does NOT have any words."))
13789 "The region has 1 word."))
13792 "The region has %d words." count))))))
13797 As written, the function works, but not in all circumstances.
13799 @node Whitespace Bug
13800 @subsection The Whitespace Bug in @code{@value{COUNT-WORDS}}
13802 The @code{@value{COUNT-WORDS}} command described in the preceding
13803 section has two bugs, or rather, one bug with two manifestations.
13804 First, if you mark a region containing only whitespace in the middle
13805 of some text, the @code{@value{COUNT-WORDS}} command tells you that the
13806 region contains one word! Second, if you mark a region containing
13807 only whitespace at the end of the buffer or the accessible portion of
13808 a narrowed buffer, the command displays an error message that looks
13812 Search failed: "\\w+\\W*"
13815 If you are reading this in Info in GNU Emacs, you can test for these
13818 First, evaluate the function in the usual manner to install it.
13820 Here is a copy of the definition. Place your cursor after the closing
13821 parenthesis and type @kbd{C-x C-e} to install it.
13825 ;; @r{First version; has bugs!}
13826 (defun @value{COUNT-WORDS} (beginning end)
13827 "Print number of words in the region.
13828 Words are defined as at least one word-constituent character followed
13829 by at least one character that is not a word-constituent. The buffer's
13830 syntax table determines which characters these are."
13834 (message "Counting words in region ... ")
13838 ;;; @r{1. Set up appropriate conditions.}
13840 (goto-char beginning)
13845 ;;; @r{2. Run the} while @r{loop.}
13846 (while (< (point) end)
13847 (re-search-forward "\\w+\\W*")
13848 (setq count (1+ count)))
13852 ;;; @r{3. Send a message to the user.}
13853 (cond ((zerop count)
13854 (message "The region does NOT have any words."))
13855 ((= 1 count) (message "The region has 1 word."))
13856 (t (message "The region has %d words." count))))))
13862 If you wish, you can also install this keybinding by evaluating it:
13865 (global-set-key "\C-c=" '@value{COUNT-WORDS})
13868 To conduct the first test, set mark and point to the beginning and end
13869 of the following line and then type @kbd{C-c =} (or @kbd{M-x
13870 @value{COUNT-WORDS}} if you have not bound @kbd{C-c =}):
13877 Emacs will tell you, correctly, that the region has three words.
13879 Repeat the test, but place mark at the beginning of the line and place
13880 point just @emph{before} the word @samp{one}. Again type the command
13881 @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}). Emacs should tell you
13882 that the region has no words, since it is composed only of the
13883 whitespace at the beginning of the line. But instead Emacs tells you
13884 that the region has one word!
13886 For the third test, copy the sample line to the end of the
13887 @file{*scratch*} buffer and then type several spaces at the end of the
13888 line. Place mark right after the word @samp{three} and point at the
13889 end of line. (The end of the line will be the end of the buffer.)
13890 Type @kbd{C-c =} (or @kbd{M-x @value{COUNT-WORDS}}) as you did before.
13891 Again, Emacs should tell you that the region has no words, since it is
13892 composed only of the whitespace at the end of the line. Instead,
13893 Emacs displays an error message saying @samp{Search failed}.
13895 The two bugs stem from the same problem.
13897 Consider the first manifestation of the bug, in which the command
13898 tells you that the whitespace at the beginning of the line contains
13899 one word. What happens is this: The @code{M-x @value{COUNT-WORDS}}
13900 command moves point to the beginning of the region. The @code{while}
13901 tests whether the value of point is smaller than the value of
13902 @code{end}, which it is. Consequently, the regular expression search
13903 looks for and finds the first word. It leaves point after the word.
13904 @code{count} is set to one. The @code{while} loop repeats; but this
13905 time the value of point is larger than the value of @code{end}, the
13906 loop is exited; and the function displays a message saying the number
13907 of words in the region is one. In brief, the regular expression
13908 search looks for and finds the word even though it is outside
13911 In the second manifestation of the bug, the region is whitespace at
13912 the end of the buffer. Emacs says @samp{Search failed}. What happens
13913 is that the true-or-false-test in the @code{while} loop tests true, so
13914 the search expression is executed. But since there are no more words
13915 in the buffer, the search fails.
13917 In both manifestations of the bug, the search extends or attempts to
13918 extend outside of the region.
13920 The solution is to limit the search to the region---this is a fairly
13921 simple action, but as you may have come to expect, it is not quite as
13922 simple as you might think.
13924 As we have seen, the @code{re-search-forward} function takes a search
13925 pattern as its first argument. But in addition to this first,
13926 mandatory argument, it accepts three optional arguments. The optional
13927 second argument bounds the search. The optional third argument, if
13928 @code{t}, causes the function to return @code{nil} rather than signal
13929 an error if the search fails. The optional fourth argument is a
13930 repeat count. (In Emacs, you can see a function's documentation by
13931 typing @kbd{C-h f}, the name of the function, and then @key{RET}.)
13933 In the @code{@value{COUNT-WORDS}} definition, the value of the end of
13934 the region is held by the variable @code{end} which is passed as an
13935 argument to the function. Thus, we can add @code{end} as an argument
13936 to the regular expression search expression:
13939 (re-search-forward "\\w+\\W*" end)
13942 However, if you make only this change to the @code{@value{COUNT-WORDS}}
13943 definition and then test the new version of the definition on a
13944 stretch of whitespace, you will receive an error message saying
13945 @samp{Search failed}.
13947 What happens is this: the search is limited to the region, and fails
13948 as you expect because there are no word-constituent characters in the
13949 region. Since it fails, we receive an error message. But we do not
13950 want to receive an error message in this case; we want to receive the
13951 message that "The region does NOT have any words."
13953 The solution to this problem is to provide @code{re-search-forward}
13954 with a third argument of @code{t}, which causes the function to return
13955 @code{nil} rather than signal an error if the search fails.
13957 However, if you make this change and try it, you will see the message
13958 ``Counting words in region ... '' and @dots{} you will keep on seeing
13959 that message @dots{}, until you type @kbd{C-g} (@code{keyboard-quit}).
13961 Here is what happens: the search is limited to the region, as before,
13962 and it fails because there are no word-constituent characters in the
13963 region, as expected. Consequently, the @code{re-search-forward}
13964 expression returns @code{nil}. It does nothing else. In particular,
13965 it does not move point, which it does as a side effect if it finds the
13966 search target. After the @code{re-search-forward} expression returns
13967 @code{nil}, the next expression in the @code{while} loop is evaluated.
13968 This expression increments the count. Then the loop repeats. The
13969 true-or-false-test tests true because the value of point is still less
13970 than the value of end, since the @code{re-search-forward} expression
13971 did not move point. @dots{} and the cycle repeats @dots{}
13973 The @code{@value{COUNT-WORDS}} definition requires yet another
13974 modification, to cause the true-or-false-test of the @code{while} loop
13975 to test false if the search fails. Put another way, there are two
13976 conditions that must be satisfied in the true-or-false-test before the
13977 word count variable is incremented: point must still be within the
13978 region and the search expression must have found a word to count.
13980 Since both the first condition and the second condition must be true
13981 together, the two expressions, the region test and the search
13982 expression, can be joined with an @code{and} special form and embedded in
13983 the @code{while} loop as the true-or-false-test, like this:
13986 (and (< (point) end) (re-search-forward "\\w+\\W*" end t))
13989 @c colon in printed section title causes problem in Info cross reference
13990 @c also trouble with an overfull hbox
13993 (For information about @code{and}, see
13994 @ref{kill-new function, , The @code{kill-new} function}.)
13998 (@xref{kill-new function, , The @code{kill-new} function}, for
13999 information about @code{and}.)
14002 The @code{re-search-forward} expression returns @code{t} if the search
14003 succeeds and as a side effect moves point. Consequently, as words are
14004 found, point is moved through the region. When the search expression
14005 fails to find another word, or when point reaches the end of the
14006 region, the true-or-false-test tests false, the @code{while} loop
14007 exits, and the @code{@value{COUNT-WORDS}} function displays one or
14008 other of its messages.
14010 After incorporating these final changes, the @code{@value{COUNT-WORDS}}
14011 works without bugs (or at least, without bugs that I have found!).
14012 Here is what it looks like:
14016 ;;; @r{Final version:} @code{while}
14017 (defun @value{COUNT-WORDS} (beginning end)
14018 "Print number of words in the region."
14020 (message "Counting words in region ... ")
14024 ;;; @r{1. Set up appropriate conditions.}
14027 (goto-char beginning)
14031 ;;; @r{2. Run the} while @r{loop.}
14032 (while (and (< (point) end)
14033 (re-search-forward "\\w+\\W*" end t))
14034 (setq count (1+ count)))
14038 ;;; @r{3. Send a message to the user.}
14039 (cond ((zerop count)
14041 "The region does NOT have any words."))
14044 "The region has 1 word."))
14047 "The region has %d words." count))))))
14051 @node recursive-count-words
14052 @section Count Words Recursively
14053 @cindex Count words recursively
14054 @cindex Recursively counting words
14055 @cindex Words, counted recursively
14057 You can write the function for counting words recursively as well as
14058 with a @code{while} loop. Let's see how this is done.
14060 First, we need to recognize that the @code{@value{COUNT-WORDS}}
14061 function has three jobs: it sets up the appropriate conditions for
14062 counting to occur; it counts the words in the region; and it sends a
14063 message to the user telling how many words there are.
14065 If we write a single recursive function to do everything, we will
14066 receive a message for every recursive call. If the region contains 13
14067 words, we will receive thirteen messages, one right after the other.
14068 We don't want this! Instead, we must write two functions to do the
14069 job, one of which (the recursive function) will be used inside of the
14070 other. One function will set up the conditions and display the
14071 message; the other will return the word count.
14073 Let us start with the function that causes the message to be displayed.
14074 We can continue to call this @code{@value{COUNT-WORDS}}.
14076 This is the function that the user will call. It will be interactive.
14077 Indeed, it will be similar to our previous versions of this
14078 function, except that it will call @code{recursive-count-words} to
14079 determine how many words are in the region.
14082 We can readily construct a template for this function, based on our
14087 ;; @r{Recursive version; uses regular expression search}
14088 (defun @value{COUNT-WORDS} (beginning end)
14089 "@var{documentation}@dots{}"
14090 (@var{interactive-expression}@dots{})
14094 ;;; @r{1. Set up appropriate conditions.}
14095 (@var{explanatory message})
14096 (@var{set-up functions}@dots{}
14100 ;;; @r{2. Count the words.}
14101 @var{recursive call}
14105 ;;; @r{3. Send a message to the user.}
14106 @var{message providing word count}))
14110 The definition looks straightforward, except that somehow the count
14111 returned by the recursive call must be passed to the message
14112 displaying the word count. A little thought suggests that this can be
14113 done by making use of a @code{let} expression: we can bind a variable
14114 in the varlist of a @code{let} expression to the number of words in
14115 the region, as returned by the recursive call; and then the
14116 @code{cond} expression, using binding, can display the value to the
14119 Often, one thinks of the binding within a @code{let} expression as
14120 somehow secondary to the `primary' work of a function. But in this
14121 case, what you might consider the `primary' job of the function,
14122 counting words, is done within the @code{let} expression.
14125 Using @code{let}, the function definition looks like this:
14129 (defun @value{COUNT-WORDS} (beginning end)
14130 "Print number of words in the region."
14135 ;;; @r{1. Set up appropriate conditions.}
14136 (message "Counting words in region ... ")
14138 (goto-char beginning)
14142 ;;; @r{2. Count the words.}
14143 (let ((count (recursive-count-words end)))
14147 ;;; @r{3. Send a message to the user.}
14148 (cond ((zerop count)
14150 "The region does NOT have any words."))
14153 "The region has 1 word."))
14156 "The region has %d words." count))))))
14160 Next, we need to write the recursive counting function.
14162 A recursive function has at least three parts: the `do-again-test', the
14163 `next-step-expression', and the recursive call.
14165 The do-again-test determines whether the function will or will not be
14166 called again. Since we are counting words in a region and can use a
14167 function that moves point forward for every word, the do-again-test
14168 can check whether point is still within the region. The do-again-test
14169 should find the value of point and determine whether point is before,
14170 at, or after the value of the end of the region. We can use the
14171 @code{point} function to locate point. Clearly, we must pass the
14172 value of the end of the region to the recursive counting function as an
14175 In addition, the do-again-test should also test whether the search finds a
14176 word. If it does not, the function should not call itself again.
14178 The next-step-expression changes a value so that when the recursive
14179 function is supposed to stop calling itself, it stops. More
14180 precisely, the next-step-expression changes a value so that at the
14181 right time, the do-again-test stops the recursive function from
14182 calling itself again. In this case, the next-step-expression can be
14183 the expression that moves point forward, word by word.
14185 The third part of a recursive function is the recursive call.
14187 Somewhere, also, we also need a part that does the `work' of the
14188 function, a part that does the counting. A vital part!
14191 But already, we have an outline of the recursive counting function:
14195 (defun recursive-count-words (region-end)
14196 "@var{documentation}@dots{}"
14197 @var{do-again-test}
14198 @var{next-step-expression}
14199 @var{recursive call})
14203 Now we need to fill in the slots. Let's start with the simplest cases
14204 first: if point is at or beyond the end of the region, there cannot
14205 be any words in the region, so the function should return zero.
14206 Likewise, if the search fails, there are no words to count, so the
14207 function should return zero.
14209 On the other hand, if point is within the region and the search
14210 succeeds, the function should call itself again.
14213 Thus, the do-again-test should look like this:
14217 (and (< (point) region-end)
14218 (re-search-forward "\\w+\\W*" region-end t))
14222 Note that the search expression is part of the do-again-test---the
14223 function returns @code{t} if its search succeeds and @code{nil} if it
14224 fails. (@xref{Whitespace Bug, , The Whitespace Bug in
14225 @code{@value{COUNT-WORDS}}}, for an explanation of how
14226 @code{re-search-forward} works.)
14228 The do-again-test is the true-or-false test of an @code{if} clause.
14229 Clearly, if the do-again-test succeeds, the then-part of the @code{if}
14230 clause should call the function again; but if it fails, the else-part
14231 should return zero since either point is outside the region or the
14232 search failed because there were no words to find.
14234 But before considering the recursive call, we need to consider the
14235 next-step-expression. What is it? Interestingly, it is the search
14236 part of the do-again-test.
14238 In addition to returning @code{t} or @code{nil} for the
14239 do-again-test, @code{re-search-forward} moves point forward as a side
14240 effect of a successful search. This is the action that changes the
14241 value of point so that the recursive function stops calling itself
14242 when point completes its movement through the region. Consequently,
14243 the @code{re-search-forward} expression is the next-step-expression.
14246 In outline, then, the body of the @code{recursive-count-words}
14247 function looks like this:
14251 (if @var{do-again-test-and-next-step-combined}
14253 @var{recursive-call-returning-count}
14259 How to incorporate the mechanism that counts?
14261 If you are not used to writing recursive functions, a question like
14262 this can be troublesome. But it can and should be approached
14265 We know that the counting mechanism should be associated in some way
14266 with the recursive call. Indeed, since the next-step-expression moves
14267 point forward by one word, and since a recursive call is made for
14268 each word, the counting mechanism must be an expression that adds one
14269 to the value returned by a call to @code{recursive-count-words}.
14272 Consider several cases:
14276 If there are two words in the region, the function should return
14277 a value resulting from adding one to the value returned when it counts
14278 the first word, plus the number returned when it counts the remaining
14279 words in the region, which in this case is one.
14282 If there is one word in the region, the function should return
14283 a value resulting from adding one to the value returned when it counts
14284 that word, plus the number returned when it counts the remaining
14285 words in the region, which in this case is zero.
14288 If there are no words in the region, the function should return zero.
14291 From the sketch we can see that the else-part of the @code{if} returns
14292 zero for the case of no words. This means that the then-part of the
14293 @code{if} must return a value resulting from adding one to the value
14294 returned from a count of the remaining words.
14297 The expression will look like this, where @code{1+} is a function that
14298 adds one to its argument.
14301 (1+ (recursive-count-words region-end))
14305 The whole @code{recursive-count-words} function will then look like
14310 (defun recursive-count-words (region-end)
14311 "@var{documentation}@dots{}"
14313 ;;; @r{1. do-again-test}
14314 (if (and (< (point) region-end)
14315 (re-search-forward "\\w+\\W*" region-end t))
14319 ;;; @r{2. then-part: the recursive call}
14320 (1+ (recursive-count-words region-end))
14322 ;;; @r{3. else-part}
14328 Let's examine how this works:
14330 If there are no words in the region, the else part of the @code{if}
14331 expression is evaluated and consequently the function returns zero.
14333 If there is one word in the region, the value of point is less than
14334 the value of @code{region-end} and the search succeeds. In this case,
14335 the true-or-false-test of the @code{if} expression tests true, and the
14336 then-part of the @code{if} expression is evaluated. The counting
14337 expression is evaluated. This expression returns a value (which will
14338 be the value returned by the whole function) that is the sum of one
14339 added to the value returned by a recursive call.
14341 Meanwhile, the next-step-expression has caused point to jump over the
14342 first (and in this case only) word in the region. This means that
14343 when @code{(recursive-count-words region-end)} is evaluated a second
14344 time, as a result of the recursive call, the value of point will be
14345 equal to or greater than the value of region end. So this time,
14346 @code{recursive-count-words} will return zero. The zero will be added
14347 to one, and the original evaluation of @code{recursive-count-words}
14348 will return one plus zero, which is one, which is the correct amount.
14350 Clearly, if there are two words in the region, the first call to
14351 @code{recursive-count-words} returns one added to the value returned
14352 by calling @code{recursive-count-words} on a region containing the
14353 remaining word---that is, it adds one to one, producing two, which is
14354 the correct amount.
14356 Similarly, if there are three words in the region, the first call to
14357 @code{recursive-count-words} returns one added to the value returned
14358 by calling @code{recursive-count-words} on a region containing the
14359 remaining two words---and so on and so on.
14363 With full documentation the two functions look like this:
14367 The recursive function:
14369 @findex recursive-count-words
14372 (defun recursive-count-words (region-end)
14373 "Number of words between point and REGION-END."
14377 ;;; @r{1. do-again-test}
14378 (if (and (< (point) region-end)
14379 (re-search-forward "\\w+\\W*" region-end t))
14383 ;;; @r{2. then-part: the recursive call}
14384 (1+ (recursive-count-words region-end))
14386 ;;; @r{3. else-part}
14397 ;;; @r{Recursive version}
14398 (defun @value{COUNT-WORDS} (beginning end)
14399 "Print number of words in the region.
14403 Words are defined as at least one word-constituent
14404 character followed by at least one character that is
14405 not a word-constituent. The buffer's syntax table
14406 determines which characters these are."
14410 (message "Counting words in region ... ")
14412 (goto-char beginning)
14413 (let ((count (recursive-count-words end)))
14416 (cond ((zerop count)
14418 "The region does NOT have any words."))
14422 (message "The region has 1 word."))
14425 "The region has %d words." count))))))
14429 @node Counting Exercise
14430 @section Exercise: Counting Punctuation
14432 Using a @code{while} loop, write a function to count the number of
14433 punctuation marks in a region---period, comma, semicolon, colon,
14434 exclamation mark, and question mark. Do the same using recursion.
14436 @node Words in a defun
14437 @chapter Counting Words in a @code{defun}
14438 @cindex Counting words in a @code{defun}
14439 @cindex Word counting in a @code{defun}
14441 Our next project is to count the number of words in a function
14442 definition. Clearly, this can be done using some variant of
14443 @code{@value{COUNT-WORDS}}. @xref{Counting Words, , Counting Words:
14444 Repetition and Regexps}. If we are just going to count the words in
14445 one definition, it is easy enough to mark the definition with the
14446 @kbd{C-M-h} (@code{mark-defun}) command, and then call
14447 @code{@value{COUNT-WORDS}}.
14449 However, I am more ambitious: I want to count the words and symbols in
14450 every definition in the Emacs sources and then print a graph that
14451 shows how many functions there are of each length: how many contain 40
14452 to 49 words or symbols, how many contain 50 to 59 words or symbols,
14453 and so on. I have often been curious how long a typical function is,
14454 and this will tell.
14457 * Divide and Conquer::
14458 * Words and Symbols:: What to count?
14459 * Syntax:: What constitutes a word or symbol?
14460 * count-words-in-defun:: Very like @code{@value{COUNT-WORDS}}.
14461 * Several defuns:: Counting several defuns in a file.
14462 * Find a File:: Do you want to look at a file?
14463 * lengths-list-file:: A list of the lengths of many definitions.
14464 * Several files:: Counting in definitions in different files.
14465 * Several files recursively:: Recursively counting in different files.
14466 * Prepare the data:: Prepare the data for display in a graph.
14470 @node Divide and Conquer
14471 @unnumberedsec Divide and Conquer
14474 Described in one phrase, the histogram project is daunting; but
14475 divided into numerous small steps, each of which we can take one at a
14476 time, the project becomes less fearsome. Let us consider what the
14481 First, write a function to count the words in one definition. This
14482 includes the problem of handling symbols as well as words.
14485 Second, write a function to list the numbers of words in each function
14486 in a file. This function can use the @code{count-words-in-defun}
14490 Third, write a function to list the numbers of words in each function
14491 in each of several files. This entails automatically finding the
14492 various files, switching to them, and counting the words in the
14493 definitions within them.
14496 Fourth, write a function to convert the list of numbers that we
14497 created in step three to a form that will be suitable for printing as
14501 Fifth, write a function to print the results as a graph.
14504 This is quite a project! But if we take each step slowly, it will not
14507 @node Words and Symbols
14508 @section What to Count?
14509 @cindex Words and symbols in defun
14511 When we first start thinking about how to count the words in a
14512 function definition, the first question is (or ought to be) what are
14513 we going to count? When we speak of `words' with respect to a Lisp
14514 function definition, we are actually speaking, in large part, of
14515 `symbols'. For example, the following @code{multiply-by-seven}
14516 function contains the five symbols @code{defun},
14517 @code{multiply-by-seven}, @code{number}, @code{*}, and @code{7}. In
14518 addition, in the documentation string, it contains the four words
14519 @samp{Multiply}, @samp{NUMBER}, @samp{by}, and @samp{seven}. The
14520 symbol @samp{number} is repeated, so the definition contains a total
14521 of ten words and symbols.
14525 (defun multiply-by-seven (number)
14526 "Multiply NUMBER by seven."
14532 However, if we mark the @code{multiply-by-seven} definition with
14533 @kbd{C-M-h} (@code{mark-defun}), and then call
14534 @code{@value{COUNT-WORDS}} on it, we will find that
14535 @code{@value{COUNT-WORDS}} claims the definition has eleven words, not
14536 ten! Something is wrong!
14538 The problem is twofold: @code{@value{COUNT-WORDS}} does not count the
14539 @samp{*} as a word, and it counts the single symbol,
14540 @code{multiply-by-seven}, as containing three words. The hyphens are
14541 treated as if they were interword spaces rather than intraword
14542 connectors: @samp{multiply-by-seven} is counted as if it were written
14543 @samp{multiply by seven}.
14545 The cause of this confusion is the regular expression search within
14546 the @code{@value{COUNT-WORDS}} definition that moves point forward word
14547 by word. In the canonical version of @code{@value{COUNT-WORDS}}, the
14555 This regular expression is a pattern defining one or more word
14556 constituent characters possibly followed by one or more characters
14557 that are not word constituents. What is meant by `word constituent
14558 characters' brings us to the issue of syntax, which is worth a section
14562 @section What Constitutes a Word or Symbol?
14563 @cindex Syntax categories and tables
14565 Emacs treats different characters as belonging to different
14566 @dfn{syntax categories}. For example, the regular expression,
14567 @samp{\\w+}, is a pattern specifying one or more @emph{word
14568 constituent} characters. Word constituent characters are members of
14569 one syntax category. Other syntax categories include the class of
14570 punctuation characters, such as the period and the comma, and the
14571 class of whitespace characters, such as the blank space and the tab
14572 character. (For more information, @pxref{Syntax Tables, , Syntax
14573 Tables, elisp, The GNU Emacs Lisp Reference Manual}.)
14575 Syntax tables specify which characters belong to which categories.
14576 Usually, a hyphen is not specified as a `word constituent character'.
14577 Instead, it is specified as being in the `class of characters that are
14578 part of symbol names but not words.' This means that the
14579 @code{@value{COUNT-WORDS}} function treats it in the same way it treats
14580 an interword white space, which is why @code{@value{COUNT-WORDS}}
14581 counts @samp{multiply-by-seven} as three words.
14583 There are two ways to cause Emacs to count @samp{multiply-by-seven} as
14584 one symbol: modify the syntax table or modify the regular expression.
14586 We could redefine a hyphen as a word constituent character by
14587 modifying the syntax table that Emacs keeps for each mode. This
14588 action would serve our purpose, except that a hyphen is merely the
14589 most common character within symbols that is not typically a word
14590 constituent character; there are others, too.
14592 Alternatively, we can redefine the regexp used in the
14593 @code{@value{COUNT-WORDS}} definition so as to include symbols. This
14594 procedure has the merit of clarity, but the task is a little tricky.
14597 The first part is simple enough: the pattern must match ``at least one
14598 character that is a word or symbol constituent''. Thus:
14601 "\\(\\w\\|\\s_\\)+"
14605 The @samp{\\(} is the first part of the grouping construct that
14606 includes the @samp{\\w} and the @samp{\\s_} as alternatives, separated
14607 by the @samp{\\|}. The @samp{\\w} matches any word-constituent
14608 character and the @samp{\\s_} matches any character that is part of a
14609 symbol name but not a word-constituent character. The @samp{+}
14610 following the group indicates that the word or symbol constituent
14611 characters must be matched at least once.
14613 However, the second part of the regexp is more difficult to design.
14614 What we want is to follow the first part with ``optionally one or more
14615 characters that are not constituents of a word or symbol''. At first,
14616 I thought I could define this with the following:
14619 "\\(\\W\\|\\S_\\)*"
14623 The upper case @samp{W} and @samp{S} match characters that are
14624 @emph{not} word or symbol constituents. Unfortunately, this
14625 expression matches any character that is either not a word constituent
14626 or not a symbol constituent. This matches any character!
14628 I then noticed that every word or symbol in my test region was
14629 followed by white space (blank space, tab, or newline). So I tried
14630 placing a pattern to match one or more blank spaces after the pattern
14631 for one or more word or symbol constituents. This failed, too. Words
14632 and symbols are often separated by whitespace, but in actual code
14633 parentheses may follow symbols and punctuation may follow words. So
14634 finally, I designed a pattern in which the word or symbol constituents
14635 are followed optionally by characters that are not white space and
14636 then followed optionally by white space.
14639 Here is the full regular expression:
14642 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
14645 @node count-words-in-defun
14646 @section The @code{count-words-in-defun} Function
14647 @cindex Counting words in a @code{defun}
14649 We have seen that there are several ways to write a
14650 @code{count-words-region} function. To write a
14651 @code{count-words-in-defun}, we need merely adapt one of these
14654 The version that uses a @code{while} loop is easy to understand, so I
14655 am going to adapt that. Because @code{count-words-in-defun} will be
14656 part of a more complex program, it need not be interactive and it need
14657 not display a message but just return the count. These considerations
14658 simplify the definition a little.
14660 On the other hand, @code{count-words-in-defun} will be used within a
14661 buffer that contains function definitions. Consequently, it is
14662 reasonable to ask that the function determine whether it is called
14663 when point is within a function definition, and if it is, to return
14664 the count for that definition. This adds complexity to the
14665 definition, but saves us from needing to pass arguments to the
14669 These considerations lead us to prepare the following template:
14673 (defun count-words-in-defun ()
14674 "@var{documentation}@dots{}"
14675 (@var{set up}@dots{}
14676 (@var{while loop}@dots{})
14677 @var{return count})
14682 As usual, our job is to fill in the slots.
14686 We are presuming that this function will be called within a buffer
14687 containing function definitions. Point will either be within a
14688 function definition or not. For @code{count-words-in-defun} to work,
14689 point must move to the beginning of the definition, a counter must
14690 start at zero, and the counting loop must stop when point reaches the
14691 end of the definition.
14693 The @code{beginning-of-defun} function searches backwards for an
14694 opening delimiter such as a @samp{(} at the beginning of a line, and
14695 moves point to that position, or else to the limit of the search. In
14696 practice, this means that @code{beginning-of-defun} moves point to the
14697 beginning of an enclosing or preceding function definition, or else to
14698 the beginning of the buffer. We can use @code{beginning-of-defun} to
14699 place point where we wish to start.
14701 The @code{while} loop requires a counter to keep track of the words or
14702 symbols being counted. A @code{let} expression can be used to create
14703 a local variable for this purpose, and bind it to an initial value of zero.
14705 The @code{end-of-defun} function works like @code{beginning-of-defun}
14706 except that it moves point to the end of the definition.
14707 @code{end-of-defun} can be used as part of an expression that
14708 determines the position of the end of the definition.
