1 \input texinfo @c -*-texinfo-*-
3 @setfilename standards.info
4 @settitle GNU Coding Standards
5 @c This date is automagically updated when you save this file:
6 @set lastupdate June 12, 2020
9 @dircategory GNU organization
11 * Standards: (standards). GNU coding standards.
14 @c @setchapternewpage odd
15 @setchapternewpage off
17 @c Put everything in one index (arbitrarily chosen to be the concept index).
23 @c This is used by a cross ref in make-stds.texi
27 The GNU coding standards, last updated @value{lastupdate}.
29 Copyright @copyright{} 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
30 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
31 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020 Free
32 Software Foundation, Inc.
34 Permission is granted to copy, distribute and/or modify this document
35 under the terms of the GNU Free Documentation License, Version 1.3 or
36 any later version published by the Free Software Foundation; with no
37 Invariant Sections, no Front-Cover Texts, and no Back-Cover
38 Texts. A copy of the license is included in the section entitled
39 ``GNU Free Documentation License''.
43 @title GNU Coding Standards
44 @author Richard Stallman, et al.
45 @author last updated @value{lastupdate}
47 @vskip 0pt plus 1filll
55 @top GNU Coding Standards
61 * Preface:: About the GNU Coding Standards.
62 * Legal Issues:: Keeping free software free.
63 * Design Advice:: General program design.
64 * Program Behavior:: Program behavior for all programs
65 * Writing C:: Making the best use of C.
66 * Documentation:: Documenting programs.
67 * Managing Releases:: The release process.
68 * References:: Mentioning non-free software or documentation.
69 * GNU Free Documentation License:: Copying and sharing this manual.
75 @chapter About the GNU Coding Standards
77 The GNU Coding Standards were written by Richard Stallman and other GNU
78 Project volunteers. Their purpose is to make the GNU system clean,
79 consistent, and easy to install. This document can also be read as a
80 guide to writing portable, robust and reliable programs. It focuses on
81 programs written in C, but many of the rules and principles are useful
82 even if you write in another programming language. The rules often
83 state reasons for writing in a certain way.
85 @cindex where to obtain @code{standards.texi}
86 @cindex downloading this manual
87 If you did not obtain this file directly from the GNU project and
88 recently, please check for a newer version. You can get the GNU
89 Coding Standards from the GNU web server in many
90 different formats, including the Texinfo source, PDF, HTML, DVI, plain
91 text, and more, at: @uref{https://www.gnu.org/prep/standards/}.
93 If you are maintaining an official GNU package, in addition to this
94 document, please read and follow the GNU maintainer information
95 (@pxref{Top, , Contents, maintain, Information for Maintainers of GNU
98 @cindex @code{gnustandards-commit@@gnu.org} mailing list
99 If you want to receive diffs for every change to these GNU documents,
100 join the mailing list @code{gnustandards-commit@@gnu.org}, via the web
102 @url{https://lists.gnu.org/mailman/listinfo/gnustandards-commit}.
103 Archives are also available there.
105 @cindex @code{bug-standards@@gnu.org} email address
106 @cindex Savannah repository for gnustandards
107 @cindex gnustandards project repository
108 Please send corrections or suggestions for this document to
109 @email{bug-standards@@gnu.org}. If you make a suggestion, please
110 include a suggested new wording for it, to help us consider the
111 suggestion efficiently. We prefer a context diff to the Texinfo
112 source, but if that's difficult for you, you can make a context diff
113 for some other version of this document, or propose it in any way that
114 makes it clear. The source repository for this document can be found
115 at @url{https://savannah.gnu.org/projects/gnustandards}.
117 These standards cover the minimum of what is important when writing a
118 GNU package. Likely, the need for additional standards will come up.
119 Sometimes, you might suggest that such standards be added to this
120 document. If you think your standards would be generally useful, please
123 You should also set standards for your package on many questions not
124 addressed or not firmly specified here. The most important point is to
125 be self-consistent---try to stick to the conventions you pick, and try
126 to document them as much as possible. That way, your program will be
127 more maintainable by others.
129 The GNU Hello program serves as an example of how to follow the GNU
130 coding standards for a trivial program.
131 @uref{https://www.gnu.org/software/hello/hello.html}.
133 This release of the GNU Coding Standards was last updated
138 @chapter Keeping Free Software Free
139 @cindex legal aspects
141 This chapter discusses how you can make sure that GNU software
142 avoids legal difficulties, and other related issues.
145 * Reading Non-Free Code:: Referring to proprietary programs.
146 * Contributions:: Accepting contributions.
147 * Trademarks:: How we deal with trademark issues.
150 @node Reading Non-Free Code
151 @section Referring to Proprietary Programs
152 @cindex proprietary programs
153 @cindex avoiding proprietary code
155 Don't in any circumstances refer to Unix source code for or during
156 your work on GNU! (Or to any other proprietary programs.)
158 If you have a vague recollection of the internals of a Unix program,
159 this does not absolutely mean you can't write an imitation of it, but
160 do try to organize the imitation internally along different lines,
161 because this is likely to make the details of the Unix version
162 irrelevant and dissimilar to your results.
164 For example, Unix utilities were generally optimized to minimize
165 memory use; if you go for speed instead, your program will be very
166 different. You could keep the entire input file in memory and scan it
167 there instead of using stdio. Use a smarter algorithm discovered more
168 recently than the Unix program. Eliminate use of temporary files. Do
169 it in one pass instead of two (we did this in the assembler).
171 Or, on the contrary, emphasize simplicity instead of speed. For some
172 applications, the speed of today's computers makes simpler algorithms
175 Or go for generality. For example, Unix programs often have static
176 tables or fixed-size strings, which make for arbitrary limits; use
177 dynamic allocation instead. Make sure your program handles NULs and
178 other funny characters in the input files. Add a programming language
179 for extensibility and write part of the program in that language.
181 Or turn some parts of the program into independently usable libraries.
182 Or use a simple garbage collector instead of tracking precisely when
183 to free memory, or use a new GNU facility such as obstacks.
187 @section Accepting Contributions
189 @cindex accepting contributions
191 If the program you are working on is copyrighted by the Free Software
192 Foundation, then when someone else sends you a piece of code to add to
193 the program, we need legal papers to use it---just as we asked you to
194 sign papers initially. @emph{Each} person who makes a nontrivial
195 contribution to a program must sign some sort of legal papers in order
196 for us to have clear title to the program; the main author alone is not
199 So, before adding in any contributions from other people, please tell
200 us, so we can arrange to get the papers. Then wait until we tell you
201 that we have received the signed papers, before you actually use the
204 This applies both before you release the program and afterward. If
205 you receive diffs to fix a bug, and they make significant changes, we
206 need legal papers for that change.
208 This also applies to comments and documentation files. For copyright
209 law, comments and code are just text. Copyright applies to all kinds of
210 text, so we need legal papers for all kinds.
212 We know it is frustrating to ask for legal papers; it's frustrating for
213 us as well. But if you don't wait, you are going out on a limb---for
214 example, what if the contributor's employer won't sign a disclaimer?
215 You might have to take that code out again!
217 You don't need papers for changes of a few lines here or there, since
218 they are not significant for copyright purposes. Also, you don't need
219 papers if all you get from the suggestion is some ideas, not actual code
220 which you use. For example, if someone sent you one implementation, but
221 you write a different implementation of the same idea, you don't need to
224 The very worst thing is if you forget to tell us about the other
225 contributor. We could be very embarrassed in court some day as a
228 We have more detailed advice for maintainers of GNU packages. If you
229 have reached the stage of maintaining a GNU program (whether released
230 or not), please take a look: @pxref{Legal Matters,,, maintain,
231 Information for GNU Maintainers}.
238 Please do not include any trademark acknowledgments in GNU software
239 packages or documentation.
241 Trademark acknowledgments are the statements that such-and-such is a
242 trademark of so-and-so. The GNU Project has no objection to the basic
243 idea of trademarks, but these acknowledgments feel like kowtowing,
244 and there is no legal requirement for them, so we don't use them.
246 What is legally required, as regards other people's trademarks, is to
247 avoid using them in ways which a reader might reasonably understand as
248 naming or labeling our own programs or activities. For example, since
249 ``Objective C'' is (or at least was) a trademark, we made sure to say
250 that we provide a ``compiler for the Objective C language'' rather
251 than an ``Objective C compiler''. The latter would have been meant as
252 a shorter way of saying the former, but it does not explicitly state
253 the relationship, so it could be misinterpreted as using ``Objective
254 C'' as a label for the compiler rather than for the language.
256 Please don't use ``win'' as an abbreviation for Microsoft Windows in
257 GNU software or documentation. In hacker terminology, calling
258 something a ``win'' is a form of praise. You're free to praise
259 Microsoft Windows on your own if you want, but please don't do so in
260 GNU packages. Please write ``Windows'' in full, or abbreviate it to
261 ``w.'' @xref{System Portability}.
264 @chapter General Program Design
265 @cindex program design
267 This chapter discusses some of the issues you should take into
268 account when designing your program.
270 @c Standard or ANSI C
272 @c In 1989 the American National Standards Institute (ANSI) standardized
273 @c C as standard X3.159-1989. In December of that year the
274 @c International Standards Organization ISO adopted the ANSI C standard
275 @c making minor changes. In 1990 ANSI then re-adopted ISO standard
276 @c C. This version of C is known as either ANSI C or Standard C.
278 @c A major revision of the C Standard appeared in 1999.
281 * Source Language:: Which languages to use.
282 * Compatibility:: Compatibility with other implementations.
283 * Using Extensions:: Using non-standard features.
284 * Standard C:: Using standard C features.
285 * Conditional Compilation:: Compiling code only if a conditional is true.
288 @node Source Language
289 @section Which Languages to Use
290 @cindex programming languages
292 When you want to use a language that gets compiled and runs at high
293 speed, the best language to use is C@. C++ is ok too, but please don't
294 make heavy use of templates. So is Java, if you compile it.
296 When highest efficiency is not required, other languages commonly used
297 in the free software community, such as Lisp, Scheme, Python, Ruby, and
298 Java, are OK too. Scheme, as implemented by GNU@tie{}Guile, plays a
299 particular role in the GNU System: it is the preferred language to
300 extend programs written in C/C++, and also a fine language for a wide
301 range of applications. The more GNU components use Guile and Scheme,
302 the more users are able to extend and combine them (@pxref{The Emacs
303 Thesis,,, guile, GNU Guile Reference Manual}).
305 Many programs are designed to be extensible: they include an interpreter
306 for a language that is higher level than C@. Often much of the program
307 is written in that language, too. The Emacs editor pioneered this
311 @cindex GNOME and Guile
312 The standard extensibility interpreter for GNU software is Guile
313 (@uref{https://www.gnu.org/@/software/@/guile/}), which implements the
314 language Scheme (an especially clean and simple dialect of Lisp).
315 Guile also includes bindings for GTK+/GNOME, making it practical to
316 write modern GUI functionality within Guile. We don't reject programs
317 written in other ``scripting languages'' such as Perl and Python, but
318 using Guile is the path that will lead to overall consistency of the
323 @section Compatibility with Other Implementations
324 @cindex compatibility with C and POSIX standards
325 @cindex C compatibility
326 @cindex POSIX compatibility
328 With occasional exceptions, utility programs and libraries for GNU
329 should be upward compatible with those in Berkeley Unix, and upward
330 compatible with Standard C if Standard C specifies their
331 behavior, and upward compatible with POSIX if POSIX specifies
334 When these standards conflict, it is useful to offer compatibility
335 modes for each of them.
337 @cindex options for compatibility
338 Standard C and POSIX prohibit many kinds of extensions. Feel
339 free to make the extensions anyway, and include a @samp{--ansi},
340 @samp{--posix}, or @samp{--compatible} option to turn them off.
341 However, if the extension has a significant chance of breaking any real
342 programs or scripts, then it is not really upward compatible. So you
343 should try to redesign its interface to make it upward compatible.
345 @cindex @code{POSIXLY_CORRECT}, environment variable
346 Many GNU programs suppress extensions that conflict with POSIX if the
347 environment variable @code{POSIXLY_CORRECT} is defined (even if it is
348 defined with a null value). Please make your program recognize this
349 variable if appropriate.
351 When a feature is used only by users (not by programs or command
352 files), and it is done poorly in Unix, feel free to replace it
353 completely with something totally different and better. (For example,
354 @code{vi} is replaced with Emacs.) But it is nice to offer a compatible
355 feature as well. (There is a free @code{vi} clone, so we offer it.)
357 Additional useful features are welcome regardless of whether
358 there is any precedent for them.
360 @node Using Extensions
361 @section Using Non-standard Features
362 @cindex non-standard extensions
364 Many GNU facilities that already exist support a number of convenient
365 extensions over the comparable Unix facilities. Whether to use these
366 extensions in implementing your program is a difficult question.
368 On the one hand, using the extensions can make a cleaner program.
369 On the other hand, people will not be able to build the program
370 unless the other GNU tools are available. This might cause the
371 program to work on fewer kinds of machines.
373 With some extensions, it might be easy to provide both alternatives.
374 For example, you can define functions with a ``keyword'' @code{INLINE}
375 and define that as a macro to expand into either @code{inline} or
376 nothing, depending on the compiler.
378 In general, perhaps it is best not to use the extensions if you can
379 straightforwardly do without them, but to use the extensions if they
380 are a big improvement.
382 An exception to this rule are the large, established programs (such as
383 Emacs) which run on a great variety of systems. Using GNU extensions in
384 such programs would make many users unhappy, so we don't do that.
386 Another exception is for programs that are used as part of compilation:
387 anything that must be compiled with other compilers in order to
388 bootstrap the GNU compilation facilities. If these require the GNU
389 compiler, then no one can compile them without having them installed
390 already. That would be extremely troublesome in certain cases.
393 @section Standard C and Pre-Standard C
394 @cindex ANSI C standard
396 1989 Standard C is widespread enough now that it is ok to use its
397 features in programs. There is one exception: do not ever use the
398 ``trigraph'' feature of Standard C.
400 The 1999 and 2011 editions of Standard C are not fully supported
401 on all platforms. If you aim to support compilation by
402 compilers other than GCC, you should not require these C
403 features in your programs. It is ok to use these features
404 conditionally when the compiler supports them.
406 If your program is only meant to compile with GCC, then you can
407 use these features if GCC supports them, when they give substantial
410 However, it is easy to support pre-standard compilers in most programs,
411 so if you know how to do that, feel free.
413 @cindex function prototypes
414 To support pre-standard C, instead of writing function definitions in
415 standard prototype form,
424 write the definition in pre-standard style like this,
434 and use a separate declaration to specify the argument prototype:
440 You need such a declaration anyway, in a header file, to get the benefit
441 of prototypes in all the files where the function is called. And once
442 you have the declaration, you normally lose nothing by writing the
443 function definition in the pre-standard style.
445 This technique does not work for integer types narrower than @code{int}.
446 If you think of an argument as being of a type narrower than @code{int},
447 declare it as @code{int} instead.
449 There are a few special cases where this technique is hard to use. For
450 example, if a function argument needs to hold the system type
451 @code{dev_t}, you run into trouble, because @code{dev_t} is shorter than
452 @code{int} on some machines; but you cannot use @code{int} instead,
453 because @code{dev_t} is wider than @code{int} on some machines. There
454 is no type you can safely use on all machines in a non-standard
455 definition. The only way to support non-standard C and pass such an
456 argument is to check the width of @code{dev_t} using Autoconf and choose
457 the argument type accordingly. This may not be worth the trouble.
459 In order to support pre-standard compilers that do not recognize
460 prototypes, you may want to use a preprocessor macro like this:
463 /* Declare the prototype for a general external function. */
464 #if defined (__STDC__) || defined (WINDOWSNT)
465 #define P_(proto) proto
471 @node Conditional Compilation
472 @section Conditional Compilation
474 When supporting configuration options already known when building your
475 program we prefer using @code{if (... )} over conditional compilation,
476 as in the former case the compiler is able to perform more extensive
477 checking of all possible code paths.
479 For example, please write
499 A modern compiler such as GCC will generate exactly the same code in
500 both cases, and we have been using similar techniques with good success
501 in several projects. Of course, the former method assumes that
502 @code{HAS_FOO} is defined as either 0 or 1.
504 While this is not a silver bullet solving all portability problems,
505 and is not always appropriate, following this policy would have saved
506 GCC developers many hours, or even days, per year.
508 In the case of function-like macros like @code{REVERSIBLE_CC_MODE} in
509 GCC which cannot be simply used in @code{if (...)} statements, there is
510 an easy workaround. Simply introduce another macro
511 @code{HAS_REVERSIBLE_CC_MODE} as in the following example:
514 #ifdef REVERSIBLE_CC_MODE
515 #define HAS_REVERSIBLE_CC_MODE 1
517 #define HAS_REVERSIBLE_CC_MODE 0
521 @node Program Behavior
522 @chapter Program Behavior for All Programs
524 This chapter describes conventions for writing robust
525 software. It also describes general standards for error messages, the
526 command line interface, and how libraries should behave.
529 * Non-GNU Standards:: We consider standards such as POSIX;
530 we don't "obey" them.
531 * Semantics:: Writing robust programs.
532 * Libraries:: Library behavior.
533 * Errors:: Formatting error messages.
534 * User Interfaces:: Standards about interfaces generally.
535 * Finding Program Files:: How to find the program's executable
536 and other files that go with it.
537 * Graphical Interfaces:: Standards for graphical interfaces.
538 * Command-Line Interfaces:: Standards for command line interfaces.
539 * Dynamic Plug-In Interfaces:: Standards for dynamic plug-in interfaces.
540 * Option Table:: Table of long options.
541 * OID Allocations:: Table of OID slots for GNU.
542 * Memory Usage:: When and how to care about memory needs.
543 * File Usage:: Which files to use, and where.
546 @node Non-GNU Standards
547 @section Non-GNU Standards
549 The GNU Project regards standards published by other organizations as
550 suggestions, not orders. We consider those standards, but we do not
551 ``obey'' them. In developing a GNU program, you should implement
552 an outside standard's specifications when that makes the GNU system
553 better overall in an objective sense. When it doesn't, you shouldn't.
555 In most cases, following published standards is convenient for
556 users---it means that their programs or scripts will work more
557 portably. For instance, GCC implements nearly all the features of
558 Standard C as specified by that standard. C program developers would
559 be unhappy if it did not. And GNU utilities mostly follow
560 specifications of POSIX.2; shell script writers and users would be
561 unhappy if our programs were incompatible.
563 But we do not follow either of these specifications rigidly, and there
564 are specific points on which we decided not to follow them, so as to
565 make the GNU system better for users.
567 For instance, Standard C says that nearly all extensions to C are
568 prohibited. How silly! GCC implements many extensions, some of which
569 were later adopted as part of the standard. If you want these
570 constructs to give an error message as ``required'' by the standard,
571 you must specify @samp{--pedantic}, which was implemented only so that
572 we can say ``GCC is a 100% implementation of the standard'', not
573 because there is any reason to actually use it.
575 POSIX.2 specifies that @samp{df} and @samp{du} must output sizes by
576 default in units of 512 bytes. What users want is units of 1k, so
577 that is what we do by default. If you want the ridiculous behavior
578 ``required'' by POSIX, you must set the environment variable
579 @samp{POSIXLY_CORRECT} (which was originally going to be named
580 @samp{POSIX_ME_HARDER}).
582 GNU utilities also depart from the letter of the POSIX.2 specification
583 when they support long-named command-line options, and intermixing
584 options with ordinary arguments. This minor incompatibility with
585 POSIX is never a problem in practice, and it is very useful.
587 In particular, don't reject a new feature, or remove an old one,
588 merely because a standard says it is ``forbidden'' or ``deprecated''.
592 @section Writing Robust Programs
594 @cindex arbitrary limits on data
595 Avoid arbitrary limits on the length or number of @emph{any} data
596 structure, including file names, lines, files, and symbols, by allocating
597 all data structures dynamically. In most Unix utilities, ``long lines
598 are silently truncated''. This is not acceptable in a GNU utility.
600 @cindex @code{NUL} characters
602 Utilities reading files should not drop NUL characters, or any other
603 nonprinting characters. Programs should work properly with multibyte
604 character encodings, such as UTF-8. You can use libiconv to deal with
605 a range of encodings.
607 @cindex error messages
608 Check every system call for an error return, unless you know you wish
609 to ignore errors. Include the system error text (from
610 @code{strerror}, or equivalent) in @emph{every} error message
611 resulting from a failing system call, as well as the name of the file
612 if any and the name of the utility. Just ``cannot open foo.c'' or
613 ``stat failed'' is not sufficient.
615 @cindex @code{malloc} return value
616 @cindex memory allocation failure
617 Check every call to @code{malloc} or @code{realloc} to see if it
618 returned @code{NULL}. Check @code{realloc} even if you are making the
619 block smaller; in a system that rounds block sizes to a power of 2,
620 @code{realloc} may get a different block if you ask for less space.
622 You must expect @code{free} to alter the contents of the block that was
623 freed. Anything you want to fetch from the block, you must fetch before
626 If @code{malloc} fails in a noninteractive program, make that a fatal
627 error. In an interactive program (one that reads commands from the
628 user), it is better to abort the command and return to the command
629 reader loop. This allows the user to kill other processes to free up
630 virtual memory, and then try the command again.
632 @cindex command-line arguments, decoding
633 Use @code{getopt_long} to decode arguments, unless the argument syntax
634 makes this unreasonable.