14710 The set up for @code{count-words-in-defun} takes shape rapidly: first
14711 we move point to the beginning of the definition, then we create a
14712 local variable to hold the count, and finally, we record the position
14713 of the end of the definition so the @code{while} loop will know when to stop
14717 The code looks like this:
14721 (beginning-of-defun)
14723 (end (save-excursion (end-of-defun) (point))))
14728 The code is simple. The only slight complication is likely to concern
14729 @code{end}: it is bound to the position of the end of the definition
14730 by a @code{save-excursion} expression that returns the value of point
14731 after @code{end-of-defun} temporarily moves it to the end of the
14734 The second part of the @code{count-words-in-defun}, after the set up,
14735 is the @code{while} loop.
14737 The loop must contain an expression that jumps point forward word by
14738 word and symbol by symbol, and another expression that counts the
14739 jumps. The true-or-false-test for the @code{while} loop should test
14740 true so long as point should jump forward, and false when point is at
14741 the end of the definition. We have already redefined the regular
14742 expression for this, so the loop is straightforward:
14746 (while (and (< (point) end)
14748 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*" end t))
14749 (setq count (1+ count)))
14753 The third part of the function definition returns the count of words
14754 and symbols. This part is the last expression within the body of the
14755 @code{let} expression, and can be, very simply, the local variable
14756 @code{count}, which when evaluated returns the count.
14759 Put together, the @code{count-words-in-defun} definition looks like this:
14761 @findex count-words-in-defun
14764 (defun count-words-in-defun ()
14765 "Return the number of words and symbols in a defun."
14766 (beginning-of-defun)
14768 (end (save-excursion (end-of-defun) (point))))
14772 (and (< (point) end)
14774 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
14776 (setq count (1+ count)))
14781 How to test this? The function is not interactive, but it is easy to
14782 put a wrapper around the function to make it interactive; we can use
14783 almost the same code as for the recursive version of
14784 @code{@value{COUNT-WORDS}}:
14788 ;;; @r{Interactive version.}
14789 (defun count-words-defun ()
14790 "Number of words and symbols in a function definition."
14793 "Counting words and symbols in function definition ... ")
14796 (let ((count (count-words-in-defun)))
14800 "The definition does NOT have any words or symbols."))
14805 "The definition has 1 word or symbol."))
14808 "The definition has %d words or symbols." count)))))
14814 Let's re-use @kbd{C-c =} as a convenient keybinding:
14817 (global-set-key "\C-c=" 'count-words-defun)
14820 Now we can try out @code{count-words-defun}: install both
14821 @code{count-words-in-defun} and @code{count-words-defun}, and set the
14822 keybinding, and then place the cursor within the following definition:
14826 (defun multiply-by-seven (number)
14827 "Multiply NUMBER by seven."
14834 Success! The definition has 10 words and symbols.
14836 The next problem is to count the numbers of words and symbols in
14837 several definitions within a single file.
14839 @node Several defuns
14840 @section Count Several @code{defuns} Within a File
14842 A file such as @file{simple.el} may have a hundred or more function
14843 definitions within it. Our long term goal is to collect statistics on
14844 many files, but as a first step, our immediate goal is to collect
14845 statistics on one file.
14847 The information will be a series of numbers, each number being the
14848 length of a function definition. We can store the numbers in a list.
14850 We know that we will want to incorporate the information regarding one
14851 file with information about many other files; this means that the
14852 function for counting definition lengths within one file need only
14853 return the list of lengths. It need not and should not display any
14856 The word count commands contain one expression to jump point forward
14857 word by word and another expression to count the jumps. The function
14858 to return the lengths of definitions can be designed to work the same
14859 way, with one expression to jump point forward definition by
14860 definition and another expression to construct the lengths' list.
14862 This statement of the problem makes it elementary to write the
14863 function definition. Clearly, we will start the count at the
14864 beginning of the file, so the first command will be @code{(goto-char
14865 (point-min))}. Next, we start the @code{while} loop; and the
14866 true-or-false test of the loop can be a regular expression search for
14867 the next function definition---so long as the search succeeds, point
14868 is moved forward and then the body of the loop is evaluated. The body
14869 needs an expression that constructs the lengths' list. @code{cons},
14870 the list construction command, can be used to create the list. That
14871 is almost all there is to it.
14874 Here is what this fragment of code looks like:
14878 (goto-char (point-min))
14879 (while (re-search-forward "^(defun" nil t)
14881 (cons (count-words-in-defun) lengths-list)))
14885 What we have left out is the mechanism for finding the file that
14886 contains the function definitions.
14888 In previous examples, we either used this, the Info file, or we
14889 switched back and forth to some other buffer, such as the
14890 @file{*scratch*} buffer.
14892 Finding a file is a new process that we have not yet discussed.
14895 @section Find a File
14896 @cindex Find a File
14898 To find a file in Emacs, you use the @kbd{C-x C-f} (@code{find-file})
14899 command. This command is almost, but not quite right for the lengths
14903 Let's look at the source for @code{find-file}:
14907 (defun find-file (filename)
14908 "Edit file FILENAME.
14909 Switch to a buffer visiting file FILENAME,
14910 creating one if none already exists."
14911 (interactive "FFind file: ")
14912 (switch-to-buffer (find-file-noselect filename)))
14917 (The most recent version of the @code{find-file} function definition
14918 permits you to specify optional wildcards to visit multiple files; that
14919 makes the definition more complex and we will not discuss it here,
14920 since it is not relevant. You can see its source using either
14921 @kbd{M-.} (@code{find-tag}) or @kbd{C-h f} (@code{describe-function}).)
14925 (defun find-file (filename &optional wildcards)
14926 "Edit file FILENAME.
14927 Switch to a buffer visiting file FILENAME,
14928 creating one if none already exists.
14929 Interactively, the default if you just type RET is the current directory,
14930 but the visited file name is available through the minibuffer history:
14931 type M-n to pull it into the minibuffer.
14933 Interactively, or if WILDCARDS is non-nil in a call from Lisp,
14934 expand wildcards (if any) and visit multiple files. You can
14935 suppress wildcard expansion by setting `find-file-wildcards' to nil.
14937 To visit a file without any kind of conversion and without
14938 automatically choosing a major mode, use \\[find-file-literally]."
14939 (interactive (find-file-read-args "Find file: " nil))
14940 (let ((value (find-file-noselect filename nil nil wildcards)))
14942 (mapcar 'switch-to-buffer (nreverse value))
14943 (switch-to-buffer value))))
14946 The definition I am showing possesses short but complete documentation
14947 and an interactive specification that prompts you for a file name when
14948 you use the command interactively. The body of the definition
14949 contains two functions, @code{find-file-noselect} and
14950 @code{switch-to-buffer}.
14952 According to its documentation as shown by @kbd{C-h f} (the
14953 @code{describe-function} command), the @code{find-file-noselect}
14954 function reads the named file into a buffer and returns the buffer.
14955 (Its most recent version includes an optional wildcards argument,
14956 too, as well as another to read a file literally and an other you
14957 suppress warning messages. These optional arguments are irrelevant.)
14959 However, the @code{find-file-noselect} function does not select the
14960 buffer in which it puts the file. Emacs does not switch its attention
14961 (or yours if you are using @code{find-file-noselect}) to the selected
14962 buffer. That is what @code{switch-to-buffer} does: it switches the
14963 buffer to which Emacs attention is directed; and it switches the
14964 buffer displayed in the window to the new buffer. We have discussed
14965 buffer switching elsewhere. (@xref{Switching Buffers}.)
14967 In this histogram project, we do not need to display each file on the
14968 screen as the program determines the length of each definition within
14969 it. Instead of employing @code{switch-to-buffer}, we can work with
14970 @code{set-buffer}, which redirects the attention of the computer
14971 program to a different buffer but does not redisplay it on the screen.
14972 So instead of calling on @code{find-file} to do the job, we must write
14973 our own expression.
14975 The task is easy: use @code{find-file-noselect} and @code{set-buffer}.
14977 @node lengths-list-file
14978 @section @code{lengths-list-file} in Detail
14980 The core of the @code{lengths-list-file} function is a @code{while}
14981 loop containing a function to move point forward `defun by defun' and
14982 a function to count the number of words and symbols in each defun.
14983 This core must be surrounded by functions that do various other tasks,
14984 including finding the file, and ensuring that point starts out at the
14985 beginning of the file. The function definition looks like this:
14986 @findex lengths-list-file
14990 (defun lengths-list-file (filename)
14991 "Return list of definitions' lengths within FILE.
14992 The returned list is a list of numbers.
14993 Each number is the number of words or
14994 symbols in one function definition."
14997 (message "Working on `%s' ... " filename)
14999 (let ((buffer (find-file-noselect filename))
15001 (set-buffer buffer)
15002 (setq buffer-read-only t)
15004 (goto-char (point-min))
15005 (while (re-search-forward "^(defun" nil t)
15007 (cons (count-words-in-defun) lengths-list)))
15008 (kill-buffer buffer)
15014 The function is passed one argument, the name of the file on which it
15015 will work. It has four lines of documentation, but no interactive
15016 specification. Since people worry that a computer is broken if they
15017 don't see anything going on, the first line of the body is a
15020 The next line contains a @code{save-excursion} that returns Emacs's
15021 attention to the current buffer when the function completes. This is
15022 useful in case you embed this function in another function that
15023 presumes point is restored to the original buffer.
15025 In the varlist of the @code{let} expression, Emacs finds the file and
15026 binds the local variable @code{buffer} to the buffer containing the
15027 file. At the same time, Emacs creates @code{lengths-list} as a local
15030 Next, Emacs switches its attention to the buffer.
15032 In the following line, Emacs makes the buffer read-only. Ideally,
15033 this line is not necessary. None of the functions for counting words
15034 and symbols in a function definition should change the buffer.
15035 Besides, the buffer is not going to be saved, even if it were changed.
15036 This line is entirely the consequence of great, perhaps excessive,
15037 caution. The reason for the caution is that this function and those
15038 it calls work on the sources for Emacs and it is inconvenient if they
15039 are inadvertently modified. It goes without saying that I did not
15040 realize a need for this line until an experiment went awry and started
15041 to modify my Emacs source files @dots{}
15043 Next comes a call to widen the buffer if it is narrowed. This
15044 function is usually not needed---Emacs creates a fresh buffer if none
15045 already exists; but if a buffer visiting the file already exists Emacs
15046 returns that one. In this case, the buffer may be narrowed and must
15047 be widened. If we wanted to be fully `user-friendly', we would
15048 arrange to save the restriction and the location of point, but we
15051 The @code{(goto-char (point-min))} expression moves point to the
15052 beginning of the buffer.
15054 Then comes a @code{while} loop in which the `work' of the function is
15055 carried out. In the loop, Emacs determines the length of each
15056 definition and constructs a lengths' list containing the information.
15058 Emacs kills the buffer after working through it. This is to save
15059 space inside of Emacs. My version of GNU Emacs 19 contained over 300
15060 source files of interest; GNU Emacs 22 contains over a thousand source
15061 files. Another function will apply @code{lengths-list-file} to each
15064 Finally, the last expression within the @code{let} expression is the
15065 @code{lengths-list} variable; its value is returned as the value of
15066 the whole function.
15068 You can try this function by installing it in the usual fashion. Then
15069 place your cursor after the following expression and type @kbd{C-x
15070 C-e} (@code{eval-last-sexp}).
15072 @c !!! 22.1.1 lisp sources location here
15075 "/usr/local/share/emacs/22.1.1/lisp/emacs-lisp/debug.el")
15079 (You may need to change the pathname of the file; the one here is for
15080 GNU Emacs version 22.1.1. To change the expression, copy it to
15081 the @file{*scratch*} buffer and edit it.
15085 (Also, to see the full length of the list, rather than a truncated
15086 version, you may have to evaluate the following:
15089 (custom-set-variables '(eval-expression-print-length nil))
15093 (@xref{defcustom, , Specifying Variables using @code{defcustom}}.
15094 Then evaluate the @code{lengths-list-file} expression.)
15097 The lengths' list for @file{debug.el} takes less than a second to
15098 produce and looks like this in GNU Emacs 22:
15101 (83 113 105 144 289 22 30 97 48 89 25 52 52 88 28 29 77 49 43 290 232 587)
15105 (Using my old machine, the version 19 lengths' list for @file{debug.el}
15106 took seven seconds to produce and looked like this:
15109 (75 41 80 62 20 45 44 68 45 12 34 235)
15112 (The newer version of @file{debug.el} contains more defuns than the
15113 earlier one; and my new machine is much faster than the old one.)
15115 Note that the length of the last definition in the file is first in
15118 @node Several files
15119 @section Count Words in @code{defuns} in Different Files
15121 In the previous section, we created a function that returns a list of
15122 the lengths of each definition in a file. Now, we want to define a
15123 function to return a master list of the lengths of the definitions in
15126 Working on each of a list of files is a repetitious act, so we can use
15127 either a @code{while} loop or recursion.
15130 * lengths-list-many-files:: Return a list of the lengths of defuns.
15131 * append:: Attach one list to another.
15135 @node lengths-list-many-files
15136 @unnumberedsubsec Determine the lengths of @code{defuns}
15139 The design using a @code{while} loop is routine. The argument passed
15140 the function is a list of files. As we saw earlier (@pxref{Loop
15141 Example}), you can write a @code{while} loop so that the body of the
15142 loop is evaluated if such a list contains elements, but to exit the
15143 loop if the list is empty. For this design to work, the body of the
15144 loop must contain an expression that shortens the list each time the
15145 body is evaluated, so that eventually the list is empty. The usual
15146 technique is to set the value of the list to the value of the @sc{cdr}
15147 of the list each time the body is evaluated.
15150 The template looks like this:
15154 (while @var{test-whether-list-is-empty}
15156 @var{set-list-to-cdr-of-list})
15160 Also, we remember that a @code{while} loop returns @code{nil} (the
15161 result of evaluating the true-or-false-test), not the result of any
15162 evaluation within its body. (The evaluations within the body of the
15163 loop are done for their side effects.) However, the expression that
15164 sets the lengths' list is part of the body---and that is the value
15165 that we want returned by the function as a whole. To do this, we
15166 enclose the @code{while} loop within a @code{let} expression, and
15167 arrange that the last element of the @code{let} expression contains
15168 the value of the lengths' list. (@xref{Incrementing Example, , Loop
15169 Example with an Incrementing Counter}.)
15171 @findex lengths-list-many-files
15173 These considerations lead us directly to the function itself:
15177 ;;; @r{Use @code{while} loop.}
15178 (defun lengths-list-many-files (list-of-files)
15179 "Return list of lengths of defuns in LIST-OF-FILES."
15182 (let (lengths-list)
15184 ;;; @r{true-or-false-test}
15185 (while list-of-files
15190 ;;; @r{Generate a lengths' list.}
15192 (expand-file-name (car list-of-files)))))
15196 ;;; @r{Make files' list shorter.}
15197 (setq list-of-files (cdr list-of-files)))
15199 ;;; @r{Return final value of lengths' list.}
15204 @code{expand-file-name} is a built-in function that converts a file
15205 name to the absolute, long, path name form. The function employs the
15206 name of the directory in which the function is called.
15208 @c !!! 22.1.1 lisp sources location here
15210 Thus, if @code{expand-file-name} is called on @code{debug.el} when
15211 Emacs is visiting the
15212 @file{/usr/local/share/emacs/22.1.1/lisp/emacs-lisp/} directory,
15222 @c !!! 22.1.1 lisp sources location here
15224 /usr/local/share/emacs/22.1.1/lisp/emacs-lisp/debug.el
15227 The only other new element of this function definition is the as yet
15228 unstudied function @code{append}, which merits a short section for
15232 @subsection The @code{append} Function
15235 The @code{append} function attaches one list to another. Thus,
15238 (append '(1 2 3 4) '(5 6 7 8))
15249 This is exactly how we want to attach two lengths' lists produced by
15250 @code{lengths-list-file} to each other. The results contrast with
15254 (cons '(1 2 3 4) '(5 6 7 8))
15259 which constructs a new list in which the first argument to @code{cons}
15260 becomes the first element of the new list:
15263 ((1 2 3 4) 5 6 7 8)
15266 @node Several files recursively
15267 @section Recursively Count Words in Different Files
15269 Besides a @code{while} loop, you can work on each of a list of files
15270 with recursion. A recursive version of @code{lengths-list-many-files}
15271 is short and simple.
15273 The recursive function has the usual parts: the `do-again-test', the
15274 `next-step-expression', and the recursive call. The `do-again-test'
15275 determines whether the function should call itself again, which it
15276 will do if the @code{list-of-files} contains any remaining elements;
15277 the `next-step-expression' resets the @code{list-of-files} to the
15278 @sc{cdr} of itself, so eventually the list will be empty; and the
15279 recursive call calls itself on the shorter list. The complete
15280 function is shorter than this description!
15281 @findex recursive-lengths-list-many-files
15285 (defun recursive-lengths-list-many-files (list-of-files)
15286 "Return list of lengths of each defun in LIST-OF-FILES."
15287 (if list-of-files ; @r{do-again-test}
15290 (expand-file-name (car list-of-files)))
15291 (recursive-lengths-list-many-files
15292 (cdr list-of-files)))))
15297 In a sentence, the function returns the lengths' list for the first of
15298 the @code{list-of-files} appended to the result of calling itself on
15299 the rest of the @code{list-of-files}.
15301 Here is a test of @code{recursive-lengths-list-many-files}, along with
15302 the results of running @code{lengths-list-file} on each of the files
15305 Install @code{recursive-lengths-list-many-files} and
15306 @code{lengths-list-file}, if necessary, and then evaluate the
15307 following expressions. You may need to change the files' pathnames;
15308 those here work when this Info file and the Emacs sources are located
15309 in their customary places. To change the expressions, copy them to
15310 the @file{*scratch*} buffer, edit them, and then evaluate them.
15312 The results are shown after the @samp{@result{}}. (These results are
15313 for files from Emacs version 22.1.1; files from other versions of
15314 Emacs may produce different results.)
15316 @c !!! 22.1.1 lisp sources location here
15319 (cd "/usr/local/share/emacs/22.1.1/")
15321 (lengths-list-file "./lisp/macros.el")
15322 @result{} (283 263 480 90)
15326 (lengths-list-file "./lisp/mail/mailalias.el")
15327 @result{} (38 32 29 95 178 180 321 218 324)
15331 (lengths-list-file "./lisp/makesum.el")
15336 (recursive-lengths-list-many-files
15337 '("./lisp/macros.el"
15338 "./lisp/mail/mailalias.el"
15339 "./lisp/makesum.el"))
15340 @result{} (283 263 480 90 38 32 29 95 178 180 321 218 324 85 181)
15344 The @code{recursive-lengths-list-many-files} function produces the
15347 The next step is to prepare the data in the list for display in a graph.
15349 @node Prepare the data
15350 @section Prepare the Data for Display in a Graph
15352 The @code{recursive-lengths-list-many-files} function returns a list
15353 of numbers. Each number records the length of a function definition.
15354 What we need to do now is transform this data into a list of numbers
15355 suitable for generating a graph. The new list will tell how many
15356 functions definitions contain less than 10 words and
15357 symbols, how many contain between 10 and 19 words and symbols, how
15358 many contain between 20 and 29 words and symbols, and so on.
15360 In brief, we need to go through the lengths' list produced by the
15361 @code{recursive-lengths-list-many-files} function and count the number
15362 of defuns within each range of lengths, and produce a list of those
15366 * Data for Display in Detail::
15367 * Sorting:: Sorting lists.
15368 * Files List:: Making a list of files.
15369 * Counting function definitions::
15373 @node Data for Display in Detail
15374 @unnumberedsubsec The Data for Display in Detail
15377 Based on what we have done before, we can readily foresee that it
15378 should not be too hard to write a function that `@sc{cdr}s' down the
15379 lengths' list, looks at each element, determines which length range it
15380 is in, and increments a counter for that range.
15382 However, before beginning to write such a function, we should consider
15383 the advantages of sorting the lengths' list first, so the numbers are
15384 ordered from smallest to largest. First, sorting will make it easier
15385 to count the numbers in each range, since two adjacent numbers will
15386 either be in the same length range or in adjacent ranges. Second, by
15387 inspecting a sorted list, we can discover the highest and lowest
15388 number, and thereby determine the largest and smallest length range
15392 @subsection Sorting Lists
15395 Emacs contains a function to sort lists, called (as you might guess)
15396 @code{sort}. The @code{sort} function takes two arguments, the list
15397 to be sorted, and a predicate that determines whether the first of
15398 two list elements is ``less'' than the second.
15400 As we saw earlier (@pxref{Wrong Type of Argument, , Using the Wrong
15401 Type Object as an Argument}), a predicate is a function that
15402 determines whether some property is true or false. The @code{sort}
15403 function will reorder a list according to whatever property the
15404 predicate uses; this means that @code{sort} can be used to sort
15405 non-numeric lists by non-numeric criteria---it can, for example,
15406 alphabetize a list.
15409 The @code{<} function is used when sorting a numeric list. For example,
15412 (sort '(4 8 21 17 33 7 21 7) '<)
15420 (4 7 7 8 17 21 21 33)
15424 (Note that in this example, both the arguments are quoted so that the
15425 symbols are not evaluated before being passed to @code{sort} as
15428 Sorting the list returned by the
15429 @code{recursive-lengths-list-many-files} function is straightforward;
15430 it uses the @code{<} function:
15434 In GNU Emacs 22, eval
15436 (cd "/usr/local/share/emacs/22.0.50/")
15438 (recursive-lengths-list-many-files
15439 '("./lisp/macros.el"
15440 "./lisp/mail/mailalias.el"
15441 "./lisp/makesum.el"))
15449 (recursive-lengths-list-many-files
15450 '("./lisp/macros.el"
15451 "./lisp/mailalias.el"
15452 "./lisp/makesum.el"))
15462 (29 32 38 85 90 95 178 180 181 218 263 283 321 324 480)
15466 (Note that in this example, the first argument to @code{sort} is not
15467 quoted, since the expression must be evaluated so as to produce the
15468 list that is passed to @code{sort}.)
15471 @subsection Making a List of Files
15473 The @code{recursive-lengths-list-many-files} function requires a list
15474 of files as its argument. For our test examples, we constructed such
15475 a list by hand; but the Emacs Lisp source directory is too large for
15476 us to do for that. Instead, we will write a function to do the job
15477 for us. In this function, we will use both a @code{while} loop and a
15480 @findex directory-files
15481 We did not have to write a function like this for older versions of
15482 GNU Emacs, since they placed all the @samp{.el} files in one
15483 directory. Instead, we were able to use the @code{directory-files}
15484 function, which lists the names of files that match a specified
15485 pattern within a single directory.
15487 However, recent versions of Emacs place Emacs Lisp files in
15488 sub-directories of the top level @file{lisp} directory. This
15489 re-arrangement eases navigation. For example, all the mail related
15490 files are in a @file{lisp} sub-directory called @file{mail}. But at
15491 the same time, this arrangement forces us to create a file listing
15492 function that descends into the sub-directories.
15494 @findex files-in-below-directory
15495 We can create this function, called @code{files-in-below-directory},
15496 using familiar functions such as @code{car}, @code{nthcdr}, and
15497 @code{substring} in conjunction with an existing function called
15498 @code{directory-files-and-attributes}. This latter function not only
15499 lists all the filenames in a directory, including the names
15500 of sub-directories, but also their attributes.
15502 To restate our goal: to create a function that will enable us
15503 to feed filenames to @code{recursive-lengths-list-many-files}
15504 as a list that looks like this (but with more elements):
15508 ("./lisp/macros.el"
15509 "./lisp/mail/rmail.el"
15510 "./lisp/makesum.el")
15514 The @code{directory-files-and-attributes} function returns a list of
15515 lists. Each of the lists within the main list consists of 13
15516 elements. The first element is a string that contains the name of the
15517 file---which, in GNU/Linux, may be a `directory file', that is to
15518 say, a file with the special attributes of a directory. The second
15519 element of the list is @code{t} for a directory, a string
15520 for symbolic link (the string is the name linked to), or @code{nil}.
15522 For example, the first @samp{.el} file in the @file{lisp/} directory
15523 is @file{abbrev.el}. Its name is
15524 @file{/usr/local/share/emacs/22.1.1/lisp/abbrev.el} and it is not a
15525 directory or a symbolic link.
15528 This is how @code{directory-files-and-attributes} lists that file and
15540 (20615 27034 579989 697000)
15542 (20615 26327 734791 805000)
15554 On the other hand, @file{mail/} is a directory within the @file{lisp/}
15555 directory. The beginning of its listing looks like this:
15566 (To learn about the different attributes, look at the documentation of
15567 @code{file-attributes}. Bear in mind that the @code{file-attributes}
15568 function does not list the filename, so its first element is
15569 @code{directory-files-and-attributes}'s second element.)
15571 We will want our new function, @code{files-in-below-directory}, to
15572 list the @samp{.el} files in the directory it is told to check, and in
15573 any directories below that directory.
15575 This gives us a hint on how to construct
15576 @code{files-in-below-directory}: within a directory, the function
15577 should add @samp{.el} filenames to a list; and if, within a directory,
15578 the function comes upon a sub-directory, it should go into that
15579 sub-directory and repeat its actions.
15581 However, we should note that every directory contains a name that
15582 refers to itself, called @file{.}, (``dot'') and a name that refers to
15583 its parent directory, called @file{..} (``double dot''). (In
15584 @file{/}, the root directory, @file{..} refers to itself, since
15585 @file{/} has no parent.) Clearly, we do not want our
15586 @code{files-in-below-directory} function to enter those directories,
15587 since they always lead us, directly or indirectly, to the current
15590 Consequently, our @code{files-in-below-directory} function must do
15595 Check to see whether it is looking at a filename that ends in
15596 @samp{.el}; and if so, add its name to a list.
15599 Check to see whether it is looking at a filename that is the name of a
15600 directory; and if so,
15604 Check to see whether it is looking at @file{.} or @file{..}; and if
15608 Or else, go into that directory and repeat the process.
15612 Let's write a function definition to do these tasks. We will use a
15613 @code{while} loop to move from one filename to another within a
15614 directory, checking what needs to be done; and we will use a recursive
15615 call to repeat the actions on each sub-directory. The recursive
15616 pattern is `accumulate'
15617 (@pxref{Accumulate, , Recursive Pattern: @emph{accumulate}}),
15618 using @code{append} as the combiner.
15621 (directory-files "/usr/local/src/emacs/lisp/" t "\\.el$")
15622 (shell-command "find /usr/local/src/emacs/lisp/ -name '*.el'")
15624 (directory-files "/usr/local/share/emacs/22.1.1/lisp/" t "\\.el$")
15625 (shell-command "find /usr/local/share/emacs/22.1.1/lisp/ -name '*.el'")
15628 @c /usr/local/share/emacs/22.1.1/lisp/
15631 Here is the function:
15635 (defun files-in-below-directory (directory)
15636 "List the .el files in DIRECTORY and in its sub-directories."
15637 ;; Although the function will be used non-interactively,
15638 ;; it will be easier to test if we make it interactive.
15639 ;; The directory will have a name such as
15640 ;; "/usr/local/share/emacs/22.1.1/lisp/"
15641 (interactive "DDirectory name: ")
15644 (let (el-files-list
15645 (current-directory-list
15646 (directory-files-and-attributes directory t)))
15647 ;; while we are in the current directory
15648 (while current-directory-list
15652 ;; check to see whether filename ends in `.el'
15653 ;; and if so, append its name to a list.
15654 ((equal ".el" (substring (car (car current-directory-list)) -3))
15655 (setq el-files-list
15656 (cons (car (car current-directory-list)) el-files-list)))
15659 ;; check whether filename is that of a directory
15660 ((eq t (car (cdr (car current-directory-list))))
15661 ;; decide whether to skip or recurse
15664 (substring (car (car current-directory-list)) -1))
15665 ;; then do nothing since filename is that of
15666 ;; current directory or parent, "." or ".."
15670 ;; else descend into the directory and repeat the process
15671 (setq el-files-list
15673 (files-in-below-directory
15674 (car (car current-directory-list)))
15676 ;; move to the next filename in the list; this also
15677 ;; shortens the list so the while loop eventually comes to an end
15678 (setq current-directory-list (cdr current-directory-list)))
15679 ;; return the filenames
15684 @c (files-in-below-directory "/usr/local/src/emacs/lisp/")
15685 @c (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/")
15687 The @code{files-in-below-directory} @code{directory-files} function
15688 takes one argument, the name of a directory.
15691 Thus, on my system,
15693 @c (length (files-in-below-directory "/usr/local/src/emacs/lisp/"))
15695 @c !!! 22.1.1 lisp sources location here
15699 (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/"))
15704 tells me that in and below my Lisp sources directory are 1031
15707 @code{files-in-below-directory} returns a list in reverse alphabetical
15708 order. An expression to sort the list in alphabetical order looks
15714 (files-in-below-directory "/usr/local/share/emacs/22.1.1/lisp/")
15721 "Test how long it takes to find lengths of all sorted elisp defuns."
15722 (insert "\n" (current-time-string) "\n")
15725 (recursive-lengths-list-many-files
15726 (files-in-below-directory "/usr/local/src/emacs/lisp/"))
15728 (insert (format "%s" (current-time-string))))
15731 @node Counting function definitions
15732 @subsection Counting function definitions
15734 Our immediate goal is to generate a list that tells us how many
15735 function definitions contain fewer than 10 words and symbols, how many
15736 contain between 10 and 19 words and symbols, how many contain between
15737 20 and 29 words and symbols, and so on.
15739 With a sorted list of numbers, this is easy: count how many elements
15740 of the list are smaller than 10, then, after moving past the numbers
15741 just counted, count how many are smaller than 20, then, after moving
15742 past the numbers just counted, count how many are smaller than 30, and
15743 so on. Each of the numbers, 10, 20, 30, 40, and the like, is one
15744 larger than the top of that range. We can call the list of such
15745 numbers the @code{top-of-ranges} list.
15748 If we wished, we could generate this list automatically, but it is
15749 simpler to write a list manually. Here it is:
15750 @vindex top-of-ranges
15754 (defvar top-of-ranges
15757 110 120 130 140 150
15758 160 170 180 190 200
15759 210 220 230 240 250
15760 260 270 280 290 300)
15761 "List specifying ranges for `defuns-per-range'.")
15765 To change the ranges, we edit this list.
15767 Next, we need to write the function that creates the list of the
15768 number of definitions within each range. Clearly, this function must
15769 take the @code{sorted-lengths} and the @code{top-of-ranges} lists
15772 The @code{defuns-per-range} function must do two things again and
15773 again: it must count the number of definitions within a range
15774 specified by the current top-of-range value; and it must shift to the
15775 next higher value in the @code{top-of-ranges} list after counting the
15776 number of definitions in the current range. Since each of these
15777 actions is repetitive, we can use @code{while} loops for the job.
15778 One loop counts the number of definitions in the range defined by the
15779 current top-of-range value, and the other loop selects each of the
15780 top-of-range values in turn.
15782 Several entries of the @code{sorted-lengths} list are counted for each
15783 range; this means that the loop for the @code{sorted-lengths} list
15784 will be inside the loop for the @code{top-of-ranges} list, like a
15785 small gear inside a big gear.
15787 The inner loop counts the number of definitions within the range. It
15788 is a simple counting loop of the type we have seen before.
15789 (@xref{Incrementing Loop, , A loop with an incrementing counter}.)
15790 The true-or-false test of the loop tests whether the value from the
15791 @code{sorted-lengths} list is smaller than the current value of the
15792 top of the range. If it is, the function increments the counter and
15793 tests the next value from the @code{sorted-lengths} list.
15796 The inner loop looks like this:
15800 (while @var{length-element-smaller-than-top-of-range}
15801 (setq number-within-range (1+ number-within-range))
15802 (setq sorted-lengths (cdr sorted-lengths)))
15806 The outer loop must start with the lowest value of the
15807 @code{top-of-ranges} list, and then be set to each of the succeeding
15808 higher values in turn. This can be done with a loop like this:
15812 (while top-of-ranges
15813 @var{body-of-loop}@dots{}
15814 (setq top-of-ranges (cdr top-of-ranges)))
15819 Put together, the two loops look like this:
15823 (while top-of-ranges
15825 ;; @r{Count the number of elements within the current range.}
15826 (while @var{length-element-smaller-than-top-of-range}
15827 (setq number-within-range (1+ number-within-range))
15828 (setq sorted-lengths (cdr sorted-lengths)))
15830 ;; @r{Move to next range.}
15831 (setq top-of-ranges (cdr top-of-ranges)))
15835 In addition, in each circuit of the outer loop, Emacs should record
15836 the number of definitions within that range (the value of
15837 @code{number-within-range}) in a list. We can use @code{cons} for
15838 this purpose. (@xref{cons, , @code{cons}}.)
15840 The @code{cons} function works fine, except that the list it
15841 constructs will contain the number of definitions for the highest
15842 range at its beginning and the number of definitions for the lowest
15843 range at its end. This is because @code{cons} attaches new elements
15844 of the list to the beginning of the list, and since the two loops are
15845 working their way through the lengths' list from the lower end first,
15846 the @code{defuns-per-range-list} will end up largest number first.
15847 But we will want to print our graph with smallest values first and the
15848 larger later. The solution is to reverse the order of the
15849 @code{defuns-per-range-list}. We can do this using the
15850 @code{nreverse} function, which reverses the order of a list.
15857 (nreverse '(1 2 3 4))
15868 Note that the @code{nreverse} function is ``destructive''---that is,
15869 it changes the list to which it is applied; this contrasts with the
15870 @code{car} and @code{cdr} functions, which are non-destructive. In
15871 this case, we do not want the original @code{defuns-per-range-list},
15872 so it does not matter that it is destroyed. (The @code{reverse}
15873 function provides a reversed copy of a list, leaving the original list
15878 Put all together, the @code{defuns-per-range} looks like this:
15882 (defun defuns-per-range (sorted-lengths top-of-ranges)
15883 "SORTED-LENGTHS defuns in each TOP-OF-RANGES range."
15884 (let ((top-of-range (car top-of-ranges))
15885 (number-within-range 0)
15886 defuns-per-range-list)
15891 (while top-of-ranges
15897 ;; @r{Need number for numeric test.}
15898 (car sorted-lengths)
15899 (< (car sorted-lengths) top-of-range))
15903 ;; @r{Count number of definitions within current range.}
15904 (setq number-within-range (1+ number-within-range))
15905 (setq sorted-lengths (cdr sorted-lengths)))
15907 ;; @r{Exit inner loop but remain within outer loop.}
15911 (setq defuns-per-range-list
15912 (cons number-within-range defuns-per-range-list))
15913 (setq number-within-range 0) ; @r{Reset count to zero.}
15917 ;; @r{Move to next range.}
15918 (setq top-of-ranges (cdr top-of-ranges))
15919 ;; @r{Specify next top of range value.}
15920 (setq top-of-range (car top-of-ranges)))
15924 ;; @r{Exit outer loop and count the number of defuns larger than}
15925 ;; @r{ the largest top-of-range value.}
15926 (setq defuns-per-range-list
15928 (length sorted-lengths)
15929 defuns-per-range-list))
15933 ;; @r{Return a list of the number of definitions within each range,}
15934 ;; @r{ smallest to largest.}
15935 (nreverse defuns-per-range-list)))
15941 The function is straightforward except for one subtle feature. The
15942 true-or-false test of the inner loop looks like this:
15946 (and (car sorted-lengths)
15947 (< (car sorted-lengths) top-of-range))
15953 instead of like this:
15956 (< (car sorted-lengths) top-of-range)
15959 The purpose of the test is to determine whether the first item in the
15960 @code{sorted-lengths} list is less than the value of the top of the
15963 The simple version of the test works fine unless the
15964 @code{sorted-lengths} list has a @code{nil} value. In that case, the
15965 @code{(car sorted-lengths)} expression function returns
15966 @code{nil}. The @code{<} function cannot compare a number to
15967 @code{nil}, which is an empty list, so Emacs signals an error and
15968 stops the function from attempting to continue to execute.
15970 The @code{sorted-lengths} list always becomes @code{nil} when the
15971 counter reaches the end of the list. This means that any attempt to
15972 use the @code{defuns-per-range} function with the simple version of
15973 the test will fail.
15975 We solve the problem by using the @code{(car sorted-lengths)}
15976 expression in conjunction with the @code{and} expression. The
15977 @code{(car sorted-lengths)} expression returns a non-@code{nil}
15978 value so long as the list has at least one number within it, but
15979 returns @code{nil} if the list is empty. The @code{and} expression
15980 first evaluates the @code{(car sorted-lengths)} expression, and
15981 if it is @code{nil}, returns false @emph{without} evaluating the
15982 @code{<} expression. But if the @code{(car sorted-lengths)}
15983 expression returns a non-@code{nil} value, the @code{and} expression
15984 evaluates the @code{<} expression, and returns that value as the value
15985 of the @code{and} expression.
15987 @c colon in printed section title causes problem in Info cross reference
15988 This way, we avoid an error.
15991 (For information about @code{and}, see
15992 @ref{kill-new function, , The @code{kill-new} function}.)
15996 (@xref{kill-new function, , The @code{kill-new} function}, for
15997 information about @code{and}.)
16000 Here is a short test of the @code{defuns-per-range} function. First,
16001 evaluate the expression that binds (a shortened)
16002 @code{top-of-ranges} list to the list of values, then evaluate the
16003 expression for binding the @code{sorted-lengths} list, and then
16004 evaluate the @code{defuns-per-range} function.
16008 ;; @r{(Shorter list than we will use later.)}
16009 (setq top-of-ranges
16010 '(110 120 130 140 150
16011 160 170 180 190 200))
16013 (setq sorted-lengths
16014 '(85 86 110 116 122 129 154 176 179 200 265 300 300))
16016 (defuns-per-range sorted-lengths top-of-ranges)
16022 The list returned looks like this:
16025 (2 2 2 0 0 1 0 2 0 0 4)
16029 Indeed, there are two elements of the @code{sorted-lengths} list
16030 smaller than 110, two elements between 110 and 119, two elements
16031 between 120 and 129, and so on. There are four elements with a value
16034 @c The next step is to turn this numbers' list into a graph.
16035 @node Readying a Graph
16036 @chapter Readying a Graph
16037 @cindex Readying a graph
16038 @cindex Graph prototype
16039 @cindex Prototype graph
16040 @cindex Body of graph
16042 Our goal is to construct a graph showing the numbers of function
16043 definitions of various lengths in the Emacs lisp sources.
16045 As a practical matter, if you were creating a graph, you would
16046 probably use a program such as @code{gnuplot} to do the job.
16047 (@code{gnuplot} is nicely integrated into GNU Emacs.) In this case,
16048 however, we create one from scratch, and in the process we will
16049 re-acquaint ourselves with some of what we learned before and learn
16052 In this chapter, we will first write a simple graph printing function.
16053 This first definition will be a @dfn{prototype}, a rapidly written
16054 function that enables us to reconnoiter this unknown graph-making
16055 territory. We will discover dragons, or find that they are myth.
16056 After scouting the terrain, we will feel more confident and enhance
16057 the function to label the axes automatically.
16060 * Columns of a graph::
16061 * graph-body-print:: How to print the body of a graph.
16062 * recursive-graph-body-print::
16064 * Line Graph Exercise::
16068 @node Columns of a graph
16069 @unnumberedsec Printing the Columns of a Graph
16072 Since Emacs is designed to be flexible and work with all kinds of
16073 terminals, including character-only terminals, the graph will need to
16074 be made from one of the `typewriter' symbols. An asterisk will do; as
16075 we enhance the graph-printing function, we can make the choice of
16076 symbol a user option.
16078 We can call this function @code{graph-body-print}; it will take a
16079 @code{numbers-list} as its only argument. At this stage, we will not
16080 label the graph, but only print its body.
16082 The @code{graph-body-print} function inserts a vertical column of
16083 asterisks for each element in the @code{numbers-list}. The height of
16084 each line is determined by the value of that element of the
16085 @code{numbers-list}.
16087 Inserting columns is a repetitive act; that means that this function can
16088 be written either with a @code{while} loop or recursively.
16090 Our first challenge is to discover how to print a column of asterisks.
16091 Usually, in Emacs, we print characters onto a screen horizontally,
16092 line by line, by typing. We have two routes we can follow: write our
16093 own column-insertion function or discover whether one exists in Emacs.
16095 To see whether there is one in Emacs, we can use the @kbd{M-x apropos}
16096 command. This command is like the @kbd{C-h a} (@code{command-apropos})
16097 command, except that the latter finds only those functions that are
16098 commands. The @kbd{M-x apropos} command lists all symbols that match
16099 a regular expression, including functions that are not interactive.
16102 What we want to look for is some command that prints or inserts
16103 columns. Very likely, the name of the function will contain either
16104 the word `print' or the word `insert' or the word `column'.
16105 Therefore, we can simply type @kbd{M-x apropos RET
16106 print\|insert\|column RET} and look at the result. On my system, this
16107 command once too takes quite some time, and then produced a list of 79
16108 functions and variables. Now it does not take much time at all and
16109 produces a list of 211 functions and variables. Scanning down the
16110 list, the only function that looks as if it might do the job is
16111 @code{insert-rectangle}.
16114 Indeed, this is the function we want; its documentation says:
16119 Insert text of RECTANGLE with upper left corner at point.
16120 RECTANGLE's first line is inserted at point,
16121 its second line is inserted at a point vertically under point, etc.
16122 RECTANGLE should be a list of strings.
16123 After this command, the mark is at the upper left corner
16124 and point is at the lower right corner.
16128 We can run a quick test, to make sure it does what we expect of it.
16130 Here is the result of placing the cursor after the
16131 @code{insert-rectangle} expression and typing @kbd{C-u C-x C-e}
16132 (@code{eval-last-sexp}). The function inserts the strings
16133 @samp{"first"}, @samp{"second"}, and @samp{"third"} at and below
16134 point. Also the function returns @code{nil}.
16138 (insert-rectangle '("first" "second" "third"))first
16145 Of course, we won't be inserting the text of the
16146 @code{insert-rectangle} expression itself into the buffer in which we
16147 are making the graph, but will call the function from our program. We
16148 shall, however, have to make sure that point is in the buffer at the
16149 place where the @code{insert-rectangle} function will insert its
16152 If you are reading this in Info, you can see how this works by
16153 switching to another buffer, such as the @file{*scratch*} buffer,
16154 placing point somewhere in the buffer, typing @kbd{M-:}, typing the
16155 @code{insert-rectangle} expression into the minibuffer at the prompt,
16156 and then typing @key{RET}. This causes Emacs to evaluate the
16157 expression in the minibuffer, but to use as the value of point the
16158 position of point in the @file{*scratch*} buffer. (@kbd{M-:} is the
16159 keybinding for @code{eval-expression}. Also, @code{nil} does not
16160 appear in the @file{*scratch*} buffer since the expression is
16161 evaluated in the minibuffer.)
16163 We find when we do this that point ends up at the end of the last
16164 inserted line---that is to say, this function moves point as a
16165 side-effect. If we were to repeat the command, with point at this
16166 position, the next insertion would be below and to the right of the
16167 previous insertion. We don't want this! If we are going to make a
16168 bar graph, the columns need to be beside each other.
16170 So we discover that each cycle of the column-inserting @code{while}
16171 loop must reposition point to the place we want it, and that place
16172 will be at the top, not the bottom, of the column. Moreover, we
16173 remember that when we print a graph, we do not expect all the columns
16174 to be the same height. This means that the top of each column may be
16175 at a different height from the previous one. We cannot simply
16176 reposition point to the same line each time, but moved over to the
16177 right---or perhaps we can@dots{}
16179 We are planning to make the columns of the bar graph out of asterisks.
16180 The number of asterisks in the column is the number specified by the
16181 current element of the @code{numbers-list}. We need to construct a
16182 list of asterisks of the right length for each call to
16183 @code{insert-rectangle}. If this list consists solely of the requisite
16184 number of asterisks, then we will have position point the right number
16185 of lines above the base for the graph to print correctly. This could
16188 Alternatively, if we can figure out some way to pass
16189 @code{insert-rectangle} a list of the same length each time, then we
16190 can place point on the same line each time, but move it over one
16191 column to the right for each new column. If we do this, however, some
16192 of the entries in the list passed to @code{insert-rectangle} must be
16193 blanks rather than asterisks. For example, if the maximum height of
16194 the graph is 5, but the height of the column is 3, then
16195 @code{insert-rectangle} requires an argument that looks like this:
16198 (" " " " "*" "*" "*")
16201 This last proposal is not so difficult, so long as we can determine
16202 the column height. There are two ways for us to specify the column
16203 height: we can arbitrarily state what it will be, which would work
16204 fine for graphs of that height; or we can search through the list of
16205 numbers and use the maximum height of the list as the maximum height
16206 of the graph. If the latter operation were difficult, then the former
16207 procedure would be easiest, but there is a function built into Emacs
16208 that determines the maximum of its arguments. We can use that
16209 function. The function is called @code{max} and it returns the
16210 largest of all its arguments, which must be numbers. Thus, for
16218 returns 7. (A corresponding function called @code{min} returns the
16219 smallest of all its arguments.)
16223 However, we cannot simply call @code{max} on the @code{numbers-list};
16224 the @code{max} function expects numbers as its argument, not a list of
16225 numbers. Thus, the following expression,
16228 (max '(3 4 6 5 7 3))
16233 produces the following error message;
16236 Wrong type of argument: number-or-marker-p, (3 4 6 5 7 3)
16240 We need a function that passes a list of arguments to a function.
16241 This function is @code{apply}. This function `applies' its first
16242 argument (a function) to its remaining arguments, the last of which
16249 (apply 'max 3 4 7 3 '(4 8 5))
16255 (Incidentally, I don't know how you would learn of this function
16256 without a book such as this. It is possible to discover other
16257 functions, like @code{search-forward} or @code{insert-rectangle}, by
16258 guessing at a part of their names and then using @code{apropos}. Even
16259 though its base in metaphor is clear---`apply' its first argument to
16260 the rest---I doubt a novice would come up with that particular word
16261 when using @code{apropos} or other aid. Of course, I could be wrong;
16262 after all, the function was first named by someone who had to invent
16265 The second and subsequent arguments to @code{apply} are optional, so
16266 we can use @code{apply} to call a function and pass the elements of a
16267 list to it, like this, which also returns 8:
16270 (apply 'max '(4 8 5))
16273 This latter way is how we will use @code{apply}. The
16274 @code{recursive-lengths-list-many-files} function returns a numbers'
16275 list to which we can apply @code{max} (we could also apply @code{max} to
16276 the sorted numbers' list; it does not matter whether the list is
16280 Hence, the operation for finding the maximum height of the graph is this:
16283 (setq max-graph-height (apply 'max numbers-list))
16286 Now we can return to the question of how to create a list of strings
16287 for a column of the graph. Told the maximum height of the graph
16288 and the number of asterisks that should appear in the column, the
16289 function should return a list of strings for the
16290 @code{insert-rectangle} command to insert.
16292 Each column is made up of asterisks or blanks. Since the function is
16293 passed the value of the height of the column and the number of
16294 asterisks in the column, the number of blanks can be found by
16295 subtracting the number of asterisks from the height of the column.
16296 Given the number of blanks and the number of asterisks, two
16297 @code{while} loops can be used to construct the list:
16301 ;;; @r{First version.}
16302 (defun column-of-graph (max-graph-height actual-height)
16303 "Return list of strings that is one column of a graph."
16304 (let ((insert-list nil)
16305 (number-of-top-blanks
16306 (- max-graph-height actual-height)))
16310 ;; @r{Fill in asterisks.}
16311 (while (> actual-height 0)
16312 (setq insert-list (cons "*" insert-list))
16313 (setq actual-height (1- actual-height)))
16317 ;; @r{Fill in blanks.}
16318 (while (> number-of-top-blanks 0)
16319 (setq insert-list (cons " " insert-list))
16320 (setq number-of-top-blanks
16321 (1- number-of-top-blanks)))
16325 ;; @r{Return whole list.}
16330 If you install this function and then evaluate the following
16331 expression you will see that it returns the list as desired:
16334 (column-of-graph 5 3)
16342 (" " " " "*" "*" "*")
16345 As written, @code{column-of-graph} contains a major flaw: the symbols
16346 used for the blank and for the marked entries in the column are
16347 `hard-coded' as a space and asterisk. This is fine for a prototype,
16348 but you, or another user, may wish to use other symbols. For example,
16349 in testing the graph function, you many want to use a period in place
16350 of the space, to make sure the point is being repositioned properly
16351 each time the @code{insert-rectangle} function is called; or you might
16352 want to substitute a @samp{+} sign or other symbol for the asterisk.
16353 You might even want to make a graph-column that is more than one
16354 display column wide. The program should be more flexible. The way to
16355 do that is to replace the blank and the asterisk with two variables
16356 that we can call @code{graph-blank} and @code{graph-symbol} and define
16357 those variables separately.
16359 Also, the documentation is not well written. These considerations
16360 lead us to the second version of the function:
16364 (defvar graph-symbol "*"
16365 "String used as symbol in graph, usually an asterisk.")
16369 (defvar graph-blank " "
16370 "String used as blank in graph, usually a blank space.
16371 graph-blank must be the same number of columns wide
16377 (For an explanation of @code{defvar}, see
16378 @ref{defvar, , Initializing a Variable with @code{defvar}}.)
16382 ;;; @r{Second version.}
16383 (defun column-of-graph (max-graph-height actual-height)
16384 "Return MAX-GRAPH-HEIGHT strings; ACTUAL-HEIGHT are graph-symbols.
16388 The graph-symbols are contiguous entries at the end
16390 The list will be inserted as one column of a graph.
16391 The strings are either graph-blank or graph-symbol."
16395 (let ((insert-list nil)
16396 (number-of-top-blanks
16397 (- max-graph-height actual-height)))
16401 ;; @r{Fill in @code{graph-symbols}.}
16402 (while (> actual-height 0)
16403 (setq insert-list (cons graph-symbol insert-list))
16404 (setq actual-height (1- actual-height)))
16408 ;; @r{Fill in @code{graph-blanks}.}
16409 (while (> number-of-top-blanks 0)
16410 (setq insert-list (cons graph-blank insert-list))
16411 (setq number-of-top-blanks
16412 (1- number-of-top-blanks)))
16414 ;; @r{Return whole list.}
16419 If we wished, we could rewrite @code{column-of-graph} a third time to
16420 provide optionally for a line graph as well as for a bar graph. This
16421 would not be hard to do. One way to think of a line graph is that it
16422 is no more than a bar graph in which the part of each bar that is
16423 below the top is blank. To construct a column for a line graph, the
16424 function first constructs a list of blanks that is one shorter than
16425 the value, then it uses @code{cons} to attach a graph symbol to the
16426 list; then it uses @code{cons} again to attach the `top blanks' to
16429 It is easy to see how to write such a function, but since we don't
16430 need it, we will not do it. But the job could be done, and if it were
16431 done, it would be done with @code{column-of-graph}. Even more
16432 important, it is worth noting that few changes would have to be made
16433 anywhere else. The enhancement, if we ever wish to make it, is
16436 Now, finally, we come to our first actual graph printing function.
16437 This prints the body of a graph, not the labels for the vertical and
16438 horizontal axes, so we can call this @code{graph-body-print}.
16440 @node graph-body-print
16441 @section The @code{graph-body-print} Function
16442 @findex graph-body-print
16444 After our preparation in the preceding section, the
16445 @code{graph-body-print} function is straightforward. The function
16446 will print column after column of asterisks and blanks, using the
16447 elements of a numbers' list to specify the number of asterisks in each
16448 column. This is a repetitive act, which means we can use a
16449 decrementing @code{while} loop or recursive function for the job. In
16450 this section, we will write the definition using a @code{while} loop.
16452 The @code{column-of-graph} function requires the height of the graph
16453 as an argument, so we should determine and record that as a local variable.
16455 This leads us to the following template for the @code{while} loop
16456 version of this function:
16460 (defun graph-body-print (numbers-list)
16461 "@var{documentation}@dots{}"
16462 (let ((height @dots{}
16467 (while numbers-list
16468 @var{insert-columns-and-reposition-point}
16469 (setq numbers-list (cdr numbers-list)))))
16474 We need to fill in the slots of the template.
16476 Clearly, we can use the @code{(apply 'max numbers-list)} expression to
16477 determine the height of the graph.
16479 The @code{while} loop will cycle through the @code{numbers-list} one
16480 element at a time. As it is shortened by the @code{(setq numbers-list
16481 (cdr numbers-list))} expression, the @sc{car} of each instance of the
16482 list is the value of the argument for @code{column-of-graph}.
16484 At each cycle of the @code{while} loop, the @code{insert-rectangle}
16485 function inserts the list returned by @code{column-of-graph}. Since
16486 the @code{insert-rectangle} function moves point to the lower right of
16487 the inserted rectangle, we need to save the location of point at the
16488 time the rectangle is inserted, move back to that position after the
16489 rectangle is inserted, and then move horizontally to the next place
16490 from which @code{insert-rectangle} is called.
16492 If the inserted columns are one character wide, as they will be if
16493 single blanks and asterisks are used, the repositioning command is
16494 simply @code{(forward-char 1)}; however, the width of a column may be
16495 greater than one. This means that the repositioning command should be
16496 written @code{(forward-char symbol-width)}. The @code{symbol-width}
16497 itself is the length of a @code{graph-blank} and can be found using
16498 the expression @code{(length graph-blank)}. The best place to bind
16499 the @code{symbol-width} variable to the value of the width of graph
16500 column is in the varlist of the @code{let} expression.
16503 These considerations lead to the following function definition:
16507 (defun graph-body-print (numbers-list)
16508 "Print a bar graph of the NUMBERS-LIST.
16509 The numbers-list consists of the Y-axis values."
16511 (let ((height (apply 'max numbers-list))
16512 (symbol-width (length graph-blank))
16517 (while numbers-list
16518 (setq from-position (point))
16520 (column-of-graph height (car numbers-list)))
16521 (goto-char from-position)
16522 (forward-char symbol-width)
16525 ;; @r{Draw graph column by column.}
16527 (setq numbers-list (cdr numbers-list)))
16530 ;; @r{Place point for X axis labels.}
16531 (forward-line height)
16538 The one unexpected expression in this function is the
16539 @w{@code{(sit-for 0)}} expression in the @code{while} loop. This
16540 expression makes the graph printing operation more interesting to
16541 watch than it would be otherwise. The expression causes Emacs to
16542 `sit' or do nothing for a zero length of time and then redraw the
16543 screen. Placed here, it causes Emacs to redraw the screen column by
16544 column. Without it, Emacs would not redraw the screen until the
16547 We can test @code{graph-body-print} with a short list of numbers.
16551 Install @code{graph-symbol}, @code{graph-blank},
16552 @code{column-of-graph}, which are in
16554 @ref{Readying a Graph, , Readying a Graph},
16557 @ref{Columns of a graph},
16559 and @code{graph-body-print}.
16563 Copy the following expression:
16566 (graph-body-print '(1 2 3 4 6 4 3 5 7 6 5 2 3))
16570 Switch to the @file{*scratch*} buffer and place the cursor where you
16571 want the graph to start.
16574 Type @kbd{M-:} (@code{eval-expression}).
16577 Yank the @code{graph-body-print} expression into the minibuffer
16578 with @kbd{C-y} (@code{yank)}.
16581 Press @key{RET} to evaluate the @code{graph-body-print} expression.
16585 Emacs will print a graph like this:
16599 @node recursive-graph-body-print
16600 @section The @code{recursive-graph-body-print} Function
16601 @findex recursive-graph-body-print
16603 The @code{graph-body-print} function may also be written recursively.
16604 The recursive solution is divided into two parts: an outside `wrapper'
16605 that uses a @code{let} expression to determine the values of several
16606 variables that need only be found once, such as the maximum height of
16607 the graph, and an inside function that is called recursively to print
16611 The `wrapper' is uncomplicated:
16615 (defun recursive-graph-body-print (numbers-list)
16616 "Print a bar graph of the NUMBERS-LIST.
16617 The numbers-list consists of the Y-axis values."
16618 (let ((height (apply 'max numbers-list))
16619 (symbol-width (length graph-blank))
16621 (recursive-graph-body-print-internal
16628 The recursive function is a little more difficult. It has four parts:
16629 the `do-again-test', the printing code, the recursive call, and the
16630 `next-step-expression'. The `do-again-test' is a @code{when}
16631 expression that determines whether the @code{numbers-list} contains
16632 any remaining elements; if it does, the function prints one column of
16633 the graph using the printing code and calls itself again. The
16634 function calls itself again according to the value produced by the
16635 `next-step-expression' which causes the call to act on a shorter
16636 version of the @code{numbers-list}.
16640 (defun recursive-graph-body-print-internal
16641 (numbers-list height symbol-width)
16642 "Print a bar graph.
16643 Used within recursive-graph-body-print function."
16648 (setq from-position (point))
16650 (column-of-graph height (car numbers-list)))
16653 (goto-char from-position)
16654 (forward-char symbol-width)
16655 (sit-for 0) ; @r{Draw graph column by column.}
16656 (recursive-graph-body-print-internal
16657 (cdr numbers-list) height symbol-width)))
16662 After installation, this expression can be tested; here is a sample:
16665 (recursive-graph-body-print '(3 2 5 6 7 5 3 4 6 4 3 2 1))
16669 Here is what @code{recursive-graph-body-print} produces:
16683 Either of these two functions, @code{graph-body-print} or
16684 @code{recursive-graph-body-print}, create the body of a graph.
16687 @section Need for Printed Axes
16689 A graph needs printed axes, so you can orient yourself. For a do-once
16690 project, it may be reasonable to draw the axes by hand using Emacs's
16691 Picture mode; but a graph drawing function may be used more than once.
16693 For this reason, I have written enhancements to the basic
16694 @code{print-graph-body} function that automatically print labels for
16695 the horizontal and vertical axes. Since the label printing functions
16696 do not contain much new material, I have placed their description in
16697 an appendix. @xref{Full Graph, , A Graph with Labeled Axes}.