636 When static storage is to be written in during program execution, use
637 explicit C code to initialize it. This way, restarting the program
638 (without reloading it), or part of it, will reinitialize those
639 variables. Reserve C initialized declarations for data that will not
643 Try to avoid low-level interfaces to obscure Unix data structures (such
644 as file directories, utmp, or the layout of kernel memory), since these
645 are less likely to work compatibly. If you need to find all the files
646 in a directory, use @code{readdir} or some other high-level interface.
647 These are supported compatibly by GNU.
649 @cindex signal handling
650 The preferred signal handling facilities are the BSD variant of
651 @code{signal}, and the POSIX @code{sigaction} function; the
652 alternative USG @code{signal} interface is an inferior design.
654 Nowadays, using the POSIX signal functions may be the easiest way
655 to make a program portable. If you use @code{signal}, then on GNU/Linux
656 systems running GNU libc version 1, you should include
657 @file{bsd/signal.h} instead of @file{signal.h}, so as to get BSD
658 behavior. It is up to you whether to support systems where
659 @code{signal} has only the USG behavior, or give up on them.
661 @cindex impossible conditions
662 In error checks that detect ``impossible'' conditions, just abort.
663 There is usually no point in printing any message. These checks
664 indicate the existence of bugs. Whoever wants to fix the bugs will have
665 to read the source code and run a debugger. So explain the problem with
666 comments in the source. The relevant data will be in variables, which
667 are easy to examine with the debugger, so there is no point moving them
670 Do not use a count of errors as the exit status for a program.
671 @emph{That does not work}, because exit status values are limited to 8
672 bits (0 through 255). A single run of the program might have 256
673 errors; if you try to return 256 as the exit status, the parent process
674 will see 0 as the status, and it will appear that the program succeeded.
676 @cindex temporary files
677 @cindex @code{TMPDIR} environment variable
678 If you make temporary files, check the @code{TMPDIR} environment
679 variable; if that variable is defined, use the specified directory
680 instead of @file{/tmp}.
682 In addition, be aware that there is a possible security problem when
683 creating temporary files in world-writable directories. In C, you can
684 avoid this problem by creating temporary files in this manner:
687 fd = open (filename, O_WRONLY | O_CREAT | O_EXCL, 0600);
691 or by using the @code{mkstemps} function from Gnulib
692 (@pxref{mkstemps,,, gnulib, Gnulib}).
694 In bash, use @code{set -C} (long name @code{noclobber}) to avoid this
695 problem. In addition, the @code{mktemp} utility is a more general
696 solution for creating temporary files from shell scripts
697 (@pxref{mktemp invocation,,, coreutils, GNU Coreutils}).
701 @section Library Behavior
704 Try to make library functions reentrant. If they need to do dynamic
705 storage allocation, at least try to avoid any nonreentrancy aside from
706 that of @code{malloc} itself.
708 Here are certain name conventions for libraries, to avoid name
711 Choose a name prefix for the library, more than two characters long.
712 All external function and variable names should start with this
713 prefix. In addition, there should only be one of these in any given
714 library member. This usually means putting each one in a separate
717 An exception can be made when two external symbols are always used
718 together, so that no reasonable program could use one without the
719 other; then they can both go in the same file.
721 External symbols that are not documented entry points for the user
722 should have names beginning with @samp{_}. The @samp{_} should be
723 followed by the chosen name prefix for the library, to prevent
724 collisions with other libraries. These can go in the same files with
725 user entry points if you like.
727 Static functions and variables can be used as you like and need not
728 fit any naming convention.
731 @section Formatting Error Messages
732 @cindex formatting error messages
733 @cindex error messages, formatting
735 Error messages from compilers should look like this:
738 @var{sourcefile}:@var{lineno}: @var{message}
742 If you want to mention the column number, use one of these formats:
745 @var{sourcefile}:@var{lineno}:@var{column}: @var{message}
746 @var{sourcefile}:@var{lineno}.@var{column}: @var{message}
751 Line numbers should start from 1 at the beginning of the file, and
752 column numbers should start from 1 at the beginning of the line.
753 (Both of these conventions are chosen for compatibility.) Calculate
754 column numbers assuming that space and all ASCII printing characters
755 have equal width, and assuming tab stops every 8 columns. For
756 non-ASCII characters, Unicode character widths should be used when in
757 a UTF-8 locale; GNU libc and GNU gnulib provide suitable
758 @code{wcwidth} functions.
760 The error message can also give both the starting and ending positions
761 of the erroneous text. There are several formats so that you can
762 avoid redundant information such as a duplicate line number.
763 Here are the possible formats:
766 @var{sourcefile}:@var{line1}.@var{column1}-@var{line2}.@var{column2}: @var{message}
767 @var{sourcefile}:@var{line1}.@var{column1}-@var{column2}: @var{message}
768 @var{sourcefile}:@var{line1}-@var{line2}: @var{message}
772 When an error is spread over several files, you can use this format:
775 @var{file1}:@var{line1}.@var{column1}-@var{file2}:@var{line2}.@var{column2}: @var{message}
778 Error messages from other noninteractive programs should look like this:
781 @var{program}:@var{sourcefile}:@var{lineno}: @var{message}
785 when there is an appropriate source file, or like this:
788 @var{program}: @var{message}
792 when there is no relevant source file.
794 If you want to mention the column number, use this format:
797 @var{program}:@var{sourcefile}:@var{lineno}:@var{column}: @var{message}
800 In an interactive program (one that is reading commands from a
801 terminal), it is better not to include the program name in an error
802 message. The place to indicate which program is running is in the
803 prompt or with the screen layout. (When the same program runs with
804 input from a source other than a terminal, it is not interactive and
805 would do best to print error messages using the noninteractive style.)
807 The string @var{message} should not begin with a capital letter when
808 it follows a program name and/or file name, because that isn't the
809 beginning of a sentence. (The sentence conceptually starts at the
810 beginning of the line.) Also, it should not end with a period.
812 Error messages from interactive programs, and other messages such as
813 usage messages, should start with a capital letter. But they should not
816 @node User Interfaces
817 @section Standards for Interfaces Generally
819 @cindex program name and its behavior
820 @cindex behavior, dependent on program's name
821 Please don't make the behavior of a utility depend on the name used to
822 invoke it. It is useful sometimes to make a link to a utility with a
823 different name, and that should not change what it does. Thus, if you
824 make @file{foo} a link to @file{ls}, the program should behave the
825 same regardless of which of those names is used to invoke it.
827 Instead, use a run time option or a compilation switch or both to
828 select among the alternate behaviors. You can also build two versions
829 of the program, with different default behaviors, and install them
830 under two different names.
832 @cindex output device and program's behavior
833 Likewise, please don't make the behavior of a command-line program
834 depend on the type of output device it gets as standard output or
835 standard input. Device independence is an important principle of the
836 system's design; do not compromise it merely to save someone from
837 typing an option now and then. (Variation in error message syntax
838 when using a terminal is ok, because that is a side issue that people
841 If you think one behavior is most useful when the output is to a
842 terminal, and another is most useful when the output is a file or a
843 pipe, then it is usually best to make the default behavior the one
844 that is useful with output to a terminal, and have an option for the
845 other behavior. You can also build two different versions of the
846 program with different names.
848 There is an exception for programs whose output in certain cases is
849 binary data. Sending such output to a terminal is useless and can
850 cause trouble. If such a program normally sends its output to stdout,
851 it should detect, in these cases, when the output is a terminal and
852 give an error message instead. The @code{-f} option should override
853 this exception, thus permitting the output to go to the terminal.
855 Compatibility requires certain programs to depend on the type of output
856 device. It would be disastrous if @code{ls} or @code{sh} did not do so
857 in the way all users expect. In some of these cases, we supplement the
858 program with a preferred alternate version that does not depend on the
859 output device type. For example, we provide a @code{dir} program much
860 like @code{ls} except that its default output format is always
863 @node Finding Program Files
864 @section Finding the Program's Executable and Associated Files
866 A program may need to find the executable file it was started with, so
867 as to relaunch the same program. It may need to find associated
868 files, either source files or files constructed by building, that
871 The way to find them starts with looking at @code{argv[0]}.
873 If that string contains a slash, it is by convention the file name of
874 the executable and its directory part is the directory that contained
875 the executable. This is the case when the program was not found
876 through @env{PATH}, which normally means it was built but not
877 installed, and run from the build directory. The program can use the
878 @code{argv[0]} file name to relaunch itself, and can look in its
879 directory part for associated files. If that file name is not
880 absolute, then it is relative to the working directory in which the
883 If @code{argv[0]} does not contain a slash, it is a command name whose
884 executable was found via @env{PATH}. The program should search for
885 that name in the directories in @env{PATH}, interpreting @file{.} as
886 the working directory that was current when the program started.
888 If this procedure finds the executable, we call the directory it was
889 found in the @dfn{invocation directory}. The program should check
890 for the presence in that directory of the associated files it needs.
892 If the program's executable is normally built in a subdirectory of the
893 main build directory, and the main build directory contains associated
894 files (perhaps including subdirectories), the program should look at
895 the parent of the invocation directory, checking for the associated
896 files and subdirectories the main build directory should contain.
898 If the invocation directory doesn't contain what's needed, but the
899 executable file name is a symbolic link, the program should try using
900 the link target's containing directory as the invocation directory.
902 If this procedure doesn't come up with an invocation directory that is
903 valid---normally the case for an installed program that was found via
904 @env{PATH}---the program should look for the associated files in the
905 directories where the program's makefile installs them.
906 @xref{Directory Variables}.
908 Providing valid information in @code{argv[0]} is a convention, not
909 guaranteed. Well-behaved programs that launch other programs, such as
910 shells, follow the convention; your code should follow it too, when
911 launching other programs. But it is always possible to launch the
912 program and give a nonsensical value in @code{argv[0]}.
914 Therefore, any program that needs to know the location of its
915 executable, or that of of other associated files, should offer the
916 user environment variables to specify those locations explicitly.
918 @strong{Don't give special privilege, such as with the @code{setuid}
919 bit, to programs that will search heuristically for associated files
920 or for their own executables when invoked that way.} Limit that
921 privilege to programs that find associated files in hard-coded
922 installed locations such as under @file{/usr} and @file{/etc}.
924 @c ??? Is even that safe, in a setuid program?
926 @xref{Bourne Shell Variables,,, bash, Bash Reference Manual},
927 for more information about @env{PATH}.
929 @node Graphical Interfaces
930 @section Standards for Graphical Interfaces
931 @cindex graphical user interface
932 @cindex interface styles
933 @cindex user interface styles
937 When you write a program that provides a graphical user interface,
938 please make it work with the X Window System, using the GTK+ toolkit
939 or the GNUstep toolkit, unless the functionality specifically requires
940 some alternative (for example, ``displaying jpeg images while in
943 In addition, please provide a command-line interface to control the
944 functionality. (In many cases, the graphical user interface can be a
945 separate program which invokes the command-line program.) This is
946 so that the same jobs can be done from scripts.