16699 @node Line Graph Exercise
16702 Write a line graph version of the graph printing functions.
16704 @node Emacs Initialization
16705 @chapter Your @file{.emacs} File
16706 @cindex @file{.emacs} file
16707 @cindex Customizing your @file{.emacs} file
16708 @cindex Initialization file
16710 ``You don't have to like Emacs to like it''---this seemingly
16711 paradoxical statement is the secret of GNU Emacs. The plain, `out of
16712 the box' Emacs is a generic tool. Most people who use it, customize
16713 it to suit themselves.
16715 GNU Emacs is mostly written in Emacs Lisp; this means that by writing
16716 expressions in Emacs Lisp you can change or extend Emacs.
16719 * Default Configuration::
16720 * Site-wide Init:: You can write site-wide init files.
16721 * defcustom:: Emacs will write code for you.
16722 * Beginning a .emacs File:: How to write a @code{.emacs file}.
16723 * Text and Auto-fill:: Automatically wrap lines.
16724 * Mail Aliases:: Use abbreviations for email addresses.
16725 * Indent Tabs Mode:: Don't use tabs with @TeX{}
16726 * Keybindings:: Create some personal keybindings.
16727 * Keymaps:: More about key binding.
16728 * Loading Files:: Load (i.e., evaluate) files automatically.
16729 * Autoload:: Make functions available.
16730 * Simple Extension:: Define a function; bind it to a key.
16731 * X11 Colors:: Colors in X.
16733 * Mode Line:: How to customize your mode line.
16737 @node Default Configuration
16738 @unnumberedsec Emacs's Default Configuration
16741 There are those who appreciate Emacs's default configuration. After
16742 all, Emacs starts you in C mode when you edit a C file, starts you in
16743 Fortran mode when you edit a Fortran file, and starts you in
16744 Fundamental mode when you edit an unadorned file. This all makes
16745 sense, if you do not know who is going to use Emacs. Who knows what a
16746 person hopes to do with an unadorned file? Fundamental mode is the
16747 right default for such a file, just as C mode is the right default for
16748 editing C code. (Enough programming languages have syntaxes
16749 that enable them to share or nearly share features, so C mode is
16750 now provided by CC mode, the `C Collection'.)
16752 But when you do know who is going to use Emacs---you,
16753 yourself---then it makes sense to customize Emacs.
16755 For example, I seldom want Fundamental mode when I edit an
16756 otherwise undistinguished file; I want Text mode. This is why I
16757 customize Emacs: so it suits me.
16759 You can customize and extend Emacs by writing or adapting a
16760 @file{~/.emacs} file. This is your personal initialization file; its
16761 contents, written in Emacs Lisp, tell Emacs what to do.@footnote{You
16762 may also add @file{.el} to @file{~/.emacs} and call it a
16763 @file{~/.emacs.el} file. In the past, you were forbidden to type the
16764 extra keystrokes that the name @file{~/.emacs.el} requires, but now
16765 you may. The new format is consistent with the Emacs Lisp file
16766 naming conventions; the old format saves typing.}
16768 A @file{~/.emacs} file contains Emacs Lisp code. You can write this
16769 code yourself; or you can use Emacs's @code{customize} feature to write
16770 the code for you. You can combine your own expressions and
16771 auto-written Customize expressions in your @file{.emacs} file.
16773 (I myself prefer to write my own expressions, except for those,
16774 particularly fonts, that I find easier to manipulate using the
16775 @code{customize} command. I combine the two methods.)
16777 Most of this chapter is about writing expressions yourself. It
16778 describes a simple @file{.emacs} file; for more information, see
16779 @ref{Init File, , The Init File, emacs, The GNU Emacs Manual}, and
16780 @ref{Init File, , The Init File, elisp, The GNU Emacs Lisp Reference
16783 @node Site-wide Init
16784 @section Site-wide Initialization Files
16786 @cindex @file{default.el} init file
16787 @cindex @file{site-init.el} init file
16788 @cindex @file{site-load.el} init file
16789 In addition to your personal initialization file, Emacs automatically
16790 loads various site-wide initialization files, if they exist. These
16791 have the same form as your @file{.emacs} file, but are loaded by
16794 Two site-wide initialization files, @file{site-load.el} and
16795 @file{site-init.el}, are loaded into Emacs and then `dumped' if a
16796 `dumped' version of Emacs is created, as is most common. (Dumped
16797 copies of Emacs load more quickly. However, once a file is loaded and
16798 dumped, a change to it does not lead to a change in Emacs unless you
16799 load it yourself or re-dump Emacs. @xref{Building Emacs, , Building
16800 Emacs, elisp, The GNU Emacs Lisp Reference Manual}, and the
16801 @file{INSTALL} file.)
16803 Three other site-wide initialization files are loaded automatically
16804 each time you start Emacs, if they exist. These are
16805 @file{site-start.el}, which is loaded @emph{before} your @file{.emacs}
16806 file, and @file{default.el}, and the terminal type file, which are both
16807 loaded @emph{after} your @file{.emacs} file.
16809 Settings and definitions in your @file{.emacs} file will overwrite
16810 conflicting settings and definitions in a @file{site-start.el} file,
16811 if it exists; but the settings and definitions in a @file{default.el}
16812 or terminal type file will overwrite those in your @file{.emacs} file.
16813 (You can prevent interference from a terminal type file by setting
16814 @code{term-file-prefix} to @code{nil}. @xref{Simple Extension, , A
16815 Simple Extension}.)
16817 @c Rewritten to avoid overfull hbox.
16818 The @file{INSTALL} file that comes in the distribution contains
16819 descriptions of the @file{site-init.el} and @file{site-load.el} files.
16821 The @file{loadup.el}, @file{startup.el}, and @file{loaddefs.el} files
16822 control loading. These files are in the @file{lisp} directory of the
16823 Emacs distribution and are worth perusing.
16825 The @file{loaddefs.el} file contains a good many suggestions as to
16826 what to put into your own @file{.emacs} file, or into a site-wide
16827 initialization file.
16830 @section Specifying Variables using @code{defcustom}
16833 You can specify variables using @code{defcustom} so that you and
16834 others can then use Emacs's @code{customize} feature to set their
16835 values. (You cannot use @code{customize} to write function
16836 definitions; but you can write @code{defuns} in your @file{.emacs}
16837 file. Indeed, you can write any Lisp expression in your @file{.emacs}
16840 The @code{customize} feature depends on the @code{defcustom} macro.
16841 Although you can use @code{defvar} or @code{setq} for variables that
16842 users set, the @code{defcustom} macro is designed for the job.
16844 You can use your knowledge of @code{defvar} for writing the
16845 first three arguments for @code{defcustom}. The first argument to
16846 @code{defcustom} is the name of the variable. The second argument is
16847 the variable's initial value, if any; and this value is set only if
16848 the value has not already been set. The third argument is the
16851 The fourth and subsequent arguments to @code{defcustom} specify types
16852 and options; these are not featured in @code{defvar}. (These
16853 arguments are optional.)
16855 Each of these arguments consists of a keyword followed by a value.
16856 Each keyword starts with the colon character @samp{:}.
16859 For example, the customizable user option variable
16860 @code{text-mode-hook} looks like this:
16864 (defcustom text-mode-hook nil
16865 "Normal hook run when entering Text mode and many related modes."
16867 :options '(turn-on-auto-fill flyspell-mode)
16873 The name of the variable is @code{text-mode-hook}; it has no default
16874 value; and its documentation string tells you what it does.
16876 The @code{:type} keyword tells Emacs the kind of data to which
16877 @code{text-mode-hook} should be set and how to display the value in a
16878 Customization buffer.
16880 The @code{:options} keyword specifies a suggested list of values for
16881 the variable. Usually, @code{:options} applies to a hook.
16882 The list is only a suggestion; it is not exclusive; a person who sets
16883 the variable may set it to other values; the list shown following the
16884 @code{:options} keyword is intended to offer convenient choices to a
16887 Finally, the @code{:group} keyword tells the Emacs Customization
16888 command in which group the variable is located. This tells where to
16891 The @code{defcustom} macro recognizes more than a dozen keywords.
16892 For more information, see @ref{Customization, , Writing Customization
16893 Definitions, elisp, The GNU Emacs Lisp Reference Manual}.
16895 Consider @code{text-mode-hook} as an example.
16897 There are two ways to customize this variable. You can use the
16898 customization command or write the appropriate expressions yourself.
16901 Using the customization command, you can type:
16908 and find that the group for editing files of data is called `data'.
16909 Enter that group. Text Mode Hook is the first member. You can click
16910 on its various options, such as @code{turn-on-auto-fill}, to set the
16911 values. After you click on the button to
16914 Save for Future Sessions
16918 Emacs will write an expression into your @file{.emacs} file.
16919 It will look like this:
16923 (custom-set-variables
16924 ;; custom-set-variables was added by Custom.
16925 ;; If you edit it by hand, you could mess it up, so be careful.
16926 ;; Your init file should contain only one such instance.
16927 ;; If there is more than one, they won't work right.
16928 '(text-mode-hook (quote (turn-on-auto-fill text-mode-hook-identify))))
16933 (The @code{text-mode-hook-identify} function tells
16934 @code{toggle-text-mode-auto-fill} which buffers are in Text mode.
16935 It comes on automatically.)
16937 The @code{custom-set-variables} function works somewhat differently
16938 than a @code{setq}. While I have never learned the differences, I
16939 modify the @code{custom-set-variables} expressions in my @file{.emacs}
16940 file by hand: I make the changes in what appears to me to be a
16941 reasonable manner and have not had any problems. Others prefer to use
16942 the Customization command and let Emacs do the work for them.
16944 Another @code{custom-set-@dots{}} function is @code{custom-set-faces}.
16945 This function sets the various font faces. Over time, I have set a
16946 considerable number of faces. Some of the time, I re-set them using
16947 @code{customize}; other times, I simply edit the
16948 @code{custom-set-faces} expression in my @file{.emacs} file itself.
16950 The second way to customize your @code{text-mode-hook} is to set it
16951 yourself in your @file{.emacs} file using code that has nothing to do
16952 with the @code{custom-set-@dots{}} functions.
16955 When you do this, and later use @code{customize}, you will see a
16959 CHANGED outside Customize; operating on it here may be unreliable.
16963 This message is only a warning. If you click on the button to
16966 Save for Future Sessions
16970 Emacs will write a @code{custom-set-@dots{}} expression near the end
16971 of your @file{.emacs} file that will be evaluated after your
16972 hand-written expression. It will, therefore, overrule your
16973 hand-written expression. No harm will be done. When you do this,
16974 however, be careful to remember which expression is active; if you
16975 forget, you may confuse yourself.
16977 So long as you remember where the values are set, you will have no
16978 trouble. In any event, the values are always set in your
16979 initialization file, which is usually called @file{.emacs}.
16981 I myself use @code{customize} for hardly anything. Mostly, I write
16982 expressions myself.
16986 Incidentally, to be more complete concerning defines: @code{defsubst}
16987 defines an inline function. The syntax is just like that of
16988 @code{defun}. @code{defconst} defines a symbol as a constant. The
16989 intent is that neither programs nor users should ever change a value
16990 set by @code{defconst}. (You can change it; the value set is a
16991 variable; but please do not.)
16993 @node Beginning a .emacs File
16994 @section Beginning a @file{.emacs} File
16995 @cindex @file{.emacs} file, beginning of
16997 When you start Emacs, it loads your @file{.emacs} file unless you tell
16998 it not to by specifying @samp{-q} on the command line. (The
16999 @code{emacs -q} command gives you a plain, out-of-the-box Emacs.)
17001 A @file{.emacs} file contains Lisp expressions. Often, these are no
17002 more than expressions to set values; sometimes they are function
17005 @xref{Init File, , The Init File @file{~/.emacs}, emacs, The GNU Emacs
17006 Manual}, for a short description of initialization files.
17008 This chapter goes over some of the same ground, but is a walk among
17009 extracts from a complete, long-used @file{.emacs} file---my own.
17011 The first part of the file consists of comments: reminders to myself.
17012 By now, of course, I remember these things, but when I started, I did
17018 ;;;; Bob's .emacs file
17019 ; Robert J. Chassell
17020 ; 26 September 1985
17025 Look at that date! I started this file a long time ago. I have been
17026 adding to it ever since.
17030 ; Each section in this file is introduced by a
17031 ; line beginning with four semicolons; and each
17032 ; entry is introduced by a line beginning with
17033 ; three semicolons.
17038 This describes the usual conventions for comments in Emacs Lisp.
17039 Everything on a line that follows a semicolon is a comment. Two,
17040 three, and four semicolons are used as subsection and section markers.
17041 (@xref{Comments, ,, elisp, The GNU Emacs Lisp Reference Manual}, for
17042 more about comments.)
17047 ; Control-h is the help key;
17048 ; after typing control-h, type a letter to
17049 ; indicate the subject about which you want help.
17050 ; For an explanation of the help facility,
17051 ; type control-h two times in a row.
17056 Just remember: type @kbd{C-h} two times for help.
17060 ; To find out about any mode, type control-h m
17061 ; while in that mode. For example, to find out
17062 ; about mail mode, enter mail mode and then type
17068 `Mode help', as I call this, is very helpful. Usually, it tells you
17069 all you need to know.
17071 Of course, you don't need to include comments like these in your
17072 @file{.emacs} file. I included them in mine because I kept forgetting
17073 about Mode help or the conventions for comments---but I was able to
17074 remember to look here to remind myself.
17076 @node Text and Auto-fill
17077 @section Text and Auto Fill Mode
17079 Now we come to the part that `turns on' Text mode and
17084 ;;; Text mode and Auto Fill mode
17085 ;; The next two lines put Emacs into Text mode
17086 ;; and Auto Fill mode, and are for writers who
17087 ;; want to start writing prose rather than code.
17088 (setq-default major-mode 'text-mode)
17089 (add-hook 'text-mode-hook 'turn-on-auto-fill)
17093 Here is the first part of this @file{.emacs} file that does something
17094 besides remind a forgetful human!
17096 The first of the two lines in parentheses tells Emacs to turn on Text
17097 mode when you find a file, @emph{unless} that file should go into some
17098 other mode, such as C mode.
17100 @cindex Per-buffer, local variables list
17101 @cindex Local variables list, per-buffer,
17102 @cindex Automatic mode selection
17103 @cindex Mode selection, automatic
17104 When Emacs reads a file, it looks at the extension to the file name,
17105 if any. (The extension is the part that comes after a @samp{.}.) If
17106 the file ends with a @samp{.c} or @samp{.h} extension then Emacs turns
17107 on C mode. Also, Emacs looks at first nonblank line of the file; if
17108 the line says @w{@samp{-*- C -*-}}, Emacs turns on C mode. Emacs
17109 possesses a list of extensions and specifications that it uses
17110 automatically. In addition, Emacs looks near the last page for a
17111 per-buffer, ``local variables list'', if any.
17114 @xref{Choosing Modes, , How Major Modes are Chosen, emacs, The GNU
17117 @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs
17121 See sections ``How Major Modes are Chosen'' and ``Local Variables in
17122 Files'' in @cite{The GNU Emacs Manual}.
17125 Now, back to the @file{.emacs} file.
17128 Here is the line again; how does it work?
17130 @cindex Text Mode turned on
17132 (setq major-mode 'text-mode)
17136 This line is a short, but complete Emacs Lisp expression.
17138 We are already familiar with @code{setq}. It sets the following variable,
17139 @code{major-mode}, to the subsequent value, which is @code{text-mode}.
17140 The single quote mark before @code{text-mode} tells Emacs to deal directly
17141 with the @code{text-mode} symbol, not with whatever it might stand for.
17142 @xref{set & setq, , Setting the Value of a Variable},
17143 for a reminder of how @code{setq} works.
17144 The main point is that there is no difference between the procedure you
17145 use to set a value in your @file{.emacs} file and the procedure you use
17146 anywhere else in Emacs.
17149 Here is the next line:
17151 @cindex Auto Fill mode turned on
17154 (add-hook 'text-mode-hook 'turn-on-auto-fill)
17158 In this line, the @code{add-hook} command adds
17159 @code{turn-on-auto-fill} to the variable.
17161 @code{turn-on-auto-fill} is the name of a program, that, you guessed
17162 it!, turns on Auto Fill mode.
17164 Every time Emacs turns on Text mode, Emacs runs the commands `hooked'
17165 onto Text mode. So every time Emacs turns on Text mode, Emacs also
17166 turns on Auto Fill mode.
17168 In brief, the first line causes Emacs to enter Text mode when you edit a
17169 file, unless the file name extension, a first non-blank line, or local
17170 variables to tell Emacs otherwise.
17172 Text mode among other actions, sets the syntax table to work
17173 conveniently for writers. In Text mode, Emacs considers an apostrophe
17174 as part of a word like a letter; but Emacs does not consider a period
17175 or a space as part of a word. Thus, @kbd{M-f} moves you over
17176 @samp{it's}. On the other hand, in C mode, @kbd{M-f} stops just after
17177 the @samp{t} of @samp{it's}.
17179 The second line causes Emacs to turn on Auto Fill mode when it turns
17180 on Text mode. In Auto Fill mode, Emacs automatically breaks a line
17181 that is too wide and brings the excessively wide part of the line down
17182 to the next line. Emacs breaks lines between words, not within them.
17184 When Auto Fill mode is turned off, lines continue to the right as you
17185 type them. Depending on how you set the value of
17186 @code{truncate-lines}, the words you type either disappear off the
17187 right side of the screen, or else are shown, in a rather ugly and
17188 unreadable manner, as a continuation line on the screen.
17191 In addition, in this part of my @file{.emacs} file, I tell the Emacs
17192 fill commands to insert two spaces after a colon:
17195 (setq colon-double-space t)
17199 @section Mail Aliases
17201 Here is a @code{setq} that `turns on' mail aliases, along with more
17207 ; To enter mail mode, type `C-x m'
17208 ; To enter RMAIL (for reading mail),
17210 (setq mail-aliases t)
17214 @cindex Mail aliases
17216 This @code{setq} command sets the value of the variable
17217 @code{mail-aliases} to @code{t}. Since @code{t} means true, the line
17218 says, in effect, ``Yes, use mail aliases.''
17220 Mail aliases are convenient short names for long email addresses or
17221 for lists of email addresses. The file where you keep your `aliases'
17222 is @file{~/.mailrc}. You write an alias like this:
17225 alias geo george@@foobar.wiz.edu
17229 When you write a message to George, address it to @samp{geo}; the
17230 mailer will automatically expand @samp{geo} to the full address.
17232 @node Indent Tabs Mode
17233 @section Indent Tabs Mode
17234 @cindex Tabs, preventing
17235 @findex indent-tabs-mode
17237 By default, Emacs inserts tabs in place of multiple spaces when it
17238 formats a region. (For example, you might indent many lines of text
17239 all at once with the @code{indent-region} command.) Tabs look fine on
17240 a terminal or with ordinary printing, but they produce badly indented
17241 output when you use @TeX{} or Texinfo since @TeX{} ignores tabs.
17244 The following turns off Indent Tabs mode:
17248 ;;; Prevent Extraneous Tabs
17249 (setq-default indent-tabs-mode nil)
17253 Note that this line uses @code{setq-default} rather than the
17254 @code{setq} command that we have seen before. The @code{setq-default}
17255 command sets values only in buffers that do not have their own local
17256 values for the variable.
17259 @xref{Just Spaces, , Tabs vs. Spaces, emacs, The GNU Emacs Manual}.
17261 @xref{File Variables, , Local Variables in Files, emacs, The GNU Emacs
17265 See sections ``Tabs vs.@: Spaces'' and ``Local Variables in
17266 Files'' in @cite{The GNU Emacs Manual}.
17271 @section Some Keybindings
17273 Now for some personal keybindings:
17277 ;;; Compare windows
17278 (global-set-key "\C-cw" 'compare-windows)
17282 @findex compare-windows
17283 @code{compare-windows} is a nifty command that compares the text in
17284 your current window with text in the next window. It makes the
17285 comparison by starting at point in each window, moving over text in
17286 each window as far as they match. I use this command all the time.
17288 This also shows how to set a key globally, for all modes.
17290 @cindex Setting a key globally
17291 @cindex Global set key
17292 @cindex Key setting globally
17293 @findex global-set-key
17294 The command is @code{global-set-key}. It is followed by the
17295 keybinding. In a @file{.emacs} file, the keybinding is written as
17296 shown: @code{\C-c} stands for `control-c', which means `press the
17297 control key and the @key{c} key at the same time'. The @code{w} means
17298 `press the @key{w} key'. The keybinding is surrounded by double
17299 quotation marks. In documentation, you would write this as
17300 @w{@kbd{C-c w}}. (If you were binding a @key{META} key, such as
17301 @kbd{M-c}, rather than a @key{CTRL} key, you would write
17302 @w{@code{\M-c}} in your @file{.emacs} file. @xref{Init Rebinding, ,
17303 Rebinding Keys in Your Init File, emacs, The GNU Emacs Manual}, for
17306 The command invoked by the keys is @code{compare-windows}. Note that
17307 @code{compare-windows} is preceded by a single quote; otherwise, Emacs
17308 would first try to evaluate the symbol to determine its value.
17310 These three things, the double quotation marks, the backslash before
17311 the @samp{C}, and the single quote mark are necessary parts of
17312 keybinding that I tend to forget. Fortunately, I have come to
17313 remember that I should look at my existing @file{.emacs} file, and
17314 adapt what is there.
17316 As for the keybinding itself: @kbd{C-c w}. This combines the prefix
17317 key, @kbd{C-c}, with a single character, in this case, @kbd{w}. This
17318 set of keys, @kbd{C-c} followed by a single character, is strictly
17319 reserved for individuals' own use. (I call these `own' keys, since
17320 these are for my own use.) You should always be able to create such a
17321 keybinding for your own use without stomping on someone else's
17322 keybinding. If you ever write an extension to Emacs, please avoid
17323 taking any of these keys for public use. Create a key like @kbd{C-c
17324 C-w} instead. Otherwise, we will run out of `own' keys.
17327 Here is another keybinding, with a comment:
17331 ;;; Keybinding for `occur'
17332 ; I use occur a lot, so let's bind it to a key:
17333 (global-set-key "\C-co" 'occur)
17338 The @code{occur} command shows all the lines in the current buffer
17339 that contain a match for a regular expression. Matching lines are
17340 shown in a buffer called @file{*Occur*}. That buffer serves as a menu
17341 to jump to occurrences.
17343 @findex global-unset-key
17344 @cindex Unbinding key
17345 @cindex Key unbinding
17347 Here is how to unbind a key, so it does not
17353 (global-unset-key "\C-xf")
17357 There is a reason for this unbinding: I found I inadvertently typed
17358 @w{@kbd{C-x f}} when I meant to type @kbd{C-x C-f}. Rather than find a
17359 file, as I intended, I accidentally set the width for filled text,
17360 almost always to a width I did not want. Since I hardly ever reset my
17361 default width, I simply unbound the key.
17363 @findex list-buffers, @r{rebound}
17364 @findex buffer-menu, @r{bound to key}
17366 The following rebinds an existing key:
17370 ;;; Rebind `C-x C-b' for `buffer-menu'
17371 (global-set-key "\C-x\C-b" 'buffer-menu)
17375 By default, @kbd{C-x C-b} runs the
17376 @code{list-buffers} command. This command lists
17377 your buffers in @emph{another} window. Since I
17378 almost always want to do something in that
17379 window, I prefer the @code{buffer-menu}
17380 command, which not only lists the buffers,
17381 but moves point into that window.
17386 @cindex Rebinding keys
17388 Emacs uses @dfn{keymaps} to record which keys call which commands.
17389 When you use @code{global-set-key} to set the keybinding for a single
17390 command in all parts of Emacs, you are specifying the keybinding in
17391 @code{current-global-map}.
17393 Specific modes, such as C mode or Text mode, have their own keymaps;
17394 the mode-specific keymaps override the global map that is shared by
17397 The @code{global-set-key} function binds, or rebinds, the global
17398 keymap. For example, the following binds the key @kbd{C-x C-b} to the
17399 function @code{buffer-menu}:
17402 (global-set-key "\C-x\C-b" 'buffer-menu)
17405 Mode-specific keymaps are bound using the @code{define-key} function,
17406 which takes a specific keymap as an argument, as well as the key and
17407 the command. For example, my @file{.emacs} file contains the
17408 following expression to bind the @code{texinfo-insert-@@group} command
17409 to @kbd{C-c C-c g}:
17413 (define-key texinfo-mode-map "\C-c\C-cg" 'texinfo-insert-@@group)
17418 The @code{texinfo-insert-@@group} function itself is a little extension
17419 to Texinfo mode that inserts @samp{@@group} into a Texinfo file. I
17420 use this command all the time and prefer to type the three strokes
17421 @kbd{C-c C-c g} rather than the six strokes @kbd{@@ g r o u p}.
17422 (@samp{@@group} and its matching @samp{@@end group} are commands that
17423 keep all enclosed text together on one page; many multi-line examples
17424 in this book are surrounded by @samp{@@group @dots{} @@end group}.)
17427 Here is the @code{texinfo-insert-@@group} function definition:
17431 (defun texinfo-insert-@@group ()
17432 "Insert the string @@group in a Texinfo buffer."
17434 (beginning-of-line)
17435 (insert "@@group\n"))
17439 (Of course, I could have used Abbrev mode to save typing, rather than
17440 write a function to insert a word; but I prefer key strokes consistent
17441 with other Texinfo mode key bindings.)
17443 You will see numerous @code{define-key} expressions in
17444 @file{loaddefs.el} as well as in the various mode libraries, such as
17445 @file{cc-mode.el} and @file{lisp-mode.el}.
17447 @xref{Key Bindings, , Customizing Key Bindings, emacs, The GNU Emacs
17448 Manual}, and @ref{Keymaps, , Keymaps, elisp, The GNU Emacs Lisp
17449 Reference Manual}, for more information about keymaps.
17451 @node Loading Files
17452 @section Loading Files
17453 @cindex Loading files
17456 Many people in the GNU Emacs community have written extensions to
17457 Emacs. As time goes by, these extensions are often included in new
17458 releases. For example, the Calendar and Diary packages are now part
17459 of the standard GNU Emacs, as is Calc.
17461 You can use a @code{load} command to evaluate a complete file and
17462 thereby install all the functions and variables in the file into Emacs.
17465 @c (auto-compression-mode t)
17468 (load "~/emacs/slowsplit")
17471 This evaluates, i.e., loads, the @file{slowsplit.el} file or if it
17472 exists, the faster, byte compiled @file{slowsplit.elc} file from the
17473 @file{emacs} sub-directory of your home directory. The file contains
17474 the function @code{split-window-quietly}, which John Robinson wrote in
17477 The @code{split-window-quietly} function splits a window with the
17478 minimum of redisplay. I installed it in 1989 because it worked well
17479 with the slow 1200 baud terminals I was then using. Nowadays, I only
17480 occasionally come across such a slow connection, but I continue to use
17481 the function because I like the way it leaves the bottom half of a
17482 buffer in the lower of the new windows and the top half in the upper
17486 To replace the key binding for the default
17487 @code{split-window-vertically}, you must also unset that key and bind
17488 the keys to @code{split-window-quietly}, like this:
17492 (global-unset-key "\C-x2")
17493 (global-set-key "\C-x2" 'split-window-quietly)
17498 If you load many extensions, as I do, then instead of specifying the
17499 exact location of the extension file, as shown above, you can specify
17500 that directory as part of Emacs's @code{load-path}. Then, when Emacs
17501 loads a file, it will search that directory as well as its default
17502 list of directories. (The default list is specified in @file{paths.h}
17503 when Emacs is built.)
17506 The following command adds your @file{~/emacs} directory to the
17507 existing load path:
17511 ;;; Emacs Load Path
17512 (setq load-path (cons "~/emacs" load-path))
17516 Incidentally, @code{load-library} is an interactive interface to the
17517 @code{load} function. The complete function looks like this:
17519 @findex load-library
17522 (defun load-library (library)
17523 "Load the library named LIBRARY.
17524 This is an interface to the function `load'."
17526 (list (completing-read "Load library: "
17527 (apply-partially 'locate-file-completion-table
17529 (get-load-suffixes)))))
17534 The name of the function, @code{load-library}, comes from the use of
17535 `library' as a conventional synonym for `file'. The source for the
17536 @code{load-library} command is in the @file{files.el} library.
17538 Another interactive command that does a slightly different job is
17539 @code{load-file}. @xref{Lisp Libraries, , Libraries of Lisp Code for
17540 Emacs, emacs, The GNU Emacs Manual}, for information on the
17541 distinction between @code{load-library} and this command.
17544 @section Autoloading
17547 Instead of installing a function by loading the file that contains it,
17548 or by evaluating the function definition, you can make the function
17549 available but not actually install it until it is first called. This
17550 is called @dfn{autoloading}.
17552 When you execute an autoloaded function, Emacs automatically evaluates
17553 the file that contains the definition, and then calls the function.
17555 Emacs starts quicker with autoloaded functions, since their libraries
17556 are not loaded right away; but you need to wait a moment when you
17557 first use such a function, while its containing file is evaluated.
17559 Rarely used functions are frequently autoloaded. The
17560 @file{loaddefs.el} library contains hundreds of autoloaded functions,
17561 from @code{bookmark-set} to @code{wordstar-mode}. Of course, you may
17562 come to use a `rare' function frequently. When you do, you should
17563 load that function's file with a @code{load} expression in your
17564 @file{.emacs} file.
17566 In my @file{.emacs} file, I load 14 libraries that contain functions
17567 that would otherwise be autoloaded. (Actually, it would have been
17568 better to include these files in my `dumped' Emacs, but I forgot.
17569 @xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp
17570 Reference Manual}, and the @file{INSTALL} file for more about
17573 You may also want to include autoloaded expressions in your @file{.emacs}
17574 file. @code{autoload} is a built-in function that takes up to five
17575 arguments, the final three of which are optional. The first argument
17576 is the name of the function to be autoloaded; the second is the name
17577 of the file to be loaded. The third argument is documentation for the
17578 function, and the fourth tells whether the function can be called
17579 interactively. The fifth argument tells what type of
17580 object---@code{autoload} can handle a keymap or macro as well as a
17581 function (the default is a function).
17584 Here is a typical example:
17588 (autoload 'html-helper-mode
17589 "html-helper-mode" "Edit HTML documents" t)
17594 (@code{html-helper-mode} is an older alternative to @code{html-mode},
17595 which is a standard part of the distribution.)
17598 This expression autoloads the @code{html-helper-mode} function. It
17599 takes it from the @file{html-helper-mode.el} file (or from the byte
17600 compiled version @file{html-helper-mode.elc}, if that exists.) The
17601 file must be located in a directory specified by @code{load-path}.
17602 The documentation says that this is a mode to help you edit documents
17603 written in the HyperText Markup Language. You can call this mode
17604 interactively by typing @kbd{M-x html-helper-mode}. (You need to
17605 duplicate the function's regular documentation in the autoload
17606 expression because the regular function is not yet loaded, so its
17607 documentation is not available.)
17609 @xref{Autoload, , Autoload, elisp, The GNU Emacs Lisp Reference
17610 Manual}, for more information.
17612 @node Simple Extension
17613 @section A Simple Extension: @code{line-to-top-of-window}
17614 @findex line-to-top-of-window
17615 @cindex Simple extension in @file{.emacs} file
17617 Here is a simple extension to Emacs that moves the line point is on to
17618 the top of the window. I use this all the time, to make text easier
17621 You can put the following code into a separate file and then load it
17622 from your @file{.emacs} file, or you can include it within your
17623 @file{.emacs} file.
17626 Here is the definition:
17630 ;;; Line to top of window;
17631 ;;; replace three keystroke sequence C-u 0 C-l
17632 (defun line-to-top-of-window ()
17633 "Move the line point is on to top of window."
17640 Now for the keybinding.
17642 Nowadays, function keys as well as mouse button events and
17643 non-@sc{ascii} characters are written within square brackets, without
17644 quotation marks. (In Emacs version 18 and before, you had to write
17645 different function key bindings for each different make of terminal.)
17647 I bind @code{line-to-top-of-window} to my @key{F6} function key like
17651 (global-set-key [f6] 'line-to-top-of-window)
17654 For more information, see @ref{Init Rebinding, , Rebinding Keys in
17655 Your Init File, emacs, The GNU Emacs Manual}.
17657 @cindex Conditional 'twixt two versions of Emacs
17658 @cindex Version of Emacs, choosing
17659 @cindex Emacs version, choosing
17660 If you run two versions of GNU Emacs, such as versions 22 and 23, and
17661 use one @file{.emacs} file, you can select which code to evaluate with
17662 the following conditional:
17667 ((= 22 emacs-major-version)
17668 ;; evaluate version 22 code
17670 ((= 23 emacs-major-version)
17671 ;; evaluate version 23 code
17676 For example, recent versions blink
17677 their cursors by default. I hate such blinking, as well as other
17678 features, so I placed the following in my @file{.emacs}
17679 file@footnote{When I start instances of Emacs that do not load my
17680 @file{.emacs} file or any site file, I also turn off blinking:
17683 emacs -q --no-site-file -eval '(blink-cursor-mode nil)'
17685 @exdent Or nowadays, using an even more sophisticated set of options,
17693 (when (>= emacs-major-version 21)
17694 (blink-cursor-mode 0)
17695 ;; Insert newline when you press `C-n' (next-line)
17696 ;; at the end of the buffer
17697 (setq next-line-add-newlines t)
17700 ;; Turn on image viewing
17701 (auto-image-file-mode t)
17704 ;; Turn on menu bar (this bar has text)
17705 ;; (Use numeric argument to turn on)
17709 ;; Turn off tool bar (this bar has icons)
17710 ;; (Use numeric argument to turn on)
17711 (tool-bar-mode nil)
17714 ;; Turn off tooltip mode for tool bar
17715 ;; (This mode causes icon explanations to pop up)
17716 ;; (Use numeric argument to turn on)
17718 ;; If tooltips turned on, make tips appear promptly
17719 (setq tooltip-delay 0.1) ; default is 0.7 second
17725 @section X11 Colors
17727 You can specify colors when you use Emacs with the MIT X Windowing
17730 I dislike the default colors and specify my own.
17733 Here are the expressions in my @file{.emacs}
17734 file that set values:
17738 ;; Set cursor color
17739 (set-cursor-color "white")
17742 (set-mouse-color "white")
17744 ;; Set foreground and background
17745 (set-foreground-color "white")
17746 (set-background-color "darkblue")
17750 ;;; Set highlighting colors for isearch and drag
17751 (set-face-foreground 'highlight "white")
17752 (set-face-background 'highlight "blue")
17756 (set-face-foreground 'region "cyan")
17757 (set-face-background 'region "blue")
17761 (set-face-foreground 'secondary-selection "skyblue")
17762 (set-face-background 'secondary-selection "darkblue")
17766 ;; Set calendar highlighting colors
17767 (setq calendar-load-hook
17769 (set-face-foreground 'diary-face "skyblue")
17770 (set-face-background 'holiday-face "slate blue")
17771 (set-face-foreground 'holiday-face "white")))
17775 The various shades of blue soothe my eye and prevent me from seeing
17776 the screen flicker.
17778 Alternatively, I could have set my specifications in various X
17779 initialization files. For example, I could set the foreground,
17780 background, cursor, and pointer (i.e., mouse) colors in my
17781 @file{~/.Xresources} file like this:
17785 Emacs*foreground: white
17786 Emacs*background: darkblue
17787 Emacs*cursorColor: white
17788 Emacs*pointerColor: white
17792 In any event, since it is not part of Emacs, I set the root color of
17793 my X window in my @file{~/.xinitrc} file, like this@footnote{I also
17794 run more modern window managers, such as Enlightenment, Gnome, or KDE;
17795 in those cases, I often specify an image rather than a plain color.}:
17798 xsetroot -solid Navy -fg white &
17802 @node Miscellaneous
17803 @section Miscellaneous Settings for a @file{.emacs} File
17806 Here are a few miscellaneous settings:
17811 Set the shape and color of the mouse cursor:
17815 ; Cursor shapes are defined in
17816 ; `/usr/include/X11/cursorfont.h';
17817 ; for example, the `target' cursor is number 128;
17818 ; the `top_left_arrow' cursor is number 132.
17822 (let ((mpointer (x-get-resource "*mpointer"
17823 "*emacs*mpointer")))
17824 ;; If you have not set your mouse pointer
17825 ;; then set it, otherwise leave as is:
17826 (if (eq mpointer nil)
17827 (setq mpointer "132")) ; top_left_arrow
17830 (setq x-pointer-shape (string-to-int mpointer))
17831 (set-mouse-color "white"))
17836 Or you can set the values of a variety of features in an alist, like
17842 default-frame-alist
17843 '((cursor-color . "white")
17844 (mouse-color . "white")
17845 (foreground-color . "white")
17846 (background-color . "DodgerBlue4")
17847 ;; (cursor-type . bar)
17848 (cursor-type . box)
17851 (tool-bar-lines . 0)
17852 (menu-bar-lines . 1)
17856 "-Misc-Fixed-Medium-R-Normal--20-200-75-75-C-100-ISO8859-1")
17862 Convert @kbd{@key{CTRL}-h} into @key{DEL} and @key{DEL}
17863 into @kbd{@key{CTRL}-h}.@*
17864 (Some older keyboards needed this, although I have not seen the
17869 ;; Translate `C-h' to <DEL>.
17870 ; (keyboard-translate ?\C-h ?\C-?)
17872 ;; Translate <DEL> to `C-h'.
17873 (keyboard-translate ?\C-? ?\C-h)
17877 @item Turn off a blinking cursor!
17881 (if (fboundp 'blink-cursor-mode)
17882 (blink-cursor-mode -1))
17887 or start GNU Emacs with the command @code{emacs -nbc}.
17890 @item When using `grep'@*
17891 @samp{-i}@w{ } Ignore case distinctions@*
17892 @samp{-n}@w{ } Prefix each line of output with line number@*
17893 @samp{-H}@w{ } Print the filename for each match.@*
17894 @samp{-e}@w{ } Protect patterns beginning with a hyphen character, @samp{-}
17897 (setq grep-command "grep -i -nH -e ")
17901 @c Evidently, no longer needed in GNU Emacs 22
17903 item Automatically uncompress compressed files when visiting them
17906 (load "uncompress")
17911 @item Find an existing buffer, even if it has a different name@*
17912 This avoids problems with symbolic links.
17915 (setq find-file-existing-other-name t)
17918 @item Set your language environment and default input method
17922 (set-language-environment "latin-1")
17923 ;; Remember you can enable or disable multilingual text input
17924 ;; with the @code{toggle-input-method'} (@kbd{C-\}) command
17925 (setq default-input-method "latin-1-prefix")
17929 If you want to write with Chinese `GB' characters, set this instead:
17933 (set-language-environment "Chinese-GB")
17934 (setq default-input-method "chinese-tonepy")
17939 @subsubheading Fixing Unpleasant Key Bindings
17940 @cindex Key bindings, fixing
17941 @cindex Bindings, key, fixing unpleasant
17943 Some systems bind keys unpleasantly. Sometimes, for example, the
17944 @key{CTRL} key appears in an awkward spot rather than at the far left
17947 Usually, when people fix these sorts of keybindings, they do not
17948 change their @file{~/.emacs} file. Instead, they bind the proper keys
17949 on their consoles with the @code{loadkeys} or @code{install-keymap}
17950 commands in their boot script and then include @code{xmodmap} commands
17951 in their @file{.xinitrc} or @file{.Xsession} file for X Windows.
17959 loadkeys /usr/share/keymaps/i386/qwerty/emacs2.kmap.gz
17961 install-keymap emacs2
17967 For a @file{.xinitrc} or @file{.Xsession} file when the @key{Caps
17968 Lock} key is at the far left of the home row:
17972 # Bind the key labeled `Caps Lock' to `Control'
17973 # (Such a broken user interface suggests that keyboard manufacturers
17974 # think that computers are typewriters from 1885.)
17976 xmodmap -e "clear Lock"
17977 xmodmap -e "add Control = Caps_Lock"
17983 In a @file{.xinitrc} or @file{.Xsession} file, to convert an @key{ALT}
17984 key to a @key{META} key:
17988 # Some ill designed keyboards have a key labeled ALT and no Meta
17989 xmodmap -e "keysym Alt_L = Meta_L Alt_L"
17995 @section A Modified Mode Line
17996 @vindex mode-line-format
17997 @cindex Mode line format
17999 Finally, a feature I really like: a modified mode line.
18001 When I work over a network, I forget which machine I am using. Also,
18002 I tend to I lose track of where I am, and which line point is on.
18004 So I reset my mode line to look like this:
18007 -:-- foo.texi rattlesnake:/home/bob/ Line 1 (Texinfo Fill) Top
18010 I am visiting a file called @file{foo.texi}, on my machine
18011 @file{rattlesnake} in my @file{/home/bob} buffer. I am on line 1, in
18012 Texinfo mode, and am at the top of the buffer.
18015 My @file{.emacs} file has a section that looks like this:
18019 ;; Set a Mode Line that tells me which machine, which directory,
18020 ;; and which line I am on, plus the other customary information.
18021 (setq-default mode-line-format
18025 "mouse-1: select window, mouse-2: delete others ..."))
18026 mode-line-mule-info
18028 mode-line-frame-identification
18032 mode-line-buffer-identification
18035 (system-name) 0 (string-match "\\..+" (system-name))))
18040 "mouse-1: select window, mouse-2: delete others ..."))
18041 (line-number-mode " Line %l ")
18047 "mouse-1: select window, mouse-2: delete others ..."))
18048 (:eval (mode-line-mode-name))
18051 #("%n" 0 2 (help-echo "mouse-2: widen" local-map (keymap ...)))
18060 Here, I redefine the default mode line. Most of the parts are from
18061 the original; but I make a few changes. I set the @emph{default} mode
18062 line format so as to permit various modes, such as Info, to override
18065 Many elements in the list are self-explanatory:
18066 @code{mode-line-modified} is a variable that tells whether the buffer
18067 has been modified, @code{mode-name} tells the name of the mode, and so
18068 on. However, the format looks complicated because of two features we
18069 have not discussed.
18071 @cindex Properties, in mode line example
18072 The first string in the mode line is a dash or hyphen, @samp{-}. In
18073 the old days, it would have been specified simply as @code{"-"}. But
18074 nowadays, Emacs can add properties to a string, such as highlighting
18075 or, as in this case, a help feature. If you place your mouse cursor
18076 over the hyphen, some help information appears (By default, you must
18077 wait seven-tenths of a second before the information appears. You can
18078 change that timing by changing the value of @code{tooltip-delay}.)
18081 The new string format has a special syntax:
18084 #("-" 0 1 (help-echo "mouse-1: select window, ..."))
18088 The @code{#(} begins a list. The first element of the list is the
18089 string itself, just one @samp{-}. The second and third
18090 elements specify the range over which the fourth element applies. A
18091 range starts @emph{after} a character, so a zero means the range
18092 starts just before the first character; a 1 means that the range ends
18093 just after the first character. The third element is the property for
18094 the range. It consists of a property list, a
18095 property name, in this case, @samp{help-echo}, followed by a value, in this
18096 case, a string. The second, third, and fourth elements of this new
18097 string format can be repeated.
18099 @xref{Text Properties, , Text Properties, elisp, The GNU Emacs Lisp
18100 Reference Manual}, and see @ref{Mode Line Format, , Mode Line Format,
18101 elisp, The GNU Emacs Lisp Reference Manual}, for more information.
18103 @code{mode-line-buffer-identification}
18104 displays the current buffer name. It is a list
18105 beginning @code{(#("%12b" 0 4 @dots{}}.
18106 The @code{#(} begins the list.
18108 The @samp{"%12b"} displays the current buffer name, using the
18109 @code{buffer-name} function with which we are familiar; the `12'
18110 specifies the maximum number of characters that will be displayed.
18111 When a name has fewer characters, whitespace is added to fill out to
18112 this number. (Buffer names can and often should be longer than 12
18113 characters; this length works well in a typical 80 column wide
18116 @code{:eval} says to evaluate the following form and use the result as
18117 a string to display. In this case, the expression displays the first
18118 component of the full system name. The end of the first component is
18119 a @samp{.} (`period'), so I use the @code{string-match} function to
18120 tell me the length of the first component. The substring from the
18121 zeroth character to that length is the name of the machine.
18124 This is the expression:
18129 (system-name) 0 (string-match "\\..+" (system-name))))
18133 @samp{%[} and @samp{%]} cause a pair of square brackets
18134 to appear for each recursive editing level. @samp{%n} says `Narrow'
18135 when narrowing is in effect. @samp{%P} tells you the percentage of
18136 the buffer that is above the bottom of the window, or `Top', `Bottom',
18137 or `All'. (A lower case @samp{p} tell you the percentage above the
18138 @emph{top} of the window.) @samp{%-} inserts enough dashes to fill
18141 Remember, ``You don't have to like Emacs to like it''---your own
18142 Emacs can have different colors, different commands, and different
18143 keys than a default Emacs.
18145 On the other hand, if you want to bring up a plain `out of the box'
18146 Emacs, with no customization, type:
18153 This will start an Emacs that does @emph{not} load your
18154 @file{~/.emacs} initialization file. A plain, default Emacs. Nothing
18161 GNU Emacs has two debuggers, @code{debug} and @code{edebug}. The
18162 first is built into the internals of Emacs and is always with you;
18163 the second requires that you instrument a function before you can use it.
18165 Both debuggers are described extensively in @ref{Debugging, ,
18166 Debugging Lisp Programs, elisp, The GNU Emacs Lisp Reference Manual}.
18167 In this chapter, I will walk through a short example of each.
18170 * debug:: How to use the built-in debugger.
18171 * debug-on-entry:: Start debugging when you call a function.
18172 * debug-on-quit:: Start debugging when you quit with @kbd{C-g}.
18173 * edebug:: How to use Edebug, a source level debugger.
18174 * Debugging Exercises::
18178 @section @code{debug}
18181 Suppose you have written a function definition that is intended to
18182 return the sum of the numbers 1 through a given number. (This is the
18183 @code{triangle} function discussed earlier. @xref{Decrementing
18184 Example, , Example with Decrementing Counter}, for a discussion.)
18185 @c xref{Decrementing Loop,, Loop with a Decrementing Counter}, for a discussion.)
18187 However, your function definition has a bug. You have mistyped
18188 @samp{1=} for @samp{1-}. Here is the broken definition:
18190 @findex triangle-bugged
18193 (defun triangle-bugged (number)
18194 "Return sum of numbers 1 through NUMBER inclusive."
18196 (while (> number 0)
18197 (setq total (+ total number))
18198 (setq number (1= number))) ; @r{Error here.}
18203 If you are reading this in Info, you can evaluate this definition in
18204 the normal fashion. You will see @code{triangle-bugged} appear in the
18208 Now evaluate the @code{triangle-bugged} function with an
18212 (triangle-bugged 4)
18216 In a recent GNU Emacs, you will create and enter a @file{*Backtrace*}
18222 ---------- Buffer: *Backtrace* ----------
18223 Debugger entered--Lisp error: (void-function 1=)
18225 (setq number (1= number))
18226 (while (> number 0) (setq total (+ total number))
18227 (setq number (1= number)))
18228 (let ((total 0)) (while (> number 0) (setq total ...)
18229 (setq number ...)) total)
18233 eval((triangle-bugged 4))
18234 eval-last-sexp-1(nil)
18235 eval-last-sexp(nil)
18236 call-interactively(eval-last-sexp)
18237 ---------- Buffer: *Backtrace* ----------
18242 (I have reformatted this example slightly; the debugger does not fold
18243 long lines. As usual, you can quit the debugger by typing @kbd{q} in
18244 the @file{*Backtrace*} buffer.)
18246 In practice, for a bug as simple as this, the `Lisp error' line will
18247 tell you what you need to know to correct the definition. The
18248 function @code{1=} is `void'.
18252 In GNU Emacs 20 and before, you will see:
18255 Symbol's function definition is void:@: 1=
18259 which has the same meaning as the @file{*Backtrace*} buffer line in
18263 However, suppose you are not quite certain what is going on?
18264 You can read the complete backtrace.
18266 In this case, you need to run a recent GNU Emacs, which automatically
18267 starts the debugger that puts you in the @file{*Backtrace*} buffer; or
18268 else, you need to start the debugger manually as described below.
18270 Read the @file{*Backtrace*} buffer from the bottom up; it tells you
18271 what Emacs did that led to the error. Emacs made an interactive call
18272 to @kbd{C-x C-e} (@code{eval-last-sexp}), which led to the evaluation
18273 of the @code{triangle-bugged} expression. Each line above tells you
18274 what the Lisp interpreter evaluated next.
18277 The third line from the top of the buffer is
18280 (setq number (1= number))
18284 Emacs tried to evaluate this expression; in order to do so, it tried
18285 to evaluate the inner expression shown on the second line from the
18294 This is where the error occurred; as the top line says:
18297 Debugger entered--Lisp error: (void-function 1=)
18301 You can correct the mistake, re-evaluate the function definition, and
18302 then run your test again.
18304 @node debug-on-entry
18305 @section @code{debug-on-entry}
18306 @findex debug-on-entry
18308 A recent GNU Emacs starts the debugger automatically when your
18309 function has an error.
18312 GNU Emacs version 20 and before did not; it simply
18313 presented you with an error message. You had to start the debugger
18317 Incidentally, you can start the debugger manually for all versions of
18318 Emacs; the advantage is that the debugger runs even if you do not have
18319 a bug in your code. Sometimes your code will be free of bugs!
18321 You can enter the debugger when you call the function by calling
18322 @code{debug-on-entry}.
18329 M-x debug-on-entry RET triangle-bugged RET
18334 Now, evaluate the following:
18337 (triangle-bugged 5)
18341 All versions of Emacs will create a @file{*Backtrace*} buffer and tell
18342 you that it is beginning to evaluate the @code{triangle-bugged}
18347 ---------- Buffer: *Backtrace* ----------
18348 Debugger entered--entering a function:
18349 * triangle-bugged(5)
18350 eval((triangle-bugged 5))
18353 eval-last-sexp-1(nil)
18354 eval-last-sexp(nil)
18355 call-interactively(eval-last-sexp)
18356 ---------- Buffer: *Backtrace* ----------
18360 In the @file{*Backtrace*} buffer, type @kbd{d}. Emacs will evaluate
18361 the first expression in @code{triangle-bugged}; the buffer will look
18366 ---------- Buffer: *Backtrace* ----------
18367 Debugger entered--beginning evaluation of function call form:
18368 * (let ((total 0)) (while (> number 0) (setq total ...)
18369 (setq number ...)) total)
18370 * triangle-bugged(5)
18371 eval((triangle-bugged 5))
18374 eval-last-sexp-1(nil)
18375 eval-last-sexp(nil)
18376 call-interactively(eval-last-sexp)
18377 ---------- Buffer: *Backtrace* ----------
18382 Now, type @kbd{d} again, eight times, slowly. Each time you type
18383 @kbd{d}, Emacs will evaluate another expression in the function
18387 Eventually, the buffer will look like this:
18391 ---------- Buffer: *Backtrace* ----------
18392 Debugger entered--beginning evaluation of function call form:
18393 * (setq number (1= number))
18394 * (while (> number 0) (setq total (+ total number))
18395 (setq number (1= number)))
18398 * (let ((total 0)) (while (> number 0) (setq total ...)
18399 (setq number ...)) total)
18400 * triangle-bugged(5)
18401 eval((triangle-bugged 5))
18404 eval-last-sexp-1(nil)
18405 eval-last-sexp(nil)
18406 call-interactively(eval-last-sexp)
18407 ---------- Buffer: *Backtrace* ----------
18413 Finally, after you type @kbd{d} two more times, Emacs will reach the
18414 error, and the top two lines of the @file{*Backtrace*} buffer will look
18419 ---------- Buffer: *Backtrace* ----------
18420 Debugger entered--Lisp error: (void-function 1=)
18423 ---------- Buffer: *Backtrace* ----------
18427 By typing @kbd{d}, you were able to step through the function.
18429 You can quit a @file{*Backtrace*} buffer by typing @kbd{q} in it; this
18430 quits the trace, but does not cancel @code{debug-on-entry}.
18432 @findex cancel-debug-on-entry
18433 To cancel the effect of @code{debug-on-entry}, call
18434 @code{cancel-debug-on-entry} and the name of the function, like this:
18437 M-x cancel-debug-on-entry RET triangle-bugged RET
18441 (If you are reading this in Info, cancel @code{debug-on-entry} now.)
18443 @node debug-on-quit
18444 @section @code{debug-on-quit} and @code{(debug)}
18446 In addition to setting @code{debug-on-error} or calling @code{debug-on-entry},
18447 there are two other ways to start @code{debug}.
18449 @findex debug-on-quit
18450 You can start @code{debug} whenever you type @kbd{C-g}
18451 (@code{keyboard-quit}) by setting the variable @code{debug-on-quit} to
18452 @code{t}. This is useful for debugging infinite loops.
18455 @cindex @code{(debug)} in code
18456 Or, you can insert a line that says @code{(debug)} into your code
18457 where you want the debugger to start, like this:
18461 (defun triangle-bugged (number)
18462 "Return sum of numbers 1 through NUMBER inclusive."
18464 (while (> number 0)
18465 (setq total (+ total number))
18466 (debug) ; @r{Start debugger.}
18467 (setq number (1= number))) ; @r{Error here.}
18472 The @code{debug} function is described in detail in @ref{Debugger, ,
18473 The Lisp Debugger, elisp, The GNU Emacs Lisp Reference Manual}.
18476 @section The @code{edebug} Source Level Debugger
18477 @cindex Source level debugger
18480 Edebug is a source level debugger. Edebug normally displays the
18481 source of the code you are debugging, with an arrow at the left that
18482 shows which line you are currently executing.
18484 You can walk through the execution of a function, line by line, or run
18485 quickly until reaching a @dfn{breakpoint} where execution stops.
18487 Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs
18488 Lisp Reference Manual}.
18491 Here is a bugged function definition for @code{triangle-recursively}.
18492 @xref{Recursive triangle function, , Recursion in place of a counter},
18493 for a review of it.
18497 (defun triangle-recursively-bugged (number)
18498 "Return sum of numbers 1 through NUMBER inclusive.
18503 (triangle-recursively-bugged
18504 (1= number))))) ; @r{Error here.}
18509 Normally, you would install this definition by positioning your cursor
18510 after the function's closing parenthesis and typing @kbd{C-x C-e}
18511 (@code{eval-last-sexp}) or else by positioning your cursor within the
18512 definition and typing @kbd{C-M-x} (@code{eval-defun}). (By default,
18513 the @code{eval-defun} command works only in Emacs Lisp mode or in Lisp
18517 However, to prepare this function definition for Edebug, you must
18518 first @dfn{instrument} the code using a different command. You can do
18519 this by positioning your cursor within or just after the definition
18523 M-x edebug-defun RET
18527 This will cause Emacs to load Edebug automatically if it is not
18528 already loaded, and properly instrument the function.
18530 After instrumenting the function, place your cursor after the
18531 following expression and type @kbd{C-x C-e} (@code{eval-last-sexp}):
18534 (triangle-recursively-bugged 3)
18538 You will be jumped back to the source for
18539 @code{triangle-recursively-bugged} and the cursor positioned at the
18540 beginning of the @code{if} line of the function. Also, you will see
18541 an arrowhead at the left hand side of that line. The arrowhead marks
18542 the line where the function is executing. (In the following examples,
18543 we show the arrowhead with @samp{=>}; in a windowing system, you may
18544 see the arrowhead as a solid triangle in the window `fringe'.)
18547 =>@point{}(if (= number 1)
18552 In the example, the location of point is displayed with a star,
18553 @samp{@point{}} (in Info, it is displayed as @samp{-!-}).
18556 In the example, the location of point is displayed as @samp{@point{}}
18557 (in a printed book, it is displayed with a five pointed star).
18560 If you now press @key{SPC}, point will move to the next expression to
18561 be executed; the line will look like this:
18564 =>(if @point{}(= number 1)
18568 As you continue to press @key{SPC}, point will move from expression to
18569 expression. At the same time, whenever an expression returns a value,
18570 that value will be displayed in the echo area. For example, after you
18571 move point past @code{number}, you will see the following:
18574 Result: 3 (#o3, #x3, ?\C-c)
18578 This means the value of @code{number} is 3, which is octal three,
18579 hexadecimal three, and @sc{ascii} `control-c' (the third letter of the
18580 alphabet, in case you need to know this information).
18582 You can continue moving through the code until you reach the line with
18583 the error. Before evaluation, that line looks like this:
18586 => @point{}(1= number))))) ; @r{Error here.}
18591 When you press @key{SPC} once again, you will produce an error message
18595 Symbol's function definition is void:@: 1=
18601 Press @kbd{q} to quit Edebug.
18603 To remove instrumentation from a function definition, simply
18604 re-evaluate it with a command that does not instrument it.
18605 For example, you could place your cursor after the definition's
18606 closing parenthesis and type @kbd{C-x C-e}.
18608 Edebug does a great deal more than walk with you through a function.
18609 You can set it so it races through on its own, stopping only at an
18610 error or at specified stopping points; you can cause it to display the
18611 changing values of various expressions; you can find out how many
18612 times a function is called, and more.
18614 Edebug is described in @ref{Edebug, , , elisp, The GNU Emacs
18615 Lisp Reference Manual}.
18618 @node Debugging Exercises
18619 @section Debugging Exercises
18623 Install the @code{@value{COUNT-WORDS}} function and then cause it to
18624 enter the built-in debugger when you call it. Run the command on a
18625 region containing two words. You will need to press @kbd{d} a
18626 remarkable number of times. On your system, is a `hook' called after
18627 the command finishes? (For information on hooks, see @ref{Command
18628 Overview, , Command Loop Overview, elisp, The GNU Emacs Lisp Reference
18632 Copy @code{@value{COUNT-WORDS}} into the @file{*scratch*} buffer,
18633 instrument the function for Edebug, and walk through its execution.