951 @cindex keyboard interface
952 @cindex library interface
953 Please also consider providing a D-bus interface for use from other
954 running programs, such as within GNOME@. (GNOME used to use CORBA
955 for this, but that is being phased out.) In addition, consider
956 providing a library interface (for use from C), and perhaps a
957 keyboard-driven console interface (for use by users from console
958 mode). Once you are doing the work to provide the functionality and
959 the graphical interface, these won't be much extra work.
961 Please make your program interoperate with access technology such as
963 @url{https://www.gnu.org/accessibility/accessibility.html}). This should
964 be automatic if you use GTK+.
966 @node Command-Line Interfaces
967 @section Standards for Command Line Interfaces
968 @cindex command-line interface
971 It is a good idea to follow the POSIX guidelines for the
972 command-line options of a program. The easiest way to do this is to use
973 @code{getopt} to parse them. Note that the GNU version of @code{getopt}
974 will normally permit options anywhere among the arguments unless the
975 special argument @samp{--} is used. This is not what POSIX
976 specifies; it is a GNU extension.
978 @cindex long-named options
979 Please define long-named options that are equivalent to the
980 single-letter Unix-style options. We hope to make GNU more user
981 friendly this way. This is easy to do with the GNU function
984 One of the advantages of long-named options is that they can be
985 consistent from program to program. For example, users should be able
986 to expect the ``verbose'' option of any GNU program which has one, to be
987 spelled precisely @samp{--verbose}. To achieve this uniformity, look at
988 the table of common long-option names when you choose the option names
989 for your program (@pxref{Option Table}).
991 It is usually a good idea for file names given as ordinary arguments to
992 be input files only; any output files would be specified using options
993 (preferably @samp{-o} or @samp{--output}). Even if you allow an output
994 file name as an ordinary argument for compatibility, try to provide an
995 option as another way to specify it. This will lead to more consistency
996 among GNU utilities, and fewer idiosyncrasies for users to remember.
998 @cindex standard command-line options
999 @cindex options, standard command-line
1000 @cindex CGI programs, standard options for
1001 @cindex PATH_INFO, specifying standard options as
1002 All programs should support two standard options: @samp{--version}
1003 and @samp{--help}. CGI programs should accept these as command-line
1004 options, and also if given as the @env{PATH_INFO}; for instance,
1005 visiting @indicateurl{http://example.org/p.cgi/--help} in a browser should
1006 output the same information as invoking @samp{p.cgi --help} from the
1010 * --version:: The standard output for --version.
1011 * --help:: The standard output for --help.
1015 @subsection @option{--version}
1017 @cindex @samp{--version} output
1019 The standard @code{--version} option should direct the program to
1020 print information about its name, version, origin and legal status,
1021 all on standard output, and then exit successfully. Other options and
1022 arguments should be ignored once this is seen, and the program should
1023 not perform its normal function.
1025 @cindex canonical name of a program
1026 @cindex program's canonical name
1027 The first line is meant to be easy for a program to parse; the version
1028 number proper starts after the last space. In addition, it contains
1029 the canonical name for this program, in this format:
1036 The program's name should be a constant string; @emph{don't} compute it
1037 from @code{argv[0]}. The idea is to state the standard or canonical
1038 name for the program, not its file name. There are other ways to find
1039 out the precise file name where a command is found in @code{PATH}.
1041 If the program is a subsidiary part of a larger package, mention the
1042 package name in parentheses, like this:
1045 emacsserver (GNU Emacs) 19.30
1049 If the package has a version number which is different from this
1050 program's version number, you can mention the package version number
1051 just before the close-parenthesis.
1053 If you @emph{need} to mention the version numbers of libraries which
1054 are distributed separately from the package which contains this program,
1055 you can do so by printing an additional line of version info for each
1056 library you want to mention. Use the same format for these lines as for
1059 Please do not mention all of the libraries that the program uses ``just
1060 for completeness''---that would produce a lot of unhelpful clutter.
1061 Please mention library version numbers only if you find in practice that
1062 they are very important to you in debugging.
1064 The following line, after the version number line or lines, should be a
1065 copyright notice. If more than one copyright notice is called for, put
1066 each on a separate line.
1068 Next should follow a line stating the license, preferably using one of
1069 abbreviations below, and a brief statement that the program is free
1070 software, and that users are free to copy and change it. Also mention
1071 that there is no warranty, to the extent permitted by law. See
1072 recommended wording below.
1074 It is ok to finish the output with a list of the major authors of the
1075 program, as a way of giving credit.
1077 Here's an example of output that follows these rules:
1081 Copyright (C) 2007 Free Software Foundation, Inc.
1082 License GPLv3+: GNU GPL version 3 or later <https://gnu.org/licenses/gpl.html>
1083 This is free software: you are free to change and redistribute it.
1084 There is NO WARRANTY, to the extent permitted by law.
1087 You should adapt this to your program, of course, filling in the proper
1088 year, copyright holder, name of program, and the references to
1089 distribution terms, and changing the rest of the wording as necessary.
1091 This copyright notice only needs to mention the most recent year in
1092 which changes were made---there's no need to list the years for previous
1093 versions' changes. You don't have to mention the name of the program in
1094 these notices, if that is inconvenient, since it appeared in the first
1095 line. (The rules are different for copyright notices in source files;
1096 @pxref{Copyright Notices,,,maintain,Information for GNU Maintainers}.)
1098 Translations of the above lines must preserve the validity of the
1099 copyright notices (@pxref{Internationalization}). If the translation's
1100 character set supports it, the @samp{(C)} should be replaced with the
1101 copyright symbol, as follows:
1104 (the official copyright symbol, which is the letter C in a circle);
1110 Write the word ``Copyright'' exactly like that, in English. Do not
1111 translate it into another language. International treaties recognize
1112 the English word ``Copyright''; translations into other languages do not
1113 have legal significance.
1115 Finally, here is the table of our suggested license abbreviations.
1116 Any abbreviation can be followed by @samp{v@var{version}[+]}, meaning
1117 that particular version, or later versions with the @samp{+}, as shown
1118 above. In the case of a GNU license, @emph{always} indicate the permitted
1119 versions in this way.
1121 In the case of exceptions for extra permissions with the GPL, we use
1122 @samp{/} for a separator; the version number can follow the license
1123 abbreviation as usual, as in the examples below.
1127 GNU General Public License, @url{https://www.gnu.org/@/licenses/@/gpl.html}.
1130 GNU Lesser General Public License, @url{https://www.gnu.org/@/licenses/@/lgpl.html}.
1133 GNU GPL with the exception for Ada.
1136 The Apache Software Foundation license,
1137 @url{https://directory.fsf.org/@/wiki/@/License:Apache2.0}.
1140 The Artistic license used for Perl,
1141 @url{https://directory.fsf.org/@/wiki/@/License:ArtisticLicense2.0}.
1144 The Expat license, @url{https://directory.fsf.org/@/wiki/@/License:Expat}.
1147 The Mozilla Public License, @url{https://directory.fsf.org/@/wiki/@/License:MPLv2.0}.
1150 The original (4-clause) BSD license, incompatible with the GNU GPL,@*
1151 @url{https://directory.fsf.org/@/wiki/@/License:BSD_4Clause}.
1154 The license used for PHP, @url{https://directory.fsf.org/@/wiki/@/License:PHPv3.01}.
1157 The non-license that is being in the public domain,@*
1158 @url{https://www.gnu.org/@/licenses/@/license-list.html#PublicDomain}.
1161 The license for Python,
1162 @url{https://directory.fsf.org/@/wiki/@/License:Python2.0.1}.
1165 The revised (3-clause) BSD, compatible with the GNU GPL,@*
1166 @url{https://directory.fsf.org/@/wiki/@/License:BSD_3Clause}.
1169 The simple non-copyleft license used for most versions of the X Window
1170 System, @url{https://directory.fsf.org/@/wiki/@/License:X11}.
1173 The license for Zlib, @url{https://directory.fsf.org/@/wiki/@/License:Zlib}.
1177 More information about these licenses and many more are on the GNU
1178 licensing web pages,
1179 @url{https://www.gnu.org/@/licenses/@/license-list.html}.
1183 @subsection @option{--help}
1185 @cindex @samp{--help} output
1187 The standard @code{--help} option should output brief documentation
1188 for how to invoke the program, on standard output, then exit
1189 successfully. Other options and arguments should be ignored once this
1190 is seen, and the program should not perform its normal function.
1192 @cindex address for bug reports
1194 Near the end of the @samp{--help} option's output, please place lines
1195 giving the email address for bug reports, the package's home page
1196 (normally @indicateurl{https://www.gnu.org/software/@var{pkg}}, and the
1197 general page for help using GNU programs. The format should be like this:
1200 Report bugs to: @var{mailing-address}
1201 @var{pkg} home page: <https://www.gnu.org/software/@var{pkg}/>
1202 General help using GNU software: <https://www.gnu.org/gethelp/>
1205 It is ok to mention other appropriate mailing lists and web pages.
1208 @node Dynamic Plug-In Interfaces
1209 @section Standards for Dynamic Plug-in Interfaces
1211 @cindex dynamic plug-ins
1213 Another aspect of keeping free programs free is encouraging
1214 development of free plug-ins, and discouraging development of
1215 proprietary plug-ins. Many GNU programs will not have anything like
1216 plug-ins at all, but those that do should follow these
1219 First, the general plug-in architecture design should closely tie the
1220 plug-in to the original code, such that the plug-in and the base
1221 program are parts of one extended program. For GCC, for example,
1222 plug-ins receive and modify GCC's internal data structures, and so
1223 clearly form an extended program with the base GCC.
1225 @vindex plugin_is_GPL_compatible
1226 Second, you should require plug-in developers to affirm that their
1227 plug-ins are released under an appropriate license. This should be
1228 enforced with a simple programmatic check. For GCC, again for
1229 example, a plug-in must define the global symbol
1230 @code{plugin_is_GPL_compatible}, thus asserting that the plug-in is
1231 released under a GPL-compatible license (@pxref{Plugins,, Plugins,
1232 gccint, GCC Internals}).
1234 By adding this check to your program you are not creating a new legal
1235 requirement. The GPL itself requires plug-ins to be free software,
1236 licensed compatibly. As long as you have followed the first rule above
1237 to keep plug-ins closely tied to your original program, the GPL and AGPL
1238 already require those plug-ins to be released under a compatible
1239 license. The symbol definition in the plug-in---or whatever equivalent
1240 works best in your program---makes it harder for anyone who might
1241 distribute proprietary plug-ins to legally defend themselves. If a case
1242 about this got to court, we can point to that symbol as evidence that
1243 the plug-in developer understood that the license had this requirement.
1247 @section Table of Long Options
1248 @cindex long option names
1249 @cindex table of long options
1251 Here is a table of long options used by GNU programs. It is surely
1252 incomplete, but we aim to list all the options that a new program might
1253 want to be compatible with. If you use names not already in the table,
1254 please send @email{bug-standards@@gnu.org} a list of them, with their
1255 meanings, so we can update the table.
1257 @c Please leave newlines between items in this table; it's much easier
1258 @c to update when it isn't completely squashed together and unreadable.
1259 @c When there is more than one short option for a long option name, put
1260 @c a semicolon between the lists of the programs that use them, not a
1261 @c period. --friedman
1265 @samp{-N} in @code{tar}.
1268 @samp{-a} in @code{du}, @code{ls}, @code{nm}, @code{stty}, @code{uname},
1269 and @code{unexpand}.
1272 @samp{-a} in @code{diff}.
1275 @samp{-A} in @code{ls}.
1278 @samp{-a} in @code{etags}, @code{tee}, @code{time};
1279 @samp{-r} in @code{tar}.
1282 @samp{-a} in @code{cp}.
1285 @samp{-n} in @code{shar}.
1288 @samp{-l} in @code{m4}.
1291 @samp{-a} in @code{diff}.
1294 @samp{-v} in @code{gawk}.
1297 @samp{-W} in @code{make}.
1300 @samp{-o} in @code{make}.
1303 @samp{-a} in @code{recode}.
1306 @samp{-a} in @code{wdiff}.
1308 @item auto-reference
1309 @samp{-A} in @code{ptx}.
1312 @samp{-n} in @code{wdiff}.
1315 For server programs, run in the background.
1317 @item backward-search
1318 @samp{-B} in @code{ctags}.
1321 @samp{-f} in @code{shar}.
1330 @samp{-b} in @code{tac}.
1333 @samp{-b} in @code{cpio} and @code{diff}.
1336 @samp{-b} in @code{shar}.
1339 Used in @code{cpio} and @code{tar}.
1342 @samp{-b} in @code{head} and @code{tail}.
1345 @samp{-b} in @code{ptx}.
1348 Used in various programs to make output shorter.
1351 @samp{-c} in @code{head}, @code{split}, and @code{tail}.
1354 @samp{-C} in @code{etags}.
1357 @samp{-A} in @code{tar}.
1360 Used in various programs to specify the directory to use.
1363 @samp{-c} in @code{chgrp} and @code{chown}.
1366 @samp{-F} in @code{ls}.
1369 @samp{-c} in @code{recode}.
1372 @samp{-c} in @code{su};
1376 @samp{-d} in @code{tar}.
1379 Used in @code{gawk}.
1382 @samp{-Z} in @code{tar} and @code{shar}.
1385 @samp{-A} in @code{tar}.
1388 @samp{-w} in @code{tar}.
1391 Used in @code{diff}.
1394 @samp{-W copyleft} in @code{gawk}.
1397 @samp{-C} in @code{ptx}, @code{recode}, and @code{wdiff};
1398 @samp{-W copyright} in @code{gawk}.
1404 @samp{-q} in @code{who}.
1407 @samp{-l} in @code{du}.
1410 Used in @code{tar} and @code{cpio}.
1413 @samp{-c} in @code{shar}.
1416 @samp{-x} in @code{ctags}.
1419 @samp{-d} in @code{touch}.
1422 @samp{-d} in @code{make} and @code{m4};
1426 @samp{-D} in @code{m4}.
1429 @samp{-d} in Bison and @code{ctags}.
1432 @samp{-D} in @code{tar}.
1435 @samp{-L} in @code{chgrp}, @code{chown}, @code{cpio}, @code{du},
1436 @code{ls}, and @code{tar}.
1438 @item dereference-args
1439 @samp{-D} in @code{du}.
1442 Specify an I/O device (special file name).
1445 @samp{-d} in @code{recode}.
1447 @item dictionary-order
1448 @samp{-d} in @code{look}.
1451 @samp{-d} in @code{tar}.
1454 @samp{-n} in @code{csplit}.
1457 Specify the directory to use, in various programs. In @code{ls}, it
1458 means to show directories themselves rather than their contents. In
1459 @code{rm} and @code{ln}, it means to not treat links to directories
1463 @samp{-x} in @code{strip}.
1465 @item discard-locals
1466 @samp{-X} in @code{strip}.
1469 @samp{-n} in @code{make}.
1472 @samp{-e} in @code{diff}.
1474 @item elide-empty-files
1475 @samp{-z} in @code{csplit}.
1478 @samp{-x} in @code{wdiff}.
1481 @samp{-z} in @code{wdiff}.
1483 @item entire-new-file
1484 @samp{-N} in @code{diff}.
1486 @item environment-overrides
1487 @samp{-e} in @code{make}.
1490 @samp{-e} in @code{xargs}.
1496 Used in @code{makeinfo}.
1499 @samp{-o} in @code{m4}.
1502 @samp{-b} in @code{ls}.
1505 @samp{-X} in @code{tar}.
1511 @samp{-x} in @code{xargs}.
1514 @samp{-e} in @code{unshar}.
1517 @samp{-t} in @code{diff}.
1520 @samp{-e} in @code{sed}.
1523 @samp{-g} in @code{nm}.
1526 @samp{-i} in @code{cpio};
1527 @samp{-x} in @code{tar}.
1530 @samp{-f} in @code{finger}.
1533 @samp{-f} in @code{su}.
1535 @item fatal-warnings
1536 @samp{-E} in @code{m4}.
1539 @samp{-f} in @code{gawk}, @code{info}, @code{make}, @code{mt},
1540 @code{sed}, and @code{tar}.
1542 @item field-separator
1543 @samp{-F} in @code{gawk}.
1549 @samp{-F} in @code{ls}.
1552 @samp{-T} in @code{tar}.
1555 Used in @code{makeinfo}.
1557 @item flag-truncation
1558 @samp{-F} in @code{ptx}.
1560 @item fixed-output-files
1564 @samp{-f} in @code{tail}.
1566 @item footnote-style
1567 Used in @code{makeinfo}.
1570 @samp{-f} in @code{cp}, @code{ln}, @code{mv}, and @code{rm}.
1573 @samp{-F} in @code{shar}.
1576 For server programs, run in the foreground;
1577 in other words, don't do anything special to run the server
1581 Used in @code{ls}, @code{time}, and @code{ptx}.
1584 @samp{-F} in @code{m4}.
1590 @samp{-g} in @code{ptx}.
1593 @samp{-x} in @code{tar}.
1596 @samp{-i} in @code{ul}.
1599 @samp{-g} in @code{recode}.
1602 @samp{-g} in @code{install}.
1605 @samp{-z} in @code{tar} and @code{shar}.
1608 @samp{-H} in @code{m4}.
1611 @samp{-h} in @code{objdump} and @code{recode}
1614 @samp{-H} in @code{who}.
1617 Used to ask for brief usage information.
1619 @item here-delimiter
1620 @samp{-d} in @code{shar}.
1622 @item hide-control-chars
1623 @samp{-q} in @code{ls}.
1626 In @code{makeinfo}, output HTML.
1629 @samp{-u} in @code{who}.
1632 @samp{-D} in @code{diff}.
1635 @samp{-I} in @code{ls};
1636 @samp{-x} in @code{recode}.
1638 @item ignore-all-space
1639 @samp{-w} in @code{diff}.
1641 @item ignore-backups
1642 @samp{-B} in @code{ls}.
1644 @item ignore-blank-lines
1645 @samp{-B} in @code{diff}.
1648 @samp{-f} in @code{look} and @code{ptx};
1649 @samp{-i} in @code{diff} and @code{wdiff}.
1652 @samp{-i} in @code{make}.
1655 @samp{-i} in @code{ptx}.
1657 @item ignore-indentation
1658 @samp{-I} in @code{etags}.
1660 @item ignore-init-file
1663 @item ignore-interrupts
1664 @samp{-i} in @code{tee}.
1666 @item ignore-matching-lines
1667 @samp{-I} in @code{diff}.
1669 @item ignore-space-change
1670 @samp{-b} in @code{diff}.
1673 @samp{-i} in @code{tar}.
1676 @samp{-i} in @code{etags};
1677 @samp{-I} in @code{m4}.
1680 @samp{-I} in @code{make}.
1683 @samp{-G} in @code{tar}.
1686 @samp{-i}, @samp{-l}, and @samp{-m} in Finger.
1689 In some programs, specify the name of the file to read as the user's
1693 @samp{-i} in @code{expand}.
1696 @samp{-T} in @code{diff}.
1699 @samp{-i} in @code{ls}.
1702 @samp{-i} in @code{cp}, @code{ln}, @code{mv}, @code{rm};
1703 @samp{-e} in @code{m4};
1704 @samp{-p} in @code{xargs};
1705 @samp{-w} in @code{tar}.
1708 @samp{-p} in @code{shar}.
1714 @samp{-j} in @code{make}.
1717 @samp{-n} in @code{make}.
1720 @samp{-k} in @code{make}.
1723 @samp{-k} in @code{csplit}.
1726 @samp{-k} in @code{du} and @code{ls}.
1729 @samp{-l} in @code{etags}.
1732 @samp{-l} in @code{wdiff}.
1734 @item level-for-gzip
1735 @samp{-g} in @code{shar}.
1738 @samp{-C} in @code{split}.
1741 Used in @code{split}, @code{head}, and @code{tail}.
1744 @samp{-l} in @code{cpio}.
1748 Used in @code{gawk}.
1751 @samp{-t} in @code{cpio};
1752 @samp{-l} in @code{recode}.
1755 @samp{-t} in @code{tar}.
1758 @samp{-N} in @code{ls}.
1761 @samp{-l} in @code{make}.
1767 Used in @code{uname}.
1770 @samp{-M} in @code{ptx}.
1773 @samp{-m} in @code{hello} and @code{uname}.
1775 @item make-directories
1776 @samp{-d} in @code{cpio}.
1779 @samp{-f} in @code{make}.
1785 @samp{-n} in @code{xargs}.
1788 @samp{-n} in @code{xargs}.
1791 @samp{-l} in @code{xargs}.
1794 @samp{-l} in @code{make}.
1797 @samp{-P} in @code{xargs}.
1800 @samp{-T} in @code{who}.
1803 @samp{-T} in @code{who}.
1806 @samp{-d} in @code{diff}.
1808 @item mixed-uuencode
1809 @samp{-M} in @code{shar}.
1812 @samp{-m} in @code{install}, @code{mkdir}, and @code{mkfifo}.
1814 @item modification-time
1815 @samp{-m} in @code{tar}.
1818 @samp{-M} in @code{tar}.
1824 @samp{-L} in @code{m4}.
1827 @samp{-a} in @code{shar}.
1830 @samp{-W} in @code{make}.
1832 @item no-builtin-rules
1833 @samp{-r} in @code{make}.
1835 @item no-character-count
1836 @samp{-w} in @code{shar}.
1838 @item no-check-existing
1839 @samp{-x} in @code{shar}.
1842 @samp{-3} in @code{wdiff}.
1845 @samp{-c} in @code{touch}.