18634 The function does not need to have a bug, although you can introduce
18635 one if you wish. If the function lacks a bug, the walk-through
18636 completes without problems.
18639 While running Edebug, type @kbd{?} to see a list of all the Edebug commands.
18640 (The @code{global-edebug-prefix} is usually @kbd{C-x X}, i.e.,
18641 @kbd{@key{CTRL}-x} followed by an upper case @kbd{X}; use this prefix
18642 for commands made outside of the Edebug debugging buffer.)
18645 In the Edebug debugging buffer, use the @kbd{p}
18646 (@code{edebug-bounce-point}) command to see where in the region the
18647 @code{@value{COUNT-WORDS}} is working.
18650 Move point to some spot further down the function and then type the
18651 @kbd{h} (@code{edebug-goto-here}) command to jump to that location.
18654 Use the @kbd{t} (@code{edebug-trace-mode}) command to cause Edebug to
18655 walk through the function on its own; use an upper case @kbd{T} for
18656 @code{edebug-Trace-fast-mode}.
18659 Set a breakpoint, then run Edebug in Trace mode until it reaches the
18664 @chapter Conclusion
18666 We have now reached the end of this Introduction. You have now
18667 learned enough about programming in Emacs Lisp to set values, to write
18668 simple @file{.emacs} files for yourself and your friends, and write
18669 simple customizations and extensions to Emacs.
18671 This is a place to stop. Or, if you wish, you can now go onward, and
18674 You have learned some of the basic nuts and bolts of programming. But
18675 only some. There are a great many more brackets and hinges that are
18676 easy to use that we have not touched.
18678 A path you can follow right now lies among the sources to GNU Emacs
18681 @cite{The GNU Emacs Lisp Reference Manual}.
18684 @ref{Top, , The GNU Emacs Lisp Reference Manual, elisp, The GNU
18685 Emacs Lisp Reference Manual}.
18688 The Emacs Lisp sources are an adventure. When you read the sources and
18689 come across a function or expression that is unfamiliar, you need to
18690 figure out or find out what it does.
18692 Go to the Reference Manual. It is a thorough, complete, and fairly
18693 easy-to-read description of Emacs Lisp. It is written not only for
18694 experts, but for people who know what you know. (The @cite{Reference
18695 Manual} comes with the standard GNU Emacs distribution. Like this
18696 introduction, it comes as a Texinfo source file, so you can read it
18697 on-line and as a typeset, printed book.)
18699 Go to the other on-line help that is part of GNU Emacs: the on-line
18700 documentation for all functions and variables, and @code{find-tag},
18701 the program that takes you to sources.
18703 Here is an example of how I explore the sources. Because of its name,
18704 @file{simple.el} is the file I looked at first, a long time ago. As
18705 it happens some of the functions in @file{simple.el} are complicated,
18706 or at least look complicated at first sight. The @code{open-line}
18707 function, for example, looks complicated.
18709 You may want to walk through this function slowly, as we did with the
18710 @code{forward-sentence} function. (@xref{forward-sentence, The
18711 @code{forward-sentence} function}.) Or you may want to skip that
18712 function and look at another, such as @code{split-line}. You don't
18713 need to read all the functions. According to
18714 @code{count-words-in-defun}, the @code{split-line} function contains
18715 102 words and symbols.
18717 Even though it is short, @code{split-line} contains expressions
18718 we have not studied: @code{skip-chars-forward}, @code{indent-to},
18719 @code{current-column} and @code{insert-and-inherit}.
18721 Consider the @code{skip-chars-forward} function. (It is part of the
18722 function definition for @code{back-to-indentation}, which is shown in
18723 @ref{Review, , Review}.)
18725 In GNU Emacs, you can find out more about @code{skip-chars-forward} by
18726 typing @kbd{C-h f} (@code{describe-function}) and the name of the
18727 function. This gives you the function documentation.
18729 You may be able to guess what is done by a well named function such as
18730 @code{indent-to}; or you can look it up, too. Incidentally, the
18731 @code{describe-function} function itself is in @file{help.el}; it is
18732 one of those long, but decipherable functions. You can look up
18733 @code{describe-function} using the @kbd{C-h f} command!
18735 In this instance, since the code is Lisp, the @file{*Help*} buffer
18736 contains the name of the library containing the function's source.
18737 You can put point over the name of the library and press the RET key,
18738 which in this situation is bound to @code{help-follow}, and be taken
18739 directly to the source, in the same way as @kbd{M-.}
18742 The definition for @code{describe-function} illustrates how to
18743 customize the @code{interactive} expression without using the standard
18744 character codes; and it shows how to create a temporary buffer.
18746 (The @code{indent-to} function is written in C rather than Emacs Lisp;
18747 it is a `built-in' function. @code{help-follow} takes you to its
18748 source as does @code{find-tag}, when properly set up.)
18750 You can look at a function's source using @code{find-tag}, which is
18751 bound to @kbd{M-.} Finally, you can find out what the Reference
18752 Manual has to say by visiting the manual in Info, and typing @kbd{i}
18753 (@code{Info-index}) and the name of the function, or by looking up the
18754 function in the index to a printed copy of the manual.
18756 Similarly, you can find out what is meant by
18757 @code{insert-and-inherit}.
18759 Other interesting source files include @file{paragraphs.el},
18760 @file{loaddefs.el}, and @file{loadup.el}. The @file{paragraphs.el}
18761 file includes short, easily understood functions as well as longer
18762 ones. The @file{loaddefs.el} file contains the many standard
18763 autoloads and many keymaps. I have never looked at it all; only at
18764 parts. @file{loadup.el} is the file that loads the standard parts of
18765 Emacs; it tells you a great deal about how Emacs is built.
18766 (@xref{Building Emacs, , Building Emacs, elisp, The GNU Emacs Lisp
18767 Reference Manual}, for more about building.)
18769 As I said, you have learned some nuts and bolts; however, and very
18770 importantly, we have hardly touched major aspects of programming; I
18771 have said nothing about how to sort information, except to use the
18772 predefined @code{sort} function; I have said nothing about how to store
18773 information, except to use variables and lists; I have said nothing
18774 about how to write programs that write programs. These are topics for
18775 another, and different kind of book, a different kind of learning.
18777 What you have done is learn enough for much practical work with GNU
18778 Emacs. What you have done is get started. This is the end of a
18781 @c ================ Appendix ================
18784 @appendix The @code{the-the} Function
18786 @cindex Duplicated words function
18787 @cindex Words, duplicated
18789 Sometimes when you you write text, you duplicate words---as with ``you
18790 you'' near the beginning of this sentence. I find that most
18791 frequently, I duplicate ``the''; hence, I call the function for
18792 detecting duplicated words, @code{the-the}.
18795 As a first step, you could use the following regular expression to
18796 search for duplicates:
18799 \\(\\w+[ \t\n]+\\)\\1
18803 This regexp matches one or more word-constituent characters followed
18804 by one or more spaces, tabs, or newlines. However, it does not detect
18805 duplicated words on different lines, since the ending of the first
18806 word, the end of the line, is different from the ending of the second
18807 word, a space. (For more information about regular expressions, see
18808 @ref{Regexp Search, , Regular Expression Searches}, as well as
18809 @ref{Regexps, , Syntax of Regular Expressions, emacs, The GNU Emacs
18810 Manual}, and @ref{Regular Expressions, , Regular Expressions, elisp,
18811 The GNU Emacs Lisp Reference Manual}.)
18813 You might try searching just for duplicated word-constituent
18814 characters but that does not work since the pattern detects doubles
18815 such as the two occurrences of `th' in `with the'.
18817 Another possible regexp searches for word-constituent characters
18818 followed by non-word-constituent characters, reduplicated. Here,
18819 @w{@samp{\\w+}} matches one or more word-constituent characters and
18820 @w{@samp{\\W*}} matches zero or more non-word-constituent characters.
18823 \\(\\(\\w+\\)\\W*\\)\\1
18829 Here is the pattern that I use. It is not perfect, but good enough.
18830 @w{@samp{\\b}} matches the empty string, provided it is at the beginning
18831 or end of a word; @w{@samp{[^@@ \n\t]+}} matches one or more occurrences of
18832 any characters that are @emph{not} an @@-sign, space, newline, or tab.
18835 \\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b
18838 One can write more complicated expressions, but I found that this
18839 expression is good enough, so I use it.
18841 Here is the @code{the-the} function, as I include it in my
18842 @file{.emacs} file, along with a handy global key binding:
18847 "Search forward for for a duplicated word."
18849 (message "Searching for for duplicated words ...")
18853 ;; This regexp is not perfect
18854 ;; but is fairly good over all:
18855 (if (re-search-forward
18856 "\\b\\([^@@ \n\t]+\\)[ \n\t]+\\1\\b" nil 'move)
18857 (message "Found duplicated word.")
18858 (message "End of buffer")))
18862 ;; Bind `the-the' to C-c \
18863 (global-set-key "\C-c\\" 'the-the)
18872 one two two three four five
18877 You can substitute the other regular expressions shown above in the
18878 function definition and try each of them on this list.
18881 @appendix Handling the Kill Ring
18882 @cindex Kill ring handling
18883 @cindex Handling the kill ring
18884 @cindex Ring, making a list like a
18886 The kill ring is a list that is transformed into a ring by the
18887 workings of the @code{current-kill} function. The @code{yank} and
18888 @code{yank-pop} commands use the @code{current-kill} function.
18890 This appendix describes the @code{current-kill} function as well as
18891 both the @code{yank} and the @code{yank-pop} commands, but first,
18892 consider the workings of the kill ring.
18895 * What the Kill Ring Does::
18897 * yank:: Paste a copy of a clipped element.
18898 * yank-pop:: Insert element pointed to.
18903 @node What the Kill Ring Does
18904 @unnumberedsec What the Kill Ring Does
18908 The kill ring has a default maximum length of sixty items; this number
18909 is too large for an explanation. Instead, set it to four. Please
18910 evaluate the following:
18914 (setq old-kill-ring-max kill-ring-max)
18915 (setq kill-ring-max 4)
18920 Then, please copy each line of the following indented example into the
18921 kill ring. You may kill each line with @kbd{C-k} or mark it and copy
18925 (In a read-only buffer, such as the @file{*info*} buffer, the kill
18926 command, @kbd{C-k} (@code{kill-line}), will not remove the text,
18927 merely copy it to the kill ring. However, your machine may beep at
18928 you. Alternatively, for silence, you may copy the region of each line
18929 with the @kbd{M-w} (@code{kill-ring-save}) command. You must mark
18930 each line for this command to succeed, but it does not matter at which
18931 end you put point or mark.)
18935 Please invoke the calls in order, so that five elements attempt to
18936 fill the kill ring:
18941 second piece of text
18943 fourth line of text
18950 Then find the value of @code{kill-ring} by evaluating
18962 ("fifth bit of text" "fourth line of text"
18963 "third line" "second piece of text")
18968 The first element, @samp{first some text}, was dropped.
18971 To return to the old value for the length of the kill ring, evaluate:
18974 (setq kill-ring-max old-kill-ring-max)
18978 @appendixsec The @code{current-kill} Function
18979 @findex current-kill
18981 The @code{current-kill} function changes the element in the kill ring
18982 to which @code{kill-ring-yank-pointer} points. (Also, the
18983 @code{kill-new} function sets @code{kill-ring-yank-pointer} to point
18984 to the latest element of the kill ring. The @code{kill-new}
18985 function is used directly or indirectly by @code{kill-append},
18986 @code{copy-region-as-kill}, @code{kill-ring-save}, @code{kill-line},
18987 and @code{kill-region}.)
18990 * Code for current-kill::
18991 * Understanding current-kill::
18995 @node Code for current-kill
18996 @unnumberedsubsec The code for @code{current-kill}
19001 The @code{current-kill} function is used by @code{yank} and by
19002 @code{yank-pop}. Here is the code for @code{current-kill}:
19006 (defun current-kill (n &optional do-not-move)
19007 "Rotate the yanking point by N places, and then return that kill.
19008 If N is zero, `interprogram-paste-function' is set, and calling it
19009 returns a string, then that string is added to the front of the
19010 kill ring and returned as the latest kill.
19013 If optional arg DO-NOT-MOVE is non-nil, then don't actually move the
19014 yanking point; just return the Nth kill forward."
19015 (let ((interprogram-paste (and (= n 0)
19016 interprogram-paste-function
19017 (funcall interprogram-paste-function))))
19020 (if interprogram-paste
19022 ;; Disable the interprogram cut function when we add the new
19023 ;; text to the kill ring, so Emacs doesn't try to own the
19024 ;; selection, with identical text.
19025 (let ((interprogram-cut-function nil))
19026 (kill-new interprogram-paste))
19027 interprogram-paste)
19030 (or kill-ring (error "Kill ring is empty"))
19031 (let ((ARGth-kill-element
19032 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19033 (length kill-ring))
19036 (setq kill-ring-yank-pointer ARGth-kill-element))
19037 (car ARGth-kill-element)))))
19041 Remember also that the @code{kill-new} function sets
19042 @code{kill-ring-yank-pointer} to the latest element of the kill
19043 ring, which means that all the functions that call it set the value
19044 indirectly: @code{kill-append}, @code{copy-region-as-kill},
19045 @code{kill-ring-save}, @code{kill-line}, and @code{kill-region}.
19048 Here is the line in @code{kill-new}, which is explained in
19049 @ref{kill-new function, , The @code{kill-new} function}.
19052 (setq kill-ring-yank-pointer kill-ring)
19056 @node Understanding current-kill
19057 @unnumberedsubsec @code{current-kill} in Outline
19060 The @code{current-kill} function looks complex, but as usual, it can
19061 be understood by taking it apart piece by piece. First look at it in
19066 (defun current-kill (n &optional do-not-move)
19067 "Rotate the yanking point by N places, and then return that kill."
19073 This function takes two arguments, one of which is optional. It has a
19074 documentation string. It is @emph{not} interactive.
19077 * Body of current-kill::
19078 * Digression concerning error:: How to mislead humans, but not computers.
19079 * Determining the Element::
19083 @node Body of current-kill
19084 @unnumberedsubsubsec The Body of @code{current-kill}
19087 The body of the function definition is a @code{let} expression, which
19088 itself has a body as well as a @var{varlist}.
19090 The @code{let} expression declares a variable that will be only usable
19091 within the bounds of this function. This variable is called
19092 @code{interprogram-paste} and is for copying to another program. It
19093 is not for copying within this instance of GNU Emacs. Most window
19094 systems provide a facility for interprogram pasting. Sadly, that
19095 facility usually provides only for the last element. Most windowing
19096 systems have not adopted a ring of many possibilities, even though
19097 Emacs has provided it for decades.
19099 The @code{if} expression has two parts, one if there exists
19100 @code{interprogram-paste} and one if not.
19103 Let us consider the `if not' or else-part of the @code{current-kill}
19104 function. (The then-part uses the @code{kill-new} function, which
19105 we have already described. @xref{kill-new function, , The
19106 @code{kill-new} function}.)
19110 (or kill-ring (error "Kill ring is empty"))
19111 (let ((ARGth-kill-element
19112 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19113 (length kill-ring))
19116 (setq kill-ring-yank-pointer ARGth-kill-element))
19117 (car ARGth-kill-element))
19122 The code first checks whether the kill ring has content; otherwise it
19126 Note that the @code{or} expression is very similar to testing length
19133 (if (zerop (length kill-ring)) ; @r{if-part}
19134 (error "Kill ring is empty")) ; @r{then-part}
19140 If there is not anything in the kill ring, its length must be zero and
19141 an error message sent to the user: @samp{Kill ring is empty}. The
19142 @code{current-kill} function uses an @code{or} expression which is
19143 simpler. But an @code{if} expression reminds us what goes on.
19145 This @code{if} expression uses the function @code{zerop} which returns
19146 true if the value it is testing is zero. When @code{zerop} tests
19147 true, the then-part of the @code{if} is evaluated. The then-part is a
19148 list starting with the function @code{error}, which is a function that
19149 is similar to the @code{message} function
19150 (@pxref{message, , The @code{message} Function}) in that
19151 it prints a one-line message in the echo area. However, in addition
19152 to printing a message, @code{error} also stops evaluation of the
19153 function within which it is embedded. This means that the rest of the
19154 function will not be evaluated if the length of the kill ring is zero.
19156 Then the @code{current-kill} function selects the element to return.
19157 The selection depends on the number of places that @code{current-kill}
19158 rotates and on where @code{kill-ring-yank-pointer} points.
19160 Next, either the optional @code{do-not-move} argument is true or the
19161 current value of @code{kill-ring-yank-pointer} is set to point to the
19162 list. Finally, another expression returns the first element of the
19163 list even if the @code{do-not-move} argument is true.
19166 @node Digression concerning error
19167 @unnumberedsubsubsec Digression about the word `error'
19170 In my opinion, it is slightly misleading, at least to humans, to use
19171 the term `error' as the name of the @code{error} function. A better
19172 term would be `cancel'. Strictly speaking, of course, you cannot
19173 point to, much less rotate a pointer to a list that has no length, so
19174 from the point of view of the computer, the word `error' is correct.
19175 But a human expects to attempt this sort of thing, if only to find out
19176 whether the kill ring is full or empty. This is an act of
19179 From the human point of view, the act of exploration and discovery is
19180 not necessarily an error, and therefore should not be labeled as one,
19181 even in the bowels of a computer. As it is, the code in Emacs implies
19182 that a human who is acting virtuously, by exploring his or her
19183 environment, is making an error. This is bad. Even though the computer
19184 takes the same steps as it does when there is an `error', a term such as
19185 `cancel' would have a clearer connotation.
19188 @node Determining the Element
19189 @unnumberedsubsubsec Determining the Element
19192 Among other actions, the else-part of the @code{if} expression sets
19193 the value of @code{kill-ring-yank-pointer} to
19194 @code{ARGth-kill-element} when the kill ring has something in it and
19195 the value of @code{do-not-move} is @code{nil}.
19198 The code looks like this:
19202 (nthcdr (mod (- n (length kill-ring-yank-pointer))
19203 (length kill-ring))
19208 This needs some examination. Unless it is not supposed to move the
19209 pointer, the @code{current-kill} function changes where
19210 @code{kill-ring-yank-pointer} points.
19212 @w{@code{(setq kill-ring-yank-pointer ARGth-kill-element))}}
19213 expression does. Also, clearly, @code{ARGth-kill-element} is being
19214 set to be equal to some @sc{cdr} of the kill ring, using the
19215 @code{nthcdr} function that is described in an earlier section.
19216 (@xref{copy-region-as-kill}.) How does it do this?
19218 As we have seen before (@pxref{nthcdr}), the @code{nthcdr} function
19219 works by repeatedly taking the @sc{cdr} of a list---it takes the
19220 @sc{cdr} of the @sc{cdr} of the @sc{cdr} @dots{}
19223 The two following expressions produce the same result:
19227 (setq kill-ring-yank-pointer (cdr kill-ring))
19229 (setq kill-ring-yank-pointer (nthcdr 1 kill-ring))
19233 However, the @code{nthcdr} expression is more complicated. It uses
19234 the @code{mod} function to determine which @sc{cdr} to select.
19236 (You will remember to look at inner functions first; indeed, we will
19237 have to go inside the @code{mod}.)
19239 The @code{mod} function returns the value of its first argument modulo
19240 the second; that is to say, it returns the remainder after dividing
19241 the first argument by the second. The value returned has the same
19242 sign as the second argument.
19250 @result{} 0 ;; @r{because there is no remainder}
19257 In this case, the first argument is often smaller than the second.
19269 We can guess what the @code{-} function does. It is like @code{+} but
19270 subtracts instead of adds; the @code{-} function subtracts its second
19271 argument from its first. Also, we already know what the @code{length}
19272 function does (@pxref{length}). It returns the length of a list.
19274 And @code{n} is the name of the required argument to the
19275 @code{current-kill} function.
19278 So when the first argument to @code{nthcdr} is zero, the @code{nthcdr}
19279 expression returns the whole list, as you can see by evaluating the
19284 ;; kill-ring-yank-pointer @r{and} kill-ring @r{have a length of four}
19285 ;; @r{and} (mod (- 0 4) 4) @result{} 0
19286 (nthcdr (mod (- 0 4) 4)
19287 '("fourth line of text"
19289 "second piece of text"
19290 "first some text"))
19295 When the first argument to the @code{current-kill} function is one,
19296 the @code{nthcdr} expression returns the list without its first
19301 (nthcdr (mod (- 1 4) 4)
19302 '("fourth line of text"
19304 "second piece of text"
19305 "first some text"))
19309 @cindex @samp{global variable} defined
19310 @cindex @samp{variable, global}, defined
19311 Incidentally, both @code{kill-ring} and @code{kill-ring-yank-pointer}
19312 are @dfn{global variables}. That means that any expression in Emacs
19313 Lisp can access them. They are not like the local variables set by
19314 @code{let} or like the symbols in an argument list.
19315 Local variables can only be accessed
19316 within the @code{let} that defines them or the function that specifies
19317 them in an argument list (and within expressions called by them).
19320 @c texi2dvi fails when the name of the section is within ifnottex ...
19321 (@xref{Prevent confusion, , @code{let} Prevents Confusion}, and
19322 @ref{defun, , The @code{defun} Macro}.)
19326 @appendixsec @code{yank}
19329 After learning about @code{current-kill}, the code for the
19330 @code{yank} function is almost easy.
19332 The @code{yank} function does not use the
19333 @code{kill-ring-yank-pointer} variable directly. It calls
19334 @code{insert-for-yank} which calls @code{current-kill} which sets the
19335 @code{kill-ring-yank-pointer} variable.
19338 The code looks like this:
19343 (defun yank (&optional arg)
19344 "Reinsert (\"paste\") the last stretch of killed text.
19345 More precisely, reinsert the stretch of killed text most recently
19346 killed OR yanked. Put point at end, and set mark at beginning.
19347 With just \\[universal-argument] as argument, same but put point at
19348 beginning (and mark at end). With argument N, reinsert the Nth most
19349 recently killed stretch of killed text.
19351 When this command inserts killed text into the buffer, it honors
19352 `yank-excluded-properties' and `yank-handler' as described in the
19353 doc string for `insert-for-yank-1', which see.
19355 See also the command \\[yank-pop]."
19359 (setq yank-window-start (window-start))
19360 ;; If we don't get all the way thru, make last-command indicate that
19361 ;; for the following command.
19362 (setq this-command t)
19363 (push-mark (point))
19366 (insert-for-yank (current-kill (cond
19371 ;; This is like exchange-point-and-mark,
19372 ;; but doesn't activate the mark.
19373 ;; It is cleaner to avoid activation, even though the command
19374 ;; loop would deactivate the mark because we inserted text.
19375 (goto-char (prog1 (mark t)
19376 (set-marker (mark-marker) (point) (current-buffer)))))
19379 ;; If we do get all the way thru, make this-command indicate that.
19380 (if (eq this-command t)
19381 (setq this-command 'yank))
19386 The key expression is @code{insert-for-yank}, which inserts the string
19387 returned by @code{current-kill}, but removes some text properties from
19390 However, before getting to that expression, the function sets the value
19391 of @code{yank-window-start} to the position returned by the
19392 @code{(window-start)} expression, the position at which the display
19393 currently starts. The @code{yank} function also sets
19394 @code{this-command} and pushes the mark.
19396 After it yanks the appropriate element, if the optional argument is a
19397 @sc{cons} rather than a number or nothing, it puts point at beginning
19398 of the yanked text and mark at its end.
19400 (The @code{prog1} function is like @code{progn} but returns the value
19401 of its first argument rather than the value of its last argument. Its
19402 first argument is forced to return the buffer's mark as an integer.
19403 You can see the documentation for these functions by placing point
19404 over them in this buffer and then typing @kbd{C-h f}
19405 (@code{describe-function}) followed by a @kbd{RET}; the default is the
19408 The last part of the function tells what to do when it succeeds.
19411 @appendixsec @code{yank-pop}
19414 After understanding @code{yank} and @code{current-kill}, you know how
19415 to approach the @code{yank-pop} function. Leaving out the
19416 documentation to save space, it looks like this:
19421 (defun yank-pop (&optional arg)
19424 (if (not (eq last-command 'yank))
19425 (error "Previous command was not a yank"))
19428 (setq this-command 'yank)
19429 (unless arg (setq arg 1))
19430 (let ((inhibit-read-only t)
19431 (before (< (point) (mark t))))
19435 (funcall (or yank-undo-function 'delete-region) (point) (mark t))
19436 (funcall (or yank-undo-function 'delete-region) (mark t) (point)))
19437 (setq yank-undo-function nil)
19440 (set-marker (mark-marker) (point) (current-buffer))
19441 (insert-for-yank (current-kill arg))
19442 ;; Set the window start back where it was in the yank command,
19444 (set-window-start (selected-window) yank-window-start t)
19448 ;; This is like exchange-point-and-mark,
19449 ;; but doesn't activate the mark.
19450 ;; It is cleaner to avoid activation, even though the command
19451 ;; loop would deactivate the mark because we inserted text.
19452 (goto-char (prog1 (mark t)
19453 (set-marker (mark-marker)
19455 (current-buffer))))))
19460 The function is interactive with a small @samp{p} so the prefix
19461 argument is processed and passed to the function. The command can
19462 only be used after a previous yank; otherwise an error message is
19463 sent. This check uses the variable @code{last-command} which is set
19464 by @code{yank} and is discussed elsewhere.
19465 (@xref{copy-region-as-kill}.)
19467 The @code{let} clause sets the variable @code{before} to true or false
19468 depending whether point is before or after mark and then the region
19469 between point and mark is deleted. This is the region that was just
19470 inserted by the previous yank and it is this text that will be
19473 @code{funcall} calls its first argument as a function, passing
19474 remaining arguments to it. The first argument is whatever the
19475 @code{or} expression returns. The two remaining arguments are the
19476 positions of point and mark set by the preceding @code{yank} command.
19478 There is more, but that is the hardest part.
19481 @appendixsec The @file{ring.el} File
19482 @cindex @file{ring.el} file
19484 Interestingly, GNU Emacs posses a file called @file{ring.el} that
19485 provides many of the features we just discussed. But functions such
19486 as @code{kill-ring-yank-pointer} do not use this library, possibly
19487 because they were written earlier.
19490 @appendix A Graph with Labeled Axes
19492 Printed axes help you understand a graph. They convey scale. In an
19493 earlier chapter (@pxref{Readying a Graph, , Readying a Graph}), we
19494 wrote the code to print the body of a graph. Here we write the code
19495 for printing and labeling vertical and horizontal axes, along with the
19499 * Labeled Example::
19500 * print-graph Varlist:: @code{let} expression in @code{print-graph}.
19501 * print-Y-axis:: Print a label for the vertical axis.
19502 * print-X-axis:: Print a horizontal label.
19503 * Print Whole Graph:: The function to print a complete graph.
19507 @node Labeled Example
19508 @unnumberedsec Labeled Example Graph
19511 Since insertions fill a buffer to the right and below point, the new
19512 graph printing function should first print the Y or vertical axis,
19513 then the body of the graph, and finally the X or horizontal axis.
19514 This sequence lays out for us the contents of the function:
19524 Print body of graph.
19531 Here is an example of how a finished graph should look:
19544 1 - ****************
19551 In this graph, both the vertical and the horizontal axes are labeled
19552 with numbers. However, in some graphs, the horizontal axis is time
19553 and would be better labeled with months, like this:
19567 Indeed, with a little thought, we can easily come up with a variety of
19568 vertical and horizontal labeling schemes. Our task could become
19569 complicated. But complications breed confusion. Rather than permit
19570 this, it is better choose a simple labeling scheme for our first
19571 effort, and to modify or replace it later.
19574 These considerations suggest the following outline for the
19575 @code{print-graph} function:
19579 (defun print-graph (numbers-list)
19580 "@var{documentation}@dots{}"
19581 (let ((height @dots{}
19585 (print-Y-axis height @dots{} )
19586 (graph-body-print numbers-list)
19587 (print-X-axis @dots{} )))
19591 We can work on each part of the @code{print-graph} function definition
19594 @node print-graph Varlist
19595 @appendixsec The @code{print-graph} Varlist
19596 @cindex @code{print-graph} varlist
19598 In writing the @code{print-graph} function, the first task is to write
19599 the varlist in the @code{let} expression. (We will leave aside for the
19600 moment any thoughts about making the function interactive or about the
19601 contents of its documentation string.)
19603 The varlist should set several values. Clearly, the top of the label
19604 for the vertical axis must be at least the height of the graph, which
19605 means that we must obtain this information here. Note that the
19606 @code{print-graph-body} function also requires this information. There
19607 is no reason to calculate the height of the graph in two different
19608 places, so we should change @code{print-graph-body} from the way we
19609 defined it earlier to take advantage of the calculation.