1848 @samp{-D} in @code{etags}.
1851 @samp{-1} in @code{wdiff}.
1853 @item no-dereference
1854 @samp{-d} in @code{cp}.
1857 @samp{-2} in @code{wdiff}.
1860 @samp{-S} in @code{make}.
1866 @samp{-P} in @code{shar}.
1869 @samp{-e} in @code{gprof}.
1872 @samp{-R} in @code{etags}.
1875 @samp{-p} in @code{nm}.
1878 Don't print a startup splash screen.
1881 Used in @code{makeinfo}.
1884 @samp{-a} in @code{gprof}.
1887 @samp{-E} in @code{gprof}.
1890 @samp{-m} in @code{shar}.
1893 Used in @code{makeinfo}.
1896 Used in @code{emacsclient}.
1899 Used in various programs to inhibit warnings.
1902 @samp{-n} in @code{info}.
1905 @samp{-n} in @code{uname}.
1908 @samp{-f} in @code{cpio}.
1911 @samp{-n} in @code{objdump}.
1914 @samp{-0} in @code{xargs}.
1917 @samp{-n} in @code{cat}.
1919 @item number-nonblank
1920 @samp{-b} in @code{cat}.
1923 @samp{-n} in @code{nm}.
1925 @item numeric-uid-gid
1926 @samp{-n} in @code{cpio} and @code{ls}.
1932 @samp{-o} in @code{tar}.
1935 @samp{-o} in @code{make}.
1937 @item one-file-system
1938 @samp{-l} in @code{tar}, @code{cp}, and @code{du}.
1941 @samp{-o} in @code{ptx}.
1944 @samp{-f} in @code{gprof}.
1947 @samp{-F} in @code{gprof}.
1950 @samp{-o} in @code{getopt}, @code{fdlist}, @code{fdmount},
1951 @code{fdmountd}, and @code{fdumount}.
1954 In various programs, specify the output file name.
1957 @samp{-o} in @code{shar}.
1960 @samp{-o} in @code{rm}.
1963 @samp{-c} in @code{unshar}.
1966 @samp{-o} in @code{install}.
1969 @samp{-l} in @code{diff}.
1971 @item paragraph-indent
1972 Used in @code{makeinfo}.
1975 @samp{-p} in @code{mkdir} and @code{rmdir}.
1978 @samp{-p} in @code{ul}.
1981 @samp{-p} in @code{cpio}.
1984 @samp{-P} in @code{finger}.
1987 @samp{-c} in @code{cpio} and @code{tar}.
1990 Used in @code{gawk}.
1992 @item prefix-builtins
1993 @samp{-P} in @code{m4}.
1996 @samp{-f} in @code{csplit}.
1999 Used in @code{tar} and @code{cp}.
2001 @item preserve-environment
2002 @samp{-p} in @code{su}.
2004 @item preserve-modification-time
2005 @samp{-m} in @code{cpio}.
2007 @item preserve-order
2008 @samp{-s} in @code{tar}.
2010 @item preserve-permissions
2011 @samp{-p} in @code{tar}.
2014 @samp{-l} in @code{diff}.
2017 @samp{-L} in @code{cmp}.
2019 @item print-data-base
2020 @samp{-p} in @code{make}.
2022 @item print-directory
2023 @samp{-w} in @code{make}.
2025 @item print-file-name
2026 @samp{-o} in @code{nm}.
2029 @samp{-s} in @code{nm}.
2032 @samp{-p} in @code{wdiff}.
2035 @samp{-p} in @code{ed}.
2038 Specify an HTTP proxy.
2041 @samp{-X} in @code{shar}.
2044 @samp{-q} in @code{make}.
2047 Used in many programs to inhibit the usual output. Every
2048 program accepting @samp{--quiet} should accept @samp{--silent} as a
2052 @samp{-Q} in @code{shar}
2055 @samp{-Q} in @code{ls}.
2058 @samp{-n} in @code{diff}.
2061 Used in @code{gawk}.
2063 @item read-full-blocks
2064 @samp{-B} in @code{tar}.
2070 @samp{-n} in @code{make}.
2073 @samp{-R} in @code{tar}.
2076 Used in @code{chgrp}, @code{chown}, @code{cp}, @code{ls}, @code{diff},
2080 @samp{-r} in @code{touch}.
2083 @samp{-r} in @code{ptx}.
2086 @samp{-r} in @code{tac} and @code{etags}.
2089 @samp{-r} in @code{uname}.
2092 @samp{-R} in @code{m4}.
2095 @samp{-r} in @code{objdump}.
2098 @samp{-r} in @code{cpio}.
2101 @samp{-i} in @code{xargs}.
2103 @item report-identical-files
2104 @samp{-s} in @code{diff}.
2106 @item reset-access-time
2107 @samp{-a} in @code{cpio}.
2110 @samp{-r} in @code{ls} and @code{nm}.
2113 @samp{-f} in @code{diff}.
2115 @item right-side-defs
2116 @samp{-R} in @code{ptx}.
2119 @samp{-s} in @code{tar}.
2121 @item same-permissions
2122 @samp{-p} in @code{tar}.
2125 @samp{-g} in @code{stty}.
2130 @item sentence-regexp
2131 @samp{-S} in @code{ptx}.
2134 @samp{-S} in @code{du}.
2137 @samp{-s} in @code{tac}.
2140 Used by @code{recode} to chose files or pipes for sequencing passes.
2143 @samp{-s} in @code{su}.
2146 @samp{-A} in @code{cat}.
2148 @item show-c-function
2149 @samp{-p} in @code{diff}.
2152 @samp{-E} in @code{cat}.
2154 @item show-function-line
2155 @samp{-F} in @code{diff}.
2158 @samp{-T} in @code{cat}.
2161 Used in many programs to inhibit the usual output.
2162 Every program accepting
2163 @samp{--silent} should accept @samp{--quiet} as a synonym.
2166 @samp{-s} in @code{ls}.
2169 Specify a file descriptor for a network server to use for its socket,
2170 instead of opening and binding a new socket. This provides a way to
2171 run, in a non-privileged process, a server that normally needs a
2172 reserved port number.
2178 @samp{-W source} in @code{gawk}.
2181 @samp{-S} in @code{tar}.
2183 @item speed-large-files
2184 @samp{-H} in @code{diff}.
2187 @samp{-E} in @code{unshar}.
2189 @item split-size-limit
2190 @samp{-L} in @code{shar}.
2193 @samp{-s} in @code{cat}.
2196 @samp{-w} in @code{wdiff}.
2199 @samp{-y} in @code{wdiff}.
2202 Used in @code{tar} and @code{diff} to specify which file within
2203 a directory to start processing with.
2206 @samp{-s} in @code{wdiff}.
2208 @item stdin-file-list
2209 @samp{-S} in @code{shar}.
2212 @samp{-S} in @code{make}.
2215 @samp{-s} in @code{recode}.
2218 @samp{-s} in @code{install}.
2221 @samp{-s} in @code{strip}.
2224 @samp{-S} in @code{strip}.
2227 @samp{-s} in @code{shar}.
2230 @samp{-S} in @code{cp}, @code{ln}, @code{mv}.
2233 @samp{-b} in @code{csplit}.
2236 @samp{-s} in @code{gprof}.
2239 @samp{-s} in @code{du}.
2242 @samp{-s} in @code{ln}.
2245 Used in GDB and @code{objdump}.
2248 @samp{-s} in @code{m4}.
2251 @samp{-s} in @code{uname}.
2254 @samp{-t} in @code{expand} and @code{unexpand}.
2257 @samp{-T} in @code{ls}.
2260 @samp{-T} in @code{tput} and @code{ul}.
2261 @samp{-t} in @code{wdiff}.
2264 @samp{-a} in @code{diff}.
2267 @samp{-T} in @code{shar}.
2270 Used in @code{ls} and @code{touch}.
2273 Specify how long to wait before giving up on some operation.
2276 @samp{-O} in @code{tar}.
2279 @samp{-c} in @code{du}.
2282 @samp{-t} in @code{make}, @code{ranlib}, and @code{recode}.
2285 @samp{-t} in @code{m4}.
2288 @samp{-t} in @code{hello};
2289 @samp{-W traditional} in @code{gawk};
2290 @samp{-G} in @code{ed}, @code{m4}, and @code{ptx}.
2296 @samp{-t} in @code{ctags}.
2298 @item typedefs-and-c++
2299 @samp{-T} in @code{ctags}.
2302 @samp{-t} in @code{ptx}.
2305 @samp{-z} in @code{tar}.
2308 @samp{-u} in @code{cpio}.
2311 @samp{-U} in @code{m4}.
2313 @item undefined-only
2314 @samp{-u} in @code{nm}.
2317 @samp{-u} in @code{cp}, @code{ctags}, @code{mv}, @code{tar}.
2320 Used in @code{gawk}; same as @samp{--help}.
2323 @samp{-B} in @code{shar}.
2325 @item vanilla-operation
2326 @samp{-V} in @code{shar}.
2329 Print more information about progress. Many programs support this.
2332 @samp{-W} in @code{tar}.
2335 Print the version number.
2337 @item version-control
2338 @samp{-V} in @code{cp}, @code{ln}, @code{mv}.
2341 @samp{-v} in @code{ctags}.
2344 @samp{-V} in @code{tar}.
2347 @samp{-W} in @code{make}.
2349 @item whole-size-limit
2350 @samp{-l} in @code{shar}.
2353 @samp{-w} in @code{ls} and @code{ptx}.
2356 @samp{-W} in @code{ptx}.
2359 @samp{-T} in @code{who}.
2362 @samp{-z} in @code{gprof}.
2365 @node OID Allocations
2366 @section OID Allocations
2367 @cindex OID allocations for GNU
2372 The OID (object identifier) 1.3.6.1.4.1.11591 has been assigned to the
2373 GNU Project (thanks to Sergey Poznyakoff). These are used for SNMP,
2374 LDAP, X.509 certificates, and so on. The web site
2375 @url{https://www.alvestrand.no/objectid} has a (voluntary) listing of
2376 many OID assignments.
2378 If you need a new slot for your GNU package, write
2379 @email{maintainers@@gnu.org}. Here is a list of arcs currently
2383 @include gnu-oids.texi
2388 @section Memory Usage
2389 @cindex memory usage
2391 If a program typically uses just a few meg of memory, don't bother making any
2392 effort to reduce memory usage. For example, if it is impractical for
2393 other reasons to operate on files more than a few meg long, it is
2394 reasonable to read entire input files into memory to operate on them.
2396 However, for programs such as @code{cat} or @code{tail}, that can
2397 usefully operate on very large files, it is important to avoid using a
2398 technique that would artificially limit the size of files it can handle.
2399 If a program works by lines and could be applied to arbitrary
2400 user-supplied input files, it should keep only a line in memory, because
2401 this is not very hard and users will want to be able to operate on input
2402 files that are bigger than will fit in memory all at once.