19611 Similarly, both the function for printing the X axis labels and the
19612 @code{print-graph-body} function need to learn the value of the width of
19613 each symbol. We can perform the calculation here and change the
19614 definition for @code{print-graph-body} from the way we defined it in the
19617 The length of the label for the horizontal axis must be at least as long
19618 as the graph. However, this information is used only in the function
19619 that prints the horizontal axis, so it does not need to be calculated here.
19621 These thoughts lead us directly to the following form for the varlist
19622 in the @code{let} for @code{print-graph}:
19626 (let ((height (apply 'max numbers-list)) ; @r{First version.}
19627 (symbol-width (length graph-blank)))
19632 As we shall see, this expression is not quite right.
19636 @appendixsec The @code{print-Y-axis} Function
19637 @cindex Axis, print vertical
19638 @cindex Y axis printing
19639 @cindex Vertical axis printing
19640 @cindex Print vertical axis
19642 The job of the @code{print-Y-axis} function is to print a label for
19643 the vertical axis that looks like this:
19661 The function should be passed the height of the graph, and then should
19662 construct and insert the appropriate numbers and marks.
19665 * print-Y-axis in Detail::
19666 * Height of label:: What height for the Y axis?
19667 * Compute a Remainder:: How to compute the remainder of a division.
19668 * Y Axis Element:: Construct a line for the Y axis.
19669 * Y-axis-column:: Generate a list of Y axis labels.
19670 * print-Y-axis Penultimate:: A not quite final version.
19674 @node print-Y-axis in Detail
19675 @unnumberedsubsec The @code{print-Y-axis} Function in Detail
19678 It is easy enough to see in the figure what the Y axis label should
19679 look like; but to say in words, and then to write a function
19680 definition to do the job is another matter. It is not quite true to
19681 say that we want a number and a tic every five lines: there are only
19682 three lines between the @samp{1} and the @samp{5} (lines 2, 3, and 4),
19683 but four lines between the @samp{5} and the @samp{10} (lines 6, 7, 8,
19684 and 9). It is better to say that we want a number and a tic mark on
19685 the base line (number 1) and then that we want a number and a tic on
19686 the fifth line from the bottom and on every line that is a multiple of
19690 @node Height of label
19691 @unnumberedsubsec What height should the label be?
19694 The next issue is what height the label should be? Suppose the maximum
19695 height of tallest column of the graph is seven. Should the highest
19696 label on the Y axis be @samp{5 -}, and should the graph stick up above
19697 the label? Or should the highest label be @samp{7 -}, and mark the peak
19698 of the graph? Or should the highest label be @code{10 -}, which is a
19699 multiple of five, and be higher than the topmost value of the graph?
19701 The latter form is preferred. Most graphs are drawn within rectangles
19702 whose sides are an integral number of steps long---5, 10, 15, and so
19703 on for a step distance of five. But as soon as we decide to use a
19704 step height for the vertical axis, we discover that the simple
19705 expression in the varlist for computing the height is wrong. The
19706 expression is @code{(apply 'max numbers-list)}. This returns the
19707 precise height, not the maximum height plus whatever is necessary to
19708 round up to the nearest multiple of five. A more complex expression
19711 As usual in cases like this, a complex problem becomes simpler if it is
19712 divided into several smaller problems.
19714 First, consider the case when the highest value of the graph is an
19715 integral multiple of five---when it is 5, 10, 15, or some higher
19716 multiple of five. We can use this value as the Y axis height.
19718 A fairly simply way to determine whether a number is a multiple of
19719 five is to divide it by five and see if the division results in a
19720 remainder. If there is no remainder, the number is a multiple of
19721 five. Thus, seven divided by five has a remainder of two, and seven
19722 is not an integral multiple of five. Put in slightly different
19723 language, more reminiscent of the classroom, five goes into seven
19724 once, with a remainder of two. However, five goes into ten twice,
19725 with no remainder: ten is an integral multiple of five.
19727 @node Compute a Remainder
19728 @appendixsubsec Side Trip: Compute a Remainder
19730 @findex % @r{(remainder function)}
19731 @cindex Remainder function, @code{%}
19732 In Lisp, the function for computing a remainder is @code{%}. The
19733 function returns the remainder of its first argument divided by its
19734 second argument. As it happens, @code{%} is a function in Emacs Lisp
19735 that you cannot discover using @code{apropos}: you find nothing if you
19736 type @kbd{M-x apropos @key{RET} remainder @key{RET}}. The only way to
19737 learn of the existence of @code{%} is to read about it in a book such
19738 as this or in the Emacs Lisp sources.
19740 You can try the @code{%} function by evaluating the following two
19752 The first expression returns 2 and the second expression returns 0.
19754 To test whether the returned value is zero or some other number, we
19755 can use the @code{zerop} function. This function returns @code{t} if
19756 its argument, which must be a number, is zero.
19768 Thus, the following expression will return @code{t} if the height
19769 of the graph is evenly divisible by five:
19772 (zerop (% height 5))
19776 (The value of @code{height}, of course, can be found from @code{(apply
19777 'max numbers-list)}.)
19779 On the other hand, if the value of @code{height} is not a multiple of
19780 five, we want to reset the value to the next higher multiple of five.
19781 This is straightforward arithmetic using functions with which we are
19782 already familiar. First, we divide the value of @code{height} by five
19783 to determine how many times five goes into the number. Thus, five
19784 goes into twelve twice. If we add one to this quotient and multiply by
19785 five, we will obtain the value of the next multiple of five that is
19786 larger than the height. Five goes into twelve twice. Add one to two,
19787 and multiply by five; the result is fifteen, which is the next multiple
19788 of five that is higher than twelve. The Lisp expression for this is:
19791 (* (1+ (/ height 5)) 5)
19795 For example, if you evaluate the following, the result is 15:
19798 (* (1+ (/ 12 5)) 5)
19801 All through this discussion, we have been using `five' as the value
19802 for spacing labels on the Y axis; but we may want to use some other
19803 value. For generality, we should replace `five' with a variable to
19804 which we can assign a value. The best name I can think of for this
19805 variable is @code{Y-axis-label-spacing}.
19808 Using this term, and an @code{if} expression, we produce the
19813 (if (zerop (% height Y-axis-label-spacing))
19816 (* (1+ (/ height Y-axis-label-spacing))
19817 Y-axis-label-spacing))
19822 This expression returns the value of @code{height} itself if the height
19823 is an even multiple of the value of the @code{Y-axis-label-spacing} or
19824 else it computes and returns a value of @code{height} that is equal to
19825 the next higher multiple of the value of the @code{Y-axis-label-spacing}.
19827 We can now include this expression in the @code{let} expression of the
19828 @code{print-graph} function (after first setting the value of
19829 @code{Y-axis-label-spacing}):
19830 @vindex Y-axis-label-spacing
19834 (defvar Y-axis-label-spacing 5
19835 "Number of lines from one Y axis label to next.")
19840 (let* ((height (apply 'max numbers-list))
19841 (height-of-top-line
19842 (if (zerop (% height Y-axis-label-spacing))
19847 (* (1+ (/ height Y-axis-label-spacing))
19848 Y-axis-label-spacing)))
19849 (symbol-width (length graph-blank))))
19855 (Note use of the @code{let*} function: the initial value of height is
19856 computed once by the @code{(apply 'max numbers-list)} expression and
19857 then the resulting value of @code{height} is used to compute its
19858 final value. @xref{fwd-para let, , The @code{let*} expression}, for
19859 more about @code{let*}.)
19861 @node Y Axis Element
19862 @appendixsubsec Construct a Y Axis Element
19864 When we print the vertical axis, we want to insert strings such as
19865 @w{@samp{5 -}} and @w{@samp{10 - }} every five lines.
19866 Moreover, we want the numbers and dashes to line up, so shorter
19867 numbers must be padded with leading spaces. If some of the strings
19868 use two digit numbers, the strings with single digit numbers must
19869 include a leading blank space before the number.
19871 @findex number-to-string
19872 To figure out the length of the number, the @code{length} function is
19873 used. But the @code{length} function works only with a string, not with
19874 a number. So the number has to be converted from being a number to
19875 being a string. This is done with the @code{number-to-string} function.
19880 (length (number-to-string 35))
19883 (length (number-to-string 100))
19889 (@code{number-to-string} is also called @code{int-to-string}; you will
19890 see this alternative name in various sources.)
19892 In addition, in each label, each number is followed by a string such
19893 as @w{@samp{ - }}, which we will call the @code{Y-axis-tic} marker.
19894 This variable is defined with @code{defvar}:
19899 (defvar Y-axis-tic " - "
19900 "String that follows number in a Y axis label.")
19904 The length of the Y label is the sum of the length of the Y axis tic
19905 mark and the length of the number of the top of the graph.
19908 (length (concat (number-to-string height) Y-axis-tic)))
19911 This value will be calculated by the @code{print-graph} function in
19912 its varlist as @code{full-Y-label-width} and passed on. (Note that we
19913 did not think to include this in the varlist when we first proposed it.)
19915 To make a complete vertical axis label, a tic mark is concatenated
19916 with a number; and the two together may be preceded by one or more
19917 spaces depending on how long the number is. The label consists of
19918 three parts: the (optional) leading spaces, the number, and the tic
19919 mark. The function is passed the value of the number for the specific
19920 row, and the value of the width of the top line, which is calculated
19921 (just once) by @code{print-graph}.
19925 (defun Y-axis-element (number full-Y-label-width)
19926 "Construct a NUMBERed label element.
19927 A numbered element looks like this ` 5 - ',
19928 and is padded as needed so all line up with
19929 the element for the largest number."
19932 (let* ((leading-spaces
19933 (- full-Y-label-width
19935 (concat (number-to-string number)
19940 (make-string leading-spaces ? )
19941 (number-to-string number)
19946 The @code{Y-axis-element} function concatenates together the leading
19947 spaces, if any; the number, as a string; and the tic mark.
19949 To figure out how many leading spaces the label will need, the
19950 function subtracts the actual length of the label---the length of the
19951 number plus the length of the tic mark---from the desired label width.
19953 @findex make-string
19954 Blank spaces are inserted using the @code{make-string} function. This
19955 function takes two arguments: the first tells it how long the string
19956 will be and the second is a symbol for the character to insert, in a
19957 special format. The format is a question mark followed by a blank
19958 space, like this, @samp{? }. @xref{Character Type, , Character Type,
19959 elisp, The GNU Emacs Lisp Reference Manual}, for a description of the
19960 syntax for characters. (Of course, you might want to replace the
19961 blank space by some other character @dots{} You know what to do.)
19963 The @code{number-to-string} function is used in the concatenation
19964 expression, to convert the number to a string that is concatenated
19965 with the leading spaces and the tic mark.
19967 @node Y-axis-column
19968 @appendixsubsec Create a Y Axis Column
19970 The preceding functions provide all the tools needed to construct a
19971 function that generates a list of numbered and blank strings to insert
19972 as the label for the vertical axis:
19974 @findex Y-axis-column
19977 (defun Y-axis-column (height width-of-label)
19978 "Construct list of Y axis labels and blank strings.
19979 For HEIGHT of line above base and WIDTH-OF-LABEL."
19983 (while (> height 1)
19984 (if (zerop (% height Y-axis-label-spacing))
19985 ;; @r{Insert label.}
19988 (Y-axis-element height width-of-label)
19992 ;; @r{Else, insert blanks.}
19995 (make-string width-of-label ? )
19997 (setq height (1- height)))
19998 ;; @r{Insert base line.}
20000 (cons (Y-axis-element 1 width-of-label) Y-axis))
20001 (nreverse Y-axis)))
20005 In this function, we start with the value of @code{height} and
20006 repetitively subtract one from its value. After each subtraction, we
20007 test to see whether the value is an integral multiple of the
20008 @code{Y-axis-label-spacing}. If it is, we construct a numbered label
20009 using the @code{Y-axis-element} function; if not, we construct a
20010 blank label using the @code{make-string} function. The base line
20011 consists of the number one followed by a tic mark.
20014 @node print-Y-axis Penultimate
20015 @appendixsubsec The Not Quite Final Version of @code{print-Y-axis}
20017 The list constructed by the @code{Y-axis-column} function is passed to
20018 the @code{print-Y-axis} function, which inserts the list as a column.
20020 @findex print-Y-axis
20023 (defun print-Y-axis (height full-Y-label-width)
20024 "Insert Y axis using HEIGHT and FULL-Y-LABEL-WIDTH.
20025 Height must be the maximum height of the graph.
20026 Full width is the width of the highest label element."
20027 ;; Value of height and full-Y-label-width
20028 ;; are passed by `print-graph'.
20031 (let ((start (point)))
20033 (Y-axis-column height full-Y-label-width))
20034 ;; @r{Place point ready for inserting graph.}
20036 ;; @r{Move point forward by value of} full-Y-label-width
20037 (forward-char full-Y-label-width)))
20041 The @code{print-Y-axis} uses the @code{insert-rectangle} function to
20042 insert the Y axis labels created by the @code{Y-axis-column} function.
20043 In addition, it places point at the correct position for printing the body of
20046 You can test @code{print-Y-axis}:
20054 Y-axis-label-spacing
20063 Copy the following expression:
20066 (print-Y-axis 12 5)
20070 Switch to the @file{*scratch*} buffer and place the cursor where you
20071 want the axis labels to start.
20074 Type @kbd{M-:} (@code{eval-expression}).
20077 Yank the @code{graph-body-print} expression into the minibuffer
20078 with @kbd{C-y} (@code{yank)}.
20081 Press @key{RET} to evaluate the expression.
20084 Emacs will print labels vertically, the top one being @w{@samp{10 -@w{
20085 }}}. (The @code{print-graph} function will pass the value of
20086 @code{height-of-top-line}, which in this case will end up as 15,
20087 thereby getting rid of what might appear as a bug.)
20091 @appendixsec The @code{print-X-axis} Function
20092 @cindex Axis, print horizontal
20093 @cindex X axis printing
20094 @cindex Print horizontal axis
20095 @cindex Horizontal axis printing
20097 X axis labels are much like Y axis labels, except that the ticks are on a
20098 line above the numbers. Labels should look like this:
20107 The first tic is under the first column of the graph and is preceded by
20108 several blank spaces. These spaces provide room in rows above for the Y
20109 axis labels. The second, third, fourth, and subsequent ticks are all
20110 spaced equally, according to the value of @code{X-axis-label-spacing}.
20112 The second row of the X axis consists of numbers, preceded by several
20113 blank spaces and also separated according to the value of the variable
20114 @code{X-axis-label-spacing}.
20116 The value of the variable @code{X-axis-label-spacing} should itself be
20117 measured in units of @code{symbol-width}, since you may want to change
20118 the width of the symbols that you are using to print the body of the
20119 graph without changing the ways the graph is labeled.
20122 * Similarities differences:: Much like @code{print-Y-axis}, but not exactly.
20123 * X Axis Tic Marks:: Create tic marks for the horizontal axis.
20127 @node Similarities differences
20128 @unnumberedsubsec Similarities and differences
20131 The @code{print-X-axis} function is constructed in more or less the
20132 same fashion as the @code{print-Y-axis} function except that it has
20133 two lines: the line of tic marks and the numbers. We will write a
20134 separate function to print each line and then combine them within the
20135 @code{print-X-axis} function.
20137 This is a three step process:
20141 Write a function to print the X axis tic marks, @code{print-X-axis-tic-line}.
20144 Write a function to print the X numbers, @code{print-X-axis-numbered-line}.
20147 Write a function to print both lines, the @code{print-X-axis} function,
20148 using @code{print-X-axis-tic-line} and
20149 @code{print-X-axis-numbered-line}.
20152 @node X Axis Tic Marks
20153 @appendixsubsec X Axis Tic Marks
20155 The first function should print the X axis tic marks. We must specify
20156 the tic marks themselves and their spacing:
20160 (defvar X-axis-label-spacing
20161 (if (boundp 'graph-blank)
20162 (* 5 (length graph-blank)) 5)
20163 "Number of units from one X axis label to next.")
20168 (Note that the value of @code{graph-blank} is set by another
20169 @code{defvar}. The @code{boundp} predicate checks whether it has
20170 already been set; @code{boundp} returns @code{nil} if it has not. If
20171 @code{graph-blank} were unbound and we did not use this conditional
20172 construction, in a recent GNU Emacs, we would enter the debugger and
20173 see an error message saying @samp{@w{Debugger entered--Lisp error:}
20174 @w{(void-variable graph-blank)}}.)
20177 Here is the @code{defvar} for @code{X-axis-tic-symbol}:
20181 (defvar X-axis-tic-symbol "|"
20182 "String to insert to point to a column in X axis.")
20187 The goal is to make a line that looks like this:
20193 The first tic is indented so that it is under the first column, which is
20194 indented to provide space for the Y axis labels.
20196 A tic element consists of the blank spaces that stretch from one tic to
20197 the next plus a tic symbol. The number of blanks is determined by the
20198 width of the tic symbol and the @code{X-axis-label-spacing}.
20201 The code looks like this:
20205 ;;; X-axis-tic-element
20209 ;; @r{Make a string of blanks.}
20210 (- (* symbol-width X-axis-label-spacing)
20211 (length X-axis-tic-symbol))
20213 ;; @r{Concatenate blanks with tic symbol.}
20219 Next, we determine how many blanks are needed to indent the first tic
20220 mark to the first column of the graph. This uses the value of
20221 @code{full-Y-label-width} passed it by the @code{print-graph} function.
20224 The code to make @code{X-axis-leading-spaces}
20229 ;; X-axis-leading-spaces
20231 (make-string full-Y-label-width ? )
20236 We also need to determine the length of the horizontal axis, which is
20237 the length of the numbers list, and the number of ticks in the horizontal
20244 (length numbers-list)
20250 (* symbol-width X-axis-label-spacing)
20254 ;; number-of-X-ticks
20255 (if (zerop (% (X-length tic-width)))
20256 (/ (X-length tic-width))
20257 (1+ (/ (X-length tic-width))))
20262 All this leads us directly to the function for printing the X axis tic line:
20264 @findex print-X-axis-tic-line
20267 (defun print-X-axis-tic-line
20268 (number-of-X-tics X-axis-leading-spaces X-axis-tic-element)
20269 "Print ticks for X axis."
20270 (insert X-axis-leading-spaces)
20271 (insert X-axis-tic-symbol) ; @r{Under first column.}
20274 ;; @r{Insert second tic in the right spot.}
20277 (- (* symbol-width X-axis-label-spacing)
20278 ;; @r{Insert white space up to second tic symbol.}
20279 (* 2 (length X-axis-tic-symbol)))
20281 X-axis-tic-symbol))
20284 ;; @r{Insert remaining ticks.}
20285 (while (> number-of-X-tics 1)
20286 (insert X-axis-tic-element)
20287 (setq number-of-X-tics (1- number-of-X-tics))))
20291 The line of numbers is equally straightforward:
20294 First, we create a numbered element with blank spaces before each number:
20296 @findex X-axis-element
20299 (defun X-axis-element (number)
20300 "Construct a numbered X axis element."
20301 (let ((leading-spaces
20302 (- (* symbol-width X-axis-label-spacing)
20303 (length (number-to-string number)))))
20304 (concat (make-string leading-spaces ? )
20305 (number-to-string number))))
20309 Next, we create the function to print the numbered line, starting with
20310 the number ``1'' under the first column:
20312 @findex print-X-axis-numbered-line
20315 (defun print-X-axis-numbered-line
20316 (number-of-X-tics X-axis-leading-spaces)
20317 "Print line of X-axis numbers"
20318 (let ((number X-axis-label-spacing))
20319 (insert X-axis-leading-spaces)
20325 ;; @r{Insert white space up to next number.}
20326 (- (* symbol-width X-axis-label-spacing) 2)
20328 (number-to-string number)))
20331 ;; @r{Insert remaining numbers.}
20332 (setq number (+ number X-axis-label-spacing))
20333 (while (> number-of-X-tics 1)
20334 (insert (X-axis-element number))
20335 (setq number (+ number X-axis-label-spacing))
20336 (setq number-of-X-tics (1- number-of-X-tics)))))
20340 Finally, we need to write the @code{print-X-axis} that uses
20341 @code{print-X-axis-tic-line} and
20342 @code{print-X-axis-numbered-line}.
20344 The function must determine the local values of the variables used by both
20345 @code{print-X-axis-tic-line} and @code{print-X-axis-numbered-line}, and
20346 then it must call them. Also, it must print the carriage return that
20347 separates the two lines.
20349 The function consists of a varlist that specifies five local variables,
20350 and calls to each of the two line printing functions:
20352 @findex print-X-axis
20355 (defun print-X-axis (numbers-list)
20356 "Print X axis labels to length of NUMBERS-LIST."
20357 (let* ((leading-spaces
20358 (make-string full-Y-label-width ? ))
20361 ;; symbol-width @r{is provided by} graph-body-print
20362 (tic-width (* symbol-width X-axis-label-spacing))
20363 (X-length (length numbers-list))
20371 ;; @r{Make a string of blanks.}
20372 (- (* symbol-width X-axis-label-spacing)
20373 (length X-axis-tic-symbol))
20377 ;; @r{Concatenate blanks with tic symbol.}
20378 X-axis-tic-symbol))
20382 (if (zerop (% X-length tic-width))
20383 (/ X-length tic-width)
20384 (1+ (/ X-length tic-width)))))
20387 (print-X-axis-tic-line tic-number leading-spaces X-tic)
20389 (print-X-axis-numbered-line tic-number leading-spaces)))
20394 You can test @code{print-X-axis}:
20398 Install @code{X-axis-tic-symbol}, @code{X-axis-label-spacing},
20399 @code{print-X-axis-tic-line}, as well as @code{X-axis-element},
20400 @code{print-X-axis-numbered-line}, and @code{print-X-axis}.
20403 Copy the following expression:
20408 (let ((full-Y-label-width 5)
20411 '(1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16))))
20416 Switch to the @file{*scratch*} buffer and place the cursor where you
20417 want the axis labels to start.
20420 Type @kbd{M-:} (@code{eval-expression}).
20423 Yank the test expression into the minibuffer
20424 with @kbd{C-y} (@code{yank)}.
20427 Press @key{RET} to evaluate the expression.
20431 Emacs will print the horizontal axis like this:
20441 @node Print Whole Graph
20442 @appendixsec Printing the Whole Graph
20443 @cindex Printing the whole graph
20444 @cindex Whole graph printing
20445 @cindex Graph, printing all
20447 Now we are nearly ready to print the whole graph.
20449 The function to print the graph with the proper labels follows the
20450 outline we created earlier (@pxref{Full Graph, , A Graph with Labeled
20451 Axes}), but with additions.
20454 Here is the outline:
20458 (defun print-graph (numbers-list)
20459 "@var{documentation}@dots{}"
20460 (let ((height @dots{}
20464 (print-Y-axis height @dots{} )
20465 (graph-body-print numbers-list)
20466 (print-X-axis @dots{} )))
20471 * The final version:: A few changes.
20472 * Test print-graph:: Run a short test.
20473 * Graphing words in defuns:: Executing the final code.
20474 * lambda:: How to write an anonymous function.
20475 * mapcar:: Apply a function to elements of a list.
20476 * Another Bug:: Yet another bug @dots{} most insidious.
20477 * Final printed graph:: The graph itself!
20481 @node The final version
20482 @unnumberedsubsec Changes for the Final Version
20485 The final version is different from what we planned in two ways:
20486 first, it contains additional values calculated once in the varlist;
20487 second, it carries an option to specify the labels' increment per row.
20488 This latter feature turns out to be essential; otherwise, a graph may
20489 have more rows than fit on a display or on a sheet of paper.
20492 This new feature requires a change to the @code{Y-axis-column}
20493 function, to add @code{vertical-step} to it. The function looks like
20496 @findex Y-axis-column @r{Final version.}
20499 ;;; @r{Final version.}
20500 (defun Y-axis-column
20501 (height width-of-label &optional vertical-step)
20502 "Construct list of labels for Y axis.
20503 HEIGHT is maximum height of graph.
20504 WIDTH-OF-LABEL is maximum width of label.
20505 VERTICAL-STEP, an option, is a positive integer
20506 that specifies how much a Y axis label increments
20507 for each line. For example, a step of 5 means
20508 that each line is five units of the graph."
20512 (number-per-line (or vertical-step 1)))
20513 (while (> height 1)
20514 (if (zerop (% height Y-axis-label-spacing))
20517 ;; @r{Insert label.}
20521 (* height number-per-line)
20526 ;; @r{Else, insert blanks.}
20529 (make-string width-of-label ? )
20531 (setq height (1- height)))
20534 ;; @r{Insert base line.}
20535 (setq Y-axis (cons (Y-axis-element
20536 (or vertical-step 1)
20539 (nreverse Y-axis)))
20543 The values for the maximum height of graph and the width of a symbol
20544 are computed by @code{print-graph} in its @code{let} expression; so
20545 @code{graph-body-print} must be changed to accept them.
20547 @findex graph-body-print @r{Final version.}
20550 ;;; @r{Final version.}
20551 (defun graph-body-print (numbers-list height symbol-width)
20552 "Print a bar graph of the NUMBERS-LIST.
20553 The numbers-list consists of the Y-axis values.
20554 HEIGHT is maximum height of graph.
20555 SYMBOL-WIDTH is number of each column."
20558 (let (from-position)
20559 (while numbers-list
20560 (setq from-position (point))
20562 (column-of-graph height (car numbers-list)))
20563 (goto-char from-position)
20564 (forward-char symbol-width)
20567 ;; @r{Draw graph column by column.}
20569 (setq numbers-list (cdr numbers-list)))
20570 ;; @r{Place point for X axis labels.}
20571 (forward-line height)
20577 Finally, the code for the @code{print-graph} function:
20579 @findex print-graph @r{Final version.}
20582 ;;; @r{Final version.}
20584 (numbers-list &optional vertical-step)
20585 "Print labeled bar graph of the NUMBERS-LIST.
20586 The numbers-list consists of the Y-axis values.
20590 Optionally, VERTICAL-STEP, a positive integer,
20591 specifies how much a Y axis label increments for
20592 each line. For example, a step of 5 means that
20593 each row is five units."
20596 (let* ((symbol-width (length graph-blank))
20597 ;; @code{height} @r{is both the largest number}
20598 ;; @r{and the number with the most digits.}
20599 (height (apply 'max numbers-list))
20602 (height-of-top-line
20603 (if (zerop (% height Y-axis-label-spacing))
20606 (* (1+ (/ height Y-axis-label-spacing))
20607 Y-axis-label-spacing)))
20610 (vertical-step (or vertical-step 1))
20611 (full-Y-label-width
20617 (* height-of-top-line vertical-step))
20623 height-of-top-line full-Y-label-width vertical-step)
20627 numbers-list height-of-top-line symbol-width)
20628 (print-X-axis numbers-list)))
20632 @node Test print-graph
20633 @appendixsubsec Testing @code{print-graph}
20636 We can test the @code{print-graph} function with a short list of numbers:
20640 Install the final versions of @code{Y-axis-column},
20641 @code{graph-body-print}, and @code{print-graph} (in addition to the
20645 Copy the following expression:
20648 (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1))
20652 Switch to the @file{*scratch*} buffer and place the cursor where you
20653 want the axis labels to start.
20656 Type @kbd{M-:} (@code{eval-expression}).
20659 Yank the test expression into the minibuffer
20660 with @kbd{C-y} (@code{yank)}.
20663 Press @key{RET} to evaluate the expression.
20667 Emacs will print a graph that looks like this:
20688 On the other hand, if you pass @code{print-graph} a
20689 @code{vertical-step} value of 2, by evaluating this expression:
20692 (print-graph '(3 2 5 6 7 5 3 4 6 4 3 2 1) 2)
20697 The graph looks like this:
20718 (A question: is the `2' on the bottom of the vertical axis a bug or a
20719 feature? If you think it is a bug, and should be a `1' instead, (or
20720 even a `0'), you can modify the sources.)
20722 @node Graphing words in defuns
20723 @appendixsubsec Graphing Numbers of Words and Symbols
20725 Now for the graph for which all this code was written: a graph that
20726 shows how many function definitions contain fewer than 10 words and
20727 symbols, how many contain between 10 and 19 words and symbols, how
20728 many contain between 20 and 29 words and symbols, and so on.
20730 This is a multi-step process. First make sure you have loaded all the
20734 It is a good idea to reset the value of @code{top-of-ranges} in case
20735 you have set it to some different value. You can evaluate the
20740 (setq top-of-ranges
20743 110 120 130 140 150
20744 160 170 180 190 200
20745 210 220 230 240 250
20746 260 270 280 290 300)
20751 Next create a list of the number of words and symbols in each range.
20755 Evaluate the following:
20759 (setq list-for-graph
20762 (recursive-lengths-list-many-files
20763 (directory-files "/usr/local/emacs/lisp"
20771 On my old machine, this took about an hour. It looked though 303 Lisp
20772 files in my copy of Emacs version 19.23. After all that computing,
20773 the @code{list-for-graph} had this value:
20777 (537 1027 955 785 594 483 349 292 224 199 166 120 116 99
20778 90 80 67 48 52 45 41 33 28 26 25 20 12 28 11 13 220)
20783 This means that my copy of Emacs had 537 function definitions with
20784 fewer than 10 words or symbols in them, 1,027 function definitions
20785 with 10 to 19 words or symbols in them, 955 function definitions with
20786 20 to 29 words or symbols in them, and so on.