2404 If your program creates complicated data structures, just make them in
2405 memory and give a fatal error if @code{malloc} returns @code{NULL}.
2409 Memory analysis tools such as @command{valgrind} can be useful, but
2410 don't complicate a program merely to avoid their false alarms. For
2411 example, if memory is used until just before a process exits, don't
2412 free it simply to silence such a tool.
2419 Programs should be prepared to operate when @file{/usr} and @file{/etc}
2420 are read-only file systems. Thus, if the program manages log files,
2421 lock files, backup files, score files, or any other files which are
2422 modified for internal purposes, these files should not be stored in
2423 @file{/usr} or @file{/etc}.
2425 There are two exceptions. @file{/etc} is used to store system
2426 configuration information; it is reasonable for a program to modify
2427 files in @file{/etc} when its job is to update the system configuration.
2428 Also, if the user explicitly asks to modify one file in a directory, it
2429 is reasonable for the program to store other files in the same
2433 @chapter Making The Best Use of C
2435 This chapter provides advice on how best to use the C language
2436 when writing GNU software.
2439 * Formatting:: Formatting your source code.
2440 * Comments:: Commenting your work.
2441 * Syntactic Conventions:: Clean use of C constructs.
2442 * Names:: Naming variables, functions, and files.
2443 * System Portability:: Portability among different operating systems.
2444 * CPU Portability:: Supporting the range of CPU types.
2445 * System Functions:: Portability and ``standard'' library functions.
2446 * Internationalization:: Techniques for internationalization.
2447 * Character Set:: Use ASCII by default.
2448 * Quote Characters:: Use "..." or '...' in the C locale.
2449 * Mmap:: How you can safely use @code{mmap}.
2453 @section Formatting Your Source Code
2454 @cindex formatting source code
2457 @cindex length of source lines
2458 Please keep the length of source lines to 79 characters or less, for
2459 maximum readability in the widest range of environments.
2462 @cindex braces, in C source
2463 @cindex function definitions, formatting
2464 It is important to put the open-brace that starts the body of a C
2465 function in column one, so that they will start a defun. Several
2466 tools look for open-braces in column one to find the beginnings of C
2467 functions. These tools will not work on code not formatted that way.
2469 Avoid putting open-brace, open-parenthesis or open-bracket in column
2470 one when they are inside a function, so that they won't start a defun.
2471 The open-brace that starts a @code{struct} body can go in column one
2472 if you find it useful to treat that definition as a defun.
2474 It is also important for function definitions to start the name of the
2475 function in column one. This helps people to search for function
2476 definitions, and may also help certain tools recognize them. Thus,
2477 using Standard C syntax, the format is this:
2481 concat (char *s1, char *s2)
2488 or, if you want to use traditional C syntax, format the definition like
2493 concat (s1, s2) /* Name starts in column one here */
2495 @{ /* Open brace in column one here */
2500 In Standard C, if the arguments don't fit nicely on one line,
2505 lots_of_args (int an_integer, long a_long, short a_short,
2506 double a_double, float a_float)
2510 @cindex @code{struct} types, formatting
2511 @cindex @code{enum} types, formatting
2512 For @code{struct} and @code{enum} types, likewise put the braces in
2513 column one, unless the whole contents fits on one line:
2521 struct foo @{ int a, b; @}
2524 The rest of this section gives our recommendations for other aspects of
2525 C formatting style, which is also the default style of the @code{indent}
2526 program in version 1.2 and newer. It corresponds to the options
2529 -nbad -bap -nbc -bbo -bl -bli2 -bls -ncdb -nce -cp1 -cs -di2
2530 -ndj -nfc1 -nfca -hnl -i2 -ip5 -lp -pcs -psl -nsc -nsob
2533 We don't think of these recommendations as requirements, because it
2534 causes no problems for users if two different programs have different
2537 But whatever style you use, please use it consistently, since a mixture
2538 of styles within one program tends to look ugly. If you are
2539 contributing changes to an existing program, please follow the style of
2542 For the body of the function, our recommended style looks like this:
2554 return ++x + bar ();
2558 @cindex spaces before open-paren
2559 We find it easier to read a program when it has spaces before the
2560 open-parentheses and after the commas. Especially after the commas.
2562 When you split an expression into multiple lines, split it
2563 before an operator, not after one. Here is the right way:
2565 @cindex expressions, splitting
2567 if (foo_this_is_long && bar > win (x, y, z)
2568 && remaining_condition)
2571 Try to avoid having two operators of different precedence at the same
2572 level of indentation. For example, don't write this:
2575 mode = (inmode[j] == VOIDmode
2576 || GET_MODE_SIZE (outmode[j]) > GET_MODE_SIZE (inmode[j])
2577 ? outmode[j] : inmode[j]);
2580 Instead, use extra parentheses so that the indentation shows the nesting:
2583 mode = ((inmode[j] == VOIDmode
2584 || (GET_MODE_SIZE (outmode[j]) > GET_MODE_SIZE (inmode[j])))
2585 ? outmode[j] : inmode[j]);
2588 Insert extra parentheses so that Emacs will indent the code properly.
2589 For example, the following indentation looks nice if you do it by hand,
2592 v = rup->ru_utime.tv_sec*1000 + rup->ru_utime.tv_usec/1000
2593 + rup->ru_stime.tv_sec*1000 + rup->ru_stime.tv_usec/1000;
2597 but Emacs would alter it. Adding a set of parentheses produces
2598 something that looks equally nice, and which Emacs will preserve:
2601 v = (rup->ru_utime.tv_sec*1000 + rup->ru_utime.tv_usec/1000
2602 + rup->ru_stime.tv_sec*1000 + rup->ru_stime.tv_usec/1000);
2605 Format do-while statements like this:
2617 Please use formfeed characters (control-L) to divide the program into
2618 pages at logical places (but not within a function). It does not matter
2619 just how long the pages are, since they do not have to fit on a printed
2620 page. The formfeeds should appear alone on lines by themselves.
2623 @section Commenting Your Work
2626 Every program should start with a comment saying briefly what it is for.
2627 Example: @samp{fmt - filter for simple filling of text}. This comment
2628 should be at the top of the source file containing the @samp{main}
2629 function of the program.
2631 Also, please write a brief comment at the start of each source file,
2632 with the file name and a line or two about the overall purpose of the
2635 Please write the comments in a GNU program in English, because English
2636 is the one language that nearly all programmers in all countries can
2637 read. If you do not write English well, please write comments in
2638 English as well as you can, then ask other people to help rewrite them.
2639 If you can't write comments in English, please find someone to work with
2640 you and translate your comments into English.
2642 Please put a comment on each function saying what the function does,
2643 what sorts of arguments it gets, and what the possible values of
2644 arguments mean and are used for. It is not necessary to duplicate in
2645 words the meaning of the C argument declarations, if a C type is being
2646 used in its customary fashion. If there is anything nonstandard about
2647 its use (such as an argument of type @code{char *} which is really the
2648 address of the second character of a string, not the first), or any
2649 possible values that would not work the way one would expect (such as,
2650 that strings containing newlines are not guaranteed to work), be sure
2653 Also explain the significance of the return value, if there is one.
2655 Please put two spaces after the end of a sentence in your comments, so
2656 that the Emacs sentence commands will work. Also, please write
2657 complete sentences and capitalize the first word. If a lower-case
2658 identifier comes at the beginning of a sentence, don't capitalize it!
2659 Changing the spelling makes it a different identifier. If you don't
2660 like starting a sentence with a lower case letter, write the sentence
2661 differently (e.g., ``The identifier lower-case is @dots{}'').
2663 The comment on a function is much clearer if you use the argument
2664 names to speak about the argument values. The variable name itself
2665 should be lower case, but write it in upper case when you are speaking
2666 about the value rather than the variable itself. Thus, ``the inode
2667 number NODE_NUM'' rather than ``an inode''.
2669 There is usually no purpose in restating the name of the function in
2670 the comment before it, because readers can see that for themselves.
2671 There might be an exception when the comment is so long that the function
2672 itself would be off the bottom of the screen.
2674 There should be a comment on each static variable as well, like this:
2677 /* Nonzero means truncate lines in the display;
2678 zero means continue them. */
2682 @cindex conditionals, comments for
2683 @cindex @code{#endif}, commenting
2684 Every @samp{#endif} should have a comment, except in the case of short
2685 conditionals (just a few lines) that are not nested. The comment should
2686 state the condition of the conditional that is ending, @emph{including
2687 its sense}. @samp{#else} should have a comment describing the condition
2688 @emph{and sense} of the code that follows. For example:
2696 #endif /* not foo */
2706 but, by contrast, write the comments this way for a @samp{#ifndef}:
2719 #endif /* not foo */
2723 @node Syntactic Conventions
2724 @section Clean Use of C Constructs
2725 @cindex syntactic conventions
2727 @cindex implicit @code{int}
2728 @cindex function argument, declaring
2729 Please explicitly declare the types of all objects. For example, you
2730 should explicitly declare all arguments to functions, and you should
2731 declare functions to return @code{int} rather than omitting the
2734 @cindex compiler warnings
2735 @cindex @samp{-Wall} compiler option
2736 Some programmers like to use the GCC @samp{-Wall} option, and change the
2737 code whenever it issues a warning. If you want to do this, then do.
2738 Other programmers prefer not to use @samp{-Wall}, because it gives
2739 warnings for valid and legitimate code which they do not want to change.
2740 If you want to do this, then do. The compiler should be your servant,
2745 Don't make the program ugly just to placate static analysis tools such
2746 as @command{lint}, @command{clang}, and GCC with extra warnings
2747 options such as @option{-Wconversion} and @option{-Wundef}. These
2748 tools can help find bugs and unclear code, but they can also generate
2749 so many false alarms that it hurts readability to silence them with
2750 unnecessary casts, wrappers, and other complications. For example,
2751 please don't insert casts to @code{void} or calls to do-nothing
2752 functions merely to pacify a lint checker.
2754 Declarations of external functions and functions to appear later in the
2755 source file should all go in one place near the beginning of the file
2756 (somewhere before the first function definition in the file), or else
2757 should go in a header file. Don't put @code{extern} declarations inside
2760 @cindex temporary variables
2761 It used to be common practice to use the same local variables (with
2762 names like @code{tem}) over and over for different values within one
2763 function. Instead of doing this, it is better to declare a separate local
2764 variable for each distinct purpose, and give it a name which is
2765 meaningful. This not only makes programs easier to understand, it also
2766 facilitates optimization by good compilers. You can also move the
2767 declaration of each local variable into the smallest scope that includes
2768 all its uses. This makes the program even cleaner.
2770 Don't use local variables or parameters that shadow global identifiers.
2771 GCC's @samp{-Wshadow} option can detect this problem.
2773 @cindex multiple variables in a line
2774 Don't declare multiple variables in one declaration that spans lines.
2775 Start a new declaration on each line, instead. For example, instead
2801 (If they are global variables, each should have a comment preceding it
2804 When you have an @code{if}-@code{else} statement nested in another
2805 @code{if} statement, always put braces around the @code{if}-@code{else}.