20788 Clearly, just by looking at this list we can see that most function
20789 definitions contain ten to thirty words and symbols.
20791 Now for printing. We do @emph{not} want to print a graph that is
20792 1,030 lines high @dots{} Instead, we should print a graph that is
20793 fewer than twenty-five lines high. A graph that height can be
20794 displayed on almost any monitor, and easily printed on a sheet of paper.
20796 This means that each value in @code{list-for-graph} must be reduced to
20797 one-fiftieth its present value.
20799 Here is a short function to do just that, using two functions we have
20800 not yet seen, @code{mapcar} and @code{lambda}.
20804 (defun one-fiftieth (full-range)
20805 "Return list, each number one-fiftieth of previous."
20806 (mapcar (lambda (arg) (/ arg 50)) full-range))
20811 @appendixsubsec A @code{lambda} Expression: Useful Anonymity
20812 @cindex Anonymous function
20815 @code{lambda} is the symbol for an anonymous function, a function
20816 without a name. Every time you use an anonymous function, you need to
20817 include its whole body.
20824 (lambda (arg) (/ arg 50))
20828 is a function definition that says `return the value resulting from
20829 dividing whatever is passed to me as @code{arg} by 50'.
20832 Earlier, for example, we had a function @code{multiply-by-seven}; it
20833 multiplied its argument by 7. This function is similar, except it
20834 divides its argument by 50; and, it has no name. The anonymous
20835 equivalent of @code{multiply-by-seven} is:
20838 (lambda (number) (* 7 number))
20842 (@xref{defun, , The @code{defun} Macro}.)
20846 If we want to multiply 3 by 7, we can write:
20848 @c clear print-postscript-figures
20849 @c lambda example diagram #1
20853 (multiply-by-seven 3)
20854 \_______________/ ^
20860 @ifset print-postscript-figures
20863 @center @image{lambda-1}
20867 @ifclear print-postscript-figures
20871 (multiply-by-seven 3)
20872 \_______________/ ^
20881 This expression returns 21.
20885 Similarly, we can write:
20887 @c lambda example diagram #2
20891 ((lambda (number) (* 7 number)) 3)
20892 \____________________________/ ^
20894 anonymous function argument
20898 @ifset print-postscript-figures
20901 @center @image{lambda-2}
20905 @ifclear print-postscript-figures
20909 ((lambda (number) (* 7 number)) 3)
20910 \____________________________/ ^
20912 anonymous function argument
20920 If we want to divide 100 by 50, we can write:
20922 @c lambda example diagram #3
20926 ((lambda (arg) (/ arg 50)) 100)
20927 \______________________/ \_/
20929 anonymous function argument
20933 @ifset print-postscript-figures
20936 @center @image{lambda-3}
20940 @ifclear print-postscript-figures
20944 ((lambda (arg) (/ arg 50)) 100)
20945 \______________________/ \_/
20947 anonymous function argument
20954 This expression returns 2. The 100 is passed to the function, which
20955 divides that number by 50.
20957 @xref{Lambda Expressions, , Lambda Expressions, elisp, The GNU Emacs
20958 Lisp Reference Manual}, for more about @code{lambda}. Lisp and lambda
20959 expressions derive from the Lambda Calculus.
20962 @appendixsubsec The @code{mapcar} Function
20965 @code{mapcar} is a function that calls its first argument with each
20966 element of its second argument, in turn. The second argument must be
20969 The @samp{map} part of the name comes from the mathematical phrase,
20970 `mapping over a domain', meaning to apply a function to each of the
20971 elements in a domain. The mathematical phrase is based on the
20972 metaphor of a surveyor walking, one step at a time, over an area he is
20973 mapping. And @samp{car}, of course, comes from the Lisp notion of the
20982 (mapcar '1+ '(2 4 6))
20988 The function @code{1+} which adds one to its argument, is executed on
20989 @emph{each} element of the list, and a new list is returned.
20991 Contrast this with @code{apply}, which applies its first argument to
20993 (@xref{Readying a Graph, , Readying a Graph}, for a explanation of
20997 In the definition of @code{one-fiftieth}, the first argument is the
20998 anonymous function:
21001 (lambda (arg) (/ arg 50))
21005 and the second argument is @code{full-range}, which will be bound to
21006 @code{list-for-graph}.
21009 The whole expression looks like this:
21012 (mapcar (lambda (arg) (/ arg 50)) full-range))
21015 @xref{Mapping Functions, , Mapping Functions, elisp, The GNU Emacs
21016 Lisp Reference Manual}, for more about @code{mapcar}.
21018 Using the @code{one-fiftieth} function, we can generate a list in
21019 which each element is one-fiftieth the size of the corresponding
21020 element in @code{list-for-graph}.
21024 (setq fiftieth-list-for-graph
21025 (one-fiftieth list-for-graph))
21030 The resulting list looks like this:
21034 (10 20 19 15 11 9 6 5 4 3 3 2 2
21035 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 4)
21040 This, we are almost ready to print! (We also notice the loss of
21041 information: many of the higher ranges are 0, meaning that fewer than
21042 50 defuns had that many words or symbols---but not necessarily meaning
21043 that none had that many words or symbols.)
21046 @appendixsubsec Another Bug @dots{} Most Insidious
21047 @cindex Bug, most insidious type
21048 @cindex Insidious type of bug
21050 I said `almost ready to print'! Of course, there is a bug in the
21051 @code{print-graph} function @dots{} It has a @code{vertical-step}
21052 option, but not a @code{horizontal-step} option. The
21053 @code{top-of-range} scale goes from 10 to 300 by tens. But the
21054 @code{print-graph} function will print only by ones.
21056 This is a classic example of what some consider the most insidious
21057 type of bug, the bug of omission. This is not the kind of bug you can
21058 find by studying the code, for it is not in the code; it is an omitted
21059 feature. Your best actions are to try your program early and often;
21060 and try to arrange, as much as you can, to write code that is easy to
21061 understand and easy to change. Try to be aware, whenever you can,
21062 that whatever you have written, @emph{will} be rewritten, if not soon,
21063 eventually. A hard maxim to follow.
21065 It is the @code{print-X-axis-numbered-line} function that needs the
21066 work; and then the @code{print-X-axis} and the @code{print-graph}
21067 functions need to be adapted. Not much needs to be done; there is one
21068 nicety: the numbers ought to line up under the tic marks. This takes
21072 Here is the corrected @code{print-X-axis-numbered-line}:
21076 (defun print-X-axis-numbered-line
21077 (number-of-X-tics X-axis-leading-spaces
21078 &optional horizontal-step)
21079 "Print line of X-axis numbers"
21080 (let ((number X-axis-label-spacing)
21081 (horizontal-step (or horizontal-step 1)))
21084 (insert X-axis-leading-spaces)
21085 ;; @r{Delete extra leading spaces.}
21088 (length (number-to-string horizontal-step)))))
21093 ;; @r{Insert white space.}
21095 X-axis-label-spacing)
21098 (number-to-string horizontal-step)))
21102 (* number horizontal-step))))
21105 ;; @r{Insert remaining numbers.}
21106 (setq number (+ number X-axis-label-spacing))
21107 (while (> number-of-X-tics 1)
21108 (insert (X-axis-element
21109 (* number horizontal-step)))
21110 (setq number (+ number X-axis-label-spacing))
21111 (setq number-of-X-tics (1- number-of-X-tics)))))
21116 If you are reading this in Info, you can see the new versions of
21117 @code{print-X-axis} @code{print-graph} and evaluate them. If you are
21118 reading this in a printed book, you can see the changed lines here
21119 (the full text is too much to print).
21124 (defun print-X-axis (numbers-list horizontal-step)
21126 (print-X-axis-numbered-line
21127 tic-number leading-spaces horizontal-step))
21135 &optional vertical-step horizontal-step)
21137 (print-X-axis numbers-list horizontal-step))
21145 (defun print-X-axis (numbers-list horizontal-step)
21146 "Print X axis labels to length of NUMBERS-LIST.
21147 Optionally, HORIZONTAL-STEP, a positive integer,
21148 specifies how much an X axis label increments for
21152 ;; Value of symbol-width and full-Y-label-width
21153 ;; are passed by `print-graph'.
21154 (let* ((leading-spaces
21155 (make-string full-Y-label-width ? ))
21156 ;; symbol-width @r{is provided by} graph-body-print
21157 (tic-width (* symbol-width X-axis-label-spacing))
21158 (X-length (length numbers-list))
21164 ;; @r{Make a string of blanks.}
21165 (- (* symbol-width X-axis-label-spacing)
21166 (length X-axis-tic-symbol))
21170 ;; @r{Concatenate blanks with tic symbol.}
21171 X-axis-tic-symbol))
21173 (if (zerop (% X-length tic-width))
21174 (/ X-length tic-width)
21175 (1+ (/ X-length tic-width)))))
21179 (print-X-axis-tic-line
21180 tic-number leading-spaces X-tic)
21182 (print-X-axis-numbered-line
21183 tic-number leading-spaces horizontal-step)))
21190 (numbers-list &optional vertical-step horizontal-step)
21191 "Print labeled bar graph of the NUMBERS-LIST.
21192 The numbers-list consists of the Y-axis values.
21196 Optionally, VERTICAL-STEP, a positive integer,
21197 specifies how much a Y axis label increments for
21198 each line. For example, a step of 5 means that
21199 each row is five units.
21203 Optionally, HORIZONTAL-STEP, a positive integer,
21204 specifies how much an X axis label increments for
21206 (let* ((symbol-width (length graph-blank))
21207 ;; @code{height} @r{is both the largest number}
21208 ;; @r{and the number with the most digits.}
21209 (height (apply 'max numbers-list))
21212 (height-of-top-line
21213 (if (zerop (% height Y-axis-label-spacing))
21216 (* (1+ (/ height Y-axis-label-spacing))
21217 Y-axis-label-spacing)))
21220 (vertical-step (or vertical-step 1))
21221 (full-Y-label-width
21225 (* height-of-top-line vertical-step))
21230 height-of-top-line full-Y-label-width vertical-step)
21232 numbers-list height-of-top-line symbol-width)
21233 (print-X-axis numbers-list horizontal-step)))
21240 Graphing Definitions Re-listed
21243 Here are all the graphing definitions in their final form:
21247 (defvar top-of-ranges
21250 110 120 130 140 150
21251 160 170 180 190 200
21252 210 220 230 240 250)
21253 "List specifying ranges for `defuns-per-range'.")
21257 (defvar graph-symbol "*"
21258 "String used as symbol in graph, usually an asterisk.")
21262 (defvar graph-blank " "
21263 "String used as blank in graph, usually a blank space.
21264 graph-blank must be the same number of columns wide
21269 (defvar Y-axis-tic " - "
21270 "String that follows number in a Y axis label.")
21274 (defvar Y-axis-label-spacing 5
21275 "Number of lines from one Y axis label to next.")
21279 (defvar X-axis-tic-symbol "|"
21280 "String to insert to point to a column in X axis.")
21284 (defvar X-axis-label-spacing
21285 (if (boundp 'graph-blank)
21286 (* 5 (length graph-blank)) 5)
21287 "Number of units from one X axis label to next.")
21293 (defun count-words-in-defun ()
21294 "Return the number of words and symbols in a defun."
21295 (beginning-of-defun)
21297 (end (save-excursion (end-of-defun) (point))))
21302 (and (< (point) end)
21304 "\\(\\w\\|\\s_\\)+[^ \t\n]*[ \t\n]*"
21306 (setq count (1+ count)))
21313 (defun lengths-list-file (filename)
21314 "Return list of definitions' lengths within FILE.
21315 The returned list is a list of numbers.
21316 Each number is the number of words or
21317 symbols in one function definition."
21321 (message "Working on `%s' ... " filename)
21323 (let ((buffer (find-file-noselect filename))
21325 (set-buffer buffer)
21326 (setq buffer-read-only t)
21328 (goto-char (point-min))
21332 (while (re-search-forward "^(defun" nil t)
21334 (cons (count-words-in-defun) lengths-list)))
21335 (kill-buffer buffer)
21342 (defun lengths-list-many-files (list-of-files)
21343 "Return list of lengths of defuns in LIST-OF-FILES."
21344 (let (lengths-list)
21345 ;;; @r{true-or-false-test}
21346 (while list-of-files
21352 ;;; @r{Generate a lengths' list.}
21354 (expand-file-name (car list-of-files)))))
21355 ;;; @r{Make files' list shorter.}
21356 (setq list-of-files (cdr list-of-files)))
21357 ;;; @r{Return final value of lengths' list.}
21364 (defun defuns-per-range (sorted-lengths top-of-ranges)
21365 "SORTED-LENGTHS defuns in each TOP-OF-RANGES range."
21366 (let ((top-of-range (car top-of-ranges))
21367 (number-within-range 0)
21368 defuns-per-range-list)
21373 (while top-of-ranges
21377 ;; @r{Need number for numeric test.}
21378 (car sorted-lengths)
21379 (< (car sorted-lengths) top-of-range))
21381 ;; @r{Count number of definitions within current range.}
21382 (setq number-within-range (1+ number-within-range))
21383 (setq sorted-lengths (cdr sorted-lengths)))
21387 ;; @r{Exit inner loop but remain within outer loop.}
21389 (setq defuns-per-range-list
21390 (cons number-within-range defuns-per-range-list))
21391 (setq number-within-range 0) ; @r{Reset count to zero.}
21393 ;; @r{Move to next range.}
21394 (setq top-of-ranges (cdr top-of-ranges))
21395 ;; @r{Specify next top of range value.}
21396 (setq top-of-range (car top-of-ranges)))
21400 ;; @r{Exit outer loop and count the number of defuns larger than}
21401 ;; @r{ the largest top-of-range value.}
21402 (setq defuns-per-range-list
21404 (length sorted-lengths)
21405 defuns-per-range-list))
21407 ;; @r{Return a list of the number of definitions within each range,}
21408 ;; @r{ smallest to largest.}
21409 (nreverse defuns-per-range-list)))
21415 (defun column-of-graph (max-graph-height actual-height)
21416 "Return list of MAX-GRAPH-HEIGHT strings;
21417 ACTUAL-HEIGHT are graph-symbols.
21418 The graph-symbols are contiguous entries at the end
21420 The list will be inserted as one column of a graph.
21421 The strings are either graph-blank or graph-symbol."
21425 (let ((insert-list nil)
21426 (number-of-top-blanks
21427 (- max-graph-height actual-height)))
21429 ;; @r{Fill in @code{graph-symbols}.}
21430 (while (> actual-height 0)
21431 (setq insert-list (cons graph-symbol insert-list))
21432 (setq actual-height (1- actual-height)))
21436 ;; @r{Fill in @code{graph-blanks}.}
21437 (while (> number-of-top-blanks 0)
21438 (setq insert-list (cons graph-blank insert-list))
21439 (setq number-of-top-blanks
21440 (1- number-of-top-blanks)))
21442 ;; @r{Return whole list.}
21449 (defun Y-axis-element (number full-Y-label-width)
21450 "Construct a NUMBERed label element.
21451 A numbered element looks like this ` 5 - ',
21452 and is padded as needed so all line up with
21453 the element for the largest number."
21456 (let* ((leading-spaces
21457 (- full-Y-label-width
21459 (concat (number-to-string number)
21464 (make-string leading-spaces ? )
21465 (number-to-string number)
21472 (defun print-Y-axis
21473 (height full-Y-label-width &optional vertical-step)
21474 "Insert Y axis by HEIGHT and FULL-Y-LABEL-WIDTH.
21475 Height must be the maximum height of the graph.
21476 Full width is the width of the highest label element.
21477 Optionally, print according to VERTICAL-STEP."
21480 ;; Value of height and full-Y-label-width
21481 ;; are passed by `print-graph'.
21482 (let ((start (point)))
21484 (Y-axis-column height full-Y-label-width vertical-step))
21487 ;; @r{Place point ready for inserting graph.}
21489 ;; @r{Move point forward by value of} full-Y-label-width
21490 (forward-char full-Y-label-width)))
21496 (defun print-X-axis-tic-line
21497 (number-of-X-tics X-axis-leading-spaces X-axis-tic-element)
21498 "Print ticks for X axis."
21499 (insert X-axis-leading-spaces)
21500 (insert X-axis-tic-symbol) ; @r{Under first column.}
21503 ;; @r{Insert second tic in the right spot.}
21506 (- (* symbol-width X-axis-label-spacing)
21507 ;; @r{Insert white space up to second tic symbol.}
21508 (* 2 (length X-axis-tic-symbol)))
21510 X-axis-tic-symbol))
21513 ;; @r{Insert remaining ticks.}
21514 (while (> number-of-X-tics 1)
21515 (insert X-axis-tic-element)
21516 (setq number-of-X-tics (1- number-of-X-tics))))
21522 (defun X-axis-element (number)
21523 "Construct a numbered X axis element."
21524 (let ((leading-spaces
21525 (- (* symbol-width X-axis-label-spacing)
21526 (length (number-to-string number)))))
21527 (concat (make-string leading-spaces ? )
21528 (number-to-string number))))
21534 (defun graph-body-print (numbers-list height symbol-width)
21535 "Print a bar graph of the NUMBERS-LIST.
21536 The numbers-list consists of the Y-axis values.
21537 HEIGHT is maximum height of graph.
21538 SYMBOL-WIDTH is number of each column."
21541 (let (from-position)
21542 (while numbers-list
21543 (setq from-position (point))
21545 (column-of-graph height (car numbers-list)))
21546 (goto-char from-position)
21547 (forward-char symbol-width)
21550 ;; @r{Draw graph column by column.}
21552 (setq numbers-list (cdr numbers-list)))
21553 ;; @r{Place point for X axis labels.}
21554 (forward-line height)
21561 (defun Y-axis-column
21562 (height width-of-label &optional vertical-step)
21563 "Construct list of labels for Y axis.
21564 HEIGHT is maximum height of graph.
21565 WIDTH-OF-LABEL is maximum width of label.
21568 VERTICAL-STEP, an option, is a positive integer
21569 that specifies how much a Y axis label increments
21570 for each line. For example, a step of 5 means
21571 that each line is five units of the graph."
21573 (number-per-line (or vertical-step 1)))
21576 (while (> height 1)
21577 (if (zerop (% height Y-axis-label-spacing))
21578 ;; @r{Insert label.}
21582 (* height number-per-line)
21587 ;; @r{Else, insert blanks.}
21590 (make-string width-of-label ? )
21592 (setq height (1- height)))
21595 ;; @r{Insert base line.}
21596 (setq Y-axis (cons (Y-axis-element
21597 (or vertical-step 1)
21600 (nreverse Y-axis)))
21606 (defun print-X-axis-numbered-line
21607 (number-of-X-tics X-axis-leading-spaces
21608 &optional horizontal-step)
21609 "Print line of X-axis numbers"
21610 (let ((number X-axis-label-spacing)
21611 (horizontal-step (or horizontal-step 1)))
21614 (insert X-axis-leading-spaces)
21616 (delete-char (- (1- (length (number-to-string horizontal-step)))))
21619 ;; @r{Insert white space up to next number.}
21620 (- (* symbol-width X-axis-label-spacing)
21621 (1- (length (number-to-string horizontal-step)))
21624 (number-to-string (* number horizontal-step))))
21627 ;; @r{Insert remaining numbers.}
21628 (setq number (+ number X-axis-label-spacing))
21629 (while (> number-of-X-tics 1)
21630 (insert (X-axis-element (* number horizontal-step)))
21631 (setq number (+ number X-axis-label-spacing))
21632 (setq number-of-X-tics (1- number-of-X-tics)))))
21638 (defun print-X-axis (numbers-list horizontal-step)
21639 "Print X axis labels to length of NUMBERS-LIST.
21640 Optionally, HORIZONTAL-STEP, a positive integer,
21641 specifies how much an X axis label increments for
21645 ;; Value of symbol-width and full-Y-label-width
21646 ;; are passed by `print-graph'.
21647 (let* ((leading-spaces
21648 (make-string full-Y-label-width ? ))
21649 ;; symbol-width @r{is provided by} graph-body-print
21650 (tic-width (* symbol-width X-axis-label-spacing))
21651 (X-length (length numbers-list))
21657 ;; @r{Make a string of blanks.}
21658 (- (* symbol-width X-axis-label-spacing)
21659 (length X-axis-tic-symbol))
21663 ;; @r{Concatenate blanks with tic symbol.}
21664 X-axis-tic-symbol))
21666 (if (zerop (% X-length tic-width))
21667 (/ X-length tic-width)
21668 (1+ (/ X-length tic-width)))))
21672 (print-X-axis-tic-line
21673 tic-number leading-spaces X-tic)
21675 (print-X-axis-numbered-line
21676 tic-number leading-spaces horizontal-step)))
21682 (defun one-fiftieth (full-range)
21683 "Return list, each number of which is 1/50th previous."
21684 (mapcar (lambda (arg) (/ arg 50)) full-range))
21691 (numbers-list &optional vertical-step horizontal-step)
21692 "Print labeled bar graph of the NUMBERS-LIST.
21693 The numbers-list consists of the Y-axis values.
21697 Optionally, VERTICAL-STEP, a positive integer,
21698 specifies how much a Y axis label increments for
21699 each line. For example, a step of 5 means that
21700 each row is five units.
21704 Optionally, HORIZONTAL-STEP, a positive integer,
21705 specifies how much an X axis label increments for
21707 (let* ((symbol-width (length graph-blank))
21708 ;; @code{height} @r{is both the largest number}
21709 ;; @r{and the number with the most digits.}
21710 (height (apply 'max numbers-list))
21713 (height-of-top-line
21714 (if (zerop (% height Y-axis-label-spacing))
21717 (* (1+ (/ height Y-axis-label-spacing))
21718 Y-axis-label-spacing)))
21721 (vertical-step (or vertical-step 1))
21722 (full-Y-label-width
21726 (* height-of-top-line vertical-step))
21732 height-of-top-line full-Y-label-width vertical-step)
21734 numbers-list height-of-top-line symbol-width)
21735 (print-X-axis numbers-list horizontal-step)))
21742 @node Final printed graph
21743 @appendixsubsec The Printed Graph
21745 When made and installed, you can call the @code{print-graph} command
21751 (print-graph fiftieth-list-for-graph 50 10)
21781 50 - ***************** * *
21783 10 50 100 150 200 250 300 350
21790 The largest group of functions contain 10--19 words and symbols each.
21792 @node Free Software and Free Manuals
21793 @appendix Free Software and Free Manuals
21795 @strong{by Richard M. Stallman}
21798 The biggest deficiency in free operating systems is not in the
21799 software---it is the lack of good free manuals that we can include in
21800 these systems. Many of our most important programs do not come with
21801 full manuals. Documentation is an essential part of any software
21802 package; when an important free software package does not come with a
21803 free manual, that is a major gap. We have many such gaps today.
21805 Once upon a time, many years ago, I thought I would learn Perl. I got
21806 a copy of a free manual, but I found it hard to read. When I asked
21807 Perl users about alternatives, they told me that there were better
21808 introductory manuals---but those were not free.
21810 Why was this? The authors of the good manuals had written them for
21811 O'Reilly Associates, which published them with restrictive terms---no
21812 copying, no modification, source files not available---which exclude
21813 them from the free software community.
21815 That wasn't the first time this sort of thing has happened, and (to
21816 our community's great loss) it was far from the last. Proprietary
21817 manual publishers have enticed a great many authors to restrict their
21818 manuals since then. Many times I have heard a GNU user eagerly tell me
21819 about a manual that he is writing, with which he expects to help the
21820 GNU project---and then had my hopes dashed, as he proceeded to explain
21821 that he had signed a contract with a publisher that would restrict it
21822 so that we cannot use it.
21824 Given that writing good English is a rare skill among programmers, we
21825 can ill afford to lose manuals this way.
21827 Free documentation, like free software, is a matter of freedom, not
21828 price. The problem with these manuals was not that O'Reilly Associates
21829 charged a price for printed copies---that in itself is fine. The Free
21830 Software Foundation @uref{http://shop.fsf.org, sells printed copies} of
21831 free @uref{http://www.gnu.org/doc/doc.html, GNU manuals}, too.
21832 But GNU manuals are available in source code form, while these manuals
21833 are available only on paper. GNU manuals come with permission to copy
21834 and modify; the Perl manuals do not. These restrictions are the
21837 The criterion for a free manual is pretty much the same as for free
21838 software: it is a matter of giving all users certain
21839 freedoms. Redistribution (including commercial redistribution) must be
21840 permitted, so that the manual can accompany every copy of the program,
21841 on-line or on paper. Permission for modification is crucial too.
21843 As a general rule, I don't believe that it is essential for people to
21844 have permission to modify all sorts of articles and books. The issues
21845 for writings are not necessarily the same as those for software. For
21846 example, I don't think you or I are obliged to give permission to
21847 modify articles like this one, which describe our actions and our
21850 But there is a particular reason why the freedom to modify is crucial
21851 for documentation for free software. When people exercise their right
21852 to modify the software, and add or change its features, if they are
21853 conscientious they will change the manual too---so they can provide
21854 accurate and usable documentation with the modified program. A manual
21855 which forbids programmers to be conscientious and finish the job, or
21856 more precisely requires them to write a new manual from scratch if
21857 they change the program, does not fill our community's needs.
21859 While a blanket prohibition on modification is unacceptable, some
21860 kinds of limits on the method of modification pose no problem. For
21861 example, requirements to preserve the original author's copyright
21862 notice, the distribution terms, or the list of authors, are ok. It is
21863 also no problem to require modified versions to include notice that
21864 they were modified, even to have entire sections that may not be
21865 deleted or changed, as long as these sections deal with nontechnical
21866 topics. (Some GNU manuals have them.)
21868 These kinds of restrictions are not a problem because, as a practical
21869 matter, they don't stop the conscientious programmer from adapting the
21870 manual to fit the modified program. In other words, they don't block
21871 the free software community from making full use of the manual.
21873 However, it must be possible to modify all the technical content of
21874 the manual, and then distribute the result in all the usual media,
21875 through all the usual channels; otherwise, the restrictions do block
21876 the community, the manual is not free, and so we need another manual.
21878 Unfortunately, it is often hard to find someone to write another
21879 manual when a proprietary manual exists. The obstacle is that many
21880 users think that a proprietary manual is good enough---so they don't
21881 see the need to write a free manual. They do not see that the free
21882 operating system has a gap that needs filling.
21884 Why do users think that proprietary manuals are good enough? Some have
21885 not considered the issue. I hope this article will do something to
21888 Other users consider proprietary manuals acceptable for the same
21889 reason so many people consider proprietary software acceptable: they
21890 judge in purely practical terms, not using freedom as a
21891 criterion. These people are entitled to their opinions, but since
21892 those opinions spring from values which do not include freedom, they
21893 are no guide for those of us who do value freedom.
21895 Please spread the word about this issue. We continue to lose manuals
21896 to proprietary publishing. If we spread the word that proprietary
21897 manuals are not sufficient, perhaps the next person who wants to help
21898 GNU by writing documentation will realize, before it is too late, that
21899 he must above all make it free.
21901 We can also encourage commercial publishers to sell free, copylefted
21902 manuals instead of proprietary ones. One way you can help this is to
21903 check the distribution terms of a manual before you buy it, and prefer
21904 copylefted manuals to non-copylefted ones.
21908 Note: The Free Software Foundation maintains a page on its Web site
21909 that lists free books available from other publishers:@*
21910 @uref{http://www.gnu.org/doc/other-free-books.html}
21912 @node GNU Free Documentation License
21913 @appendix GNU Free Documentation License
21915 @cindex FDL, GNU Free Documentation License
21916 @include doclicense.texi
21922 MENU ENTRY: NODE NAME.
21928 @c Place biographical information on right-hand (verso) page
21931 \par\vfill\supereject
21933 \global\evenheadline={\hfil} \global\evenfootline={\hfil}
21934 \global\oddheadline={\hfil} \global\oddfootline={\hfil}
21937 % \par\vfill\supereject
21938 \global\evenheadline={\hfil} \global\evenfootline={\hfil}
21939 \global\oddheadline={\hfil} \global\oddfootline={\hfil}
21940 %\page\hbox{}%\page
21941 %\page\hbox{}%\page
21948 @c ================ Biographical information ================
21952 @center About the Author
21957 @node About the Author
21958 @unnumbered About the Author
21962 Robert J. Chassell has worked with GNU Emacs since 1985. He writes
21963 and edits, teaches Emacs and Emacs Lisp, and speaks throughout the
21964 world on software freedom. Chassell was a founding Director and
21965 Treasurer of the Free Software Foundation, Inc. He is co-author of
21966 the @cite{Texinfo} manual, and has edited more than a dozen other
21967 books. He graduated from Cambridge University, in England. He has an
21968 abiding interest in social and economic history and flies his own
21975 @c @c Prevent page number on blank verso, so eject it first.
21977 @c \par\vfill\supereject
21982 @c @evenheading @thispage @| @| @thistitle
21983 @c @oddheading @| @| @thispage