2806 Thus, never write like this:
2829 If you have an @code{if} statement nested inside of an @code{else}
2830 statement, either write @code{else if} on one line, like this,
2840 with its @code{then}-part indented like the preceding @code{then}-part,
2841 or write the nested @code{if} within braces like this:
2853 Don't declare both a structure tag and variables or typedefs in the
2854 same declaration. Instead, declare the structure tag separately
2855 and then use it to declare the variables or typedefs.
2857 Try to avoid assignments inside @code{if}-conditions (assignments
2858 inside @code{while}-conditions are ok). For example, don't write
2862 if ((foo = (char *) malloc (sizeof *foo)) == NULL)
2863 fatal ("virtual memory exhausted");
2867 instead, write this:
2870 foo = (char *) malloc (sizeof *foo);
2872 fatal ("virtual memory exhausted");
2876 @section Naming Variables, Functions, and Files
2878 @cindex names of variables, functions, and files
2879 The names of global variables and functions in a program serve as
2880 comments of a sort. So don't choose terse names---instead, look for
2881 names that give useful information about the meaning of the variable or
2882 function. In a GNU program, names should be English, like other
2885 Local variable names can be shorter, because they are used only within
2886 one context, where (presumably) comments explain their purpose.
2888 Try to limit your use of abbreviations in symbol names. It is ok to
2889 make a few abbreviations, explain what they mean, and then use them
2890 frequently, but don't use lots of obscure abbreviations.
2892 Please use underscores to separate words in a name, so that the Emacs
2893 word commands can be useful within them. Stick to lower case; reserve
2894 upper case for macros and @code{enum} constants, and for name-prefixes
2895 that follow a uniform convention.
2897 For example, you should use names like @code{ignore_space_change_flag};
2898 don't use names like @code{iCantReadThis}.
2900 Variables that indicate whether command-line options have been
2901 specified should be named after the meaning of the option, not after
2902 the option-letter. A comment should state both the exact meaning of
2903 the option and its letter. For example,
2907 /* Ignore changes in horizontal whitespace (-b). */
2908 int ignore_space_change_flag;
2912 When you want to define names with constant integer values, use
2913 @code{enum} rather than @samp{#define}. GDB knows about enumeration
2916 @cindex file-name limitations
2918 You might want to make sure that none of the file names would conflict
2919 if the files were loaded onto an MS-DOS file system which shortens the
2920 names. You can use the program @code{doschk} to test for this.
2922 Some GNU programs were designed to limit themselves to file names of 14
2923 characters or less, to avoid file name conflicts if they are read into
2924 older System V systems. Please preserve this feature in the existing
2925 GNU programs that have it, but there is no need to do this in new GNU
2926 programs. @code{doschk} also reports file names longer than 14
2930 @node System Portability
2931 @section Portability between System Types
2932 @cindex portability, between system types
2934 In the Unix world, ``portability'' refers to porting to different Unix
2935 versions. For a GNU program, this kind of portability is desirable, but
2938 The primary purpose of GNU software is to run on top of the GNU kernel,
2939 compiled with the GNU C compiler, on various types of CPU@. So the
2940 kinds of portability that are absolutely necessary are quite limited.
2941 But it is important to support Linux-based GNU systems, since they
2942 are the form of GNU that is popular.
2944 Beyond that, it is good to support the other free operating systems
2945 (*BSD), and it is nice to support other Unix-like systems if you want
2946 to. Supporting a variety of Unix-like systems is desirable, although
2947 not paramount. It is usually not too hard, so you may as well do it.
2948 But you don't have to consider it an obligation, if it does turn out to
2952 The easiest way to achieve portability to most Unix-like systems is to
2953 use Autoconf. It's unlikely that your program needs to know more
2954 information about the host platform than Autoconf can provide, simply
2955 because most of the programs that need such knowledge have already been
2958 Avoid using the format of semi-internal data bases (e.g., directories)
2959 when there is a higher-level alternative (@code{readdir}).
2961 @cindex non-POSIX systems, and portability
2962 As for systems that are not like Unix, such as MS-DOS, Windows, VMS, MVS,
2963 and older Macintosh systems, supporting them is often a lot of work.
2964 When that is the case, it is better to spend your time adding features
2965 that will be useful on GNU and GNU/Linux, rather than on supporting
2966 other incompatible systems.
2968 If you do support Windows, please do not abbreviate it as ``win''.
2971 Usually we write the name ``Windows'' in full, but when brevity is
2972 very important (as in file names and some symbol names), we abbreviate
2973 it to ``w''. In GNU Emacs, for instance, we use @samp{w32} in file
2974 names of Windows-specific files, but the macro for Windows
2975 conditionals is called @code{WINDOWSNT}. In principle there could
2978 It is a good idea to define the ``feature test macro''
2979 @code{_GNU_SOURCE} when compiling your C files. When you compile on GNU
2980 or GNU/Linux, this will enable the declarations of GNU library extension
2981 functions, and that will usually give you a compiler error message if
2982 you define the same function names in some other way in your program.
2983 (You don't have to actually @emph{use} these functions, if you prefer
2984 to make the program more portable to other systems.)
2986 But whether or not you use these GNU extensions, you should avoid
2987 using their names for any other meanings. Doing so would make it hard
2988 to move your code into other GNU programs.
2990 @node CPU Portability
2991 @section Portability between CPUs
2993 @cindex data types, and portability
2994 @cindex portability, and data types
2995 Even GNU systems will differ because of differences among CPU
2996 types---for example, difference in byte ordering and alignment
2997 requirements. It is absolutely essential to handle these differences.
2998 However, don't make any effort to cater to the possibility that an
2999 @code{int} will be less than 32 bits. We don't support 16-bit machines
3002 You need not cater to the possibility that @code{long} will be smaller
3003 than pointers and @code{size_t}. We know of one such platform: 64-bit
3004 programs on Microsoft Windows. If you care about making your package
3005 run on Windows using Mingw64, you would need to deal with 8-byte
3006 pointers and 4-byte @code{long}, which would break this code:
3009 printf ("size = %lu\n", (unsigned long) sizeof array);
3010 printf ("diff = %ld\n", (long) (pointer2 - pointer1));
3013 Whether to support Mingw64, and Windows in general, in your package is
3014 your choice. The GNU Project doesn't say you have any responsibility to
3015 do so. Our goal is to replace proprietary systems, including Windows,
3016 not to enhance them. If people pressure you to make your program run
3017 on Windows, and you are not interested, you can respond with, ``Switch
3018 to GNU/Linux --- your freedom depends on it.''
3020 Predefined file-size types like @code{off_t} are an exception: they are
3021 longer than @code{long} on many platforms, so code like the above won't
3022 work with them. One way to print an @code{off_t} value portably is to
3023 print its digits yourself, one by one.
3025 Don't assume that the address of an @code{int} object is also the
3026 address of its least-significant byte. This is false on big-endian
3027 machines. Thus, don't make the following mistake:
3032 while ((c = getchar ()) != EOF)
3033 write (file_descriptor, &c, 1);
3036 @noindent Instead, use @code{unsigned char} as follows. (The @code{unsigned}
3037 is for portability to unusual systems where @code{char} is signed and
3038 where there is integer overflow checking.)
3042 while ((c = getchar ()) != EOF)
3044 unsigned char u = c;
3045 write (file_descriptor, &u, 1);
3049 @cindex casting pointers to integers
3050 Avoid casting pointers to integers if you can. Such casts greatly
3051 reduce portability, and in most programs they are easy to avoid. In the
3052 cases where casting pointers to integers is essential---such as, a Lisp
3053 interpreter which stores type information as well as an address in one
3054 word---you'll have to make explicit provisions to handle different word
3055 sizes. You will also need to make provision for systems in which the
3056 normal range of addresses you can get from @code{malloc} starts far away
3060 @node System Functions
3061 @section Calling System Functions
3063 @cindex C library functions, and portability
3064 @cindex POSIX functions, and portability
3065 @cindex library functions, and portability
3066 @cindex portability, and library functions
3068 Historically, C implementations differed substantially, and many
3069 systems lacked a full implementation of ANSI/ISO C89. Nowadays,
3070 however, all practical systems have a C89 compiler and GNU C supports
3071 almost all of C99 and some of C11. Similarly, most systems implement
3072 POSIX.1-2001 libraries and tools, and many have POSIX.1-2008.
3074 Hence, there is little reason to support old C or non-POSIX systems,
3075 and you may want to take advantage of standard C and POSIX to write
3076 clearer, more portable, or faster code. You should use standard
3077 interfaces where possible; but if GNU extensions make your program
3078 more maintainable, powerful, or otherwise better, don't hesitate to
3079 use them. In any case, don't make your own declaration of system
3080 functions; that's a recipe for conflict.
3082 Despite the standards, nearly every library function has some sort of
3083 portability issue on some system or another. Here are some examples:
3087 Names with trailing @code{/}'s are mishandled on many platforms.
3090 @code{long double} may be unimplemented; floating values Infinity and
3091 NaN are often mishandled; output for large precisions may be
3095 May return @code{int} instead of @code{ssize_t}.
3098 On Windows, @code{errno} is not set on failure.
3102 @uref{https://www.gnu.org/software/gnulib/, Gnulib} is a big help in
3103 this regard. Gnulib provides implementations of standard interfaces
3104 on many of the systems that lack them, including portable
3105 implementations of enhanced GNU interfaces, thereby making their use
3106 portable, and of POSIX-1.2008 interfaces, some of which are missing
3107 even on up-to-date GNU systems.
3109 @findex xmalloc, in Gnulib
3110 @findex error messages, in Gnulib
3111 @findex data structures, in Gnulib
3112 Gnulib also provides many useful non-standard interfaces; for example,
3113 C implementations of standard data structures (hash tables, binary
3114 trees), error-checking type-safe wrappers for memory allocation
3115 functions (@code{xmalloc}, @code{xrealloc}), and output of error
3118 Gnulib integrates with GNU Autoconf and Automake to remove much of the
3119 burden of writing portable code from the programmer: Gnulib makes your
3120 configure script automatically determine what features are missing and
3121 use the Gnulib code to supply the missing pieces.
3123 The Gnulib and Autoconf manuals have extensive sections on
3124 portability: @ref{Top,, Introduction, gnulib, Gnulib} and
3125 @pxref{Portable C and C++,,, autoconf, Autoconf}. Please consult them
3126 for many more details.
3129 @node Internationalization
3130 @section Internationalization
3131 @cindex internationalization
3134 GNU has a library called GNU gettext that makes it easy to translate the
3135 messages in a program into various languages. You should use this
3136 library in every program. Use English for the messages as they appear
3137 in the program, and let gettext provide the way to translate them into
3140 Using GNU gettext involves putting a call to the @code{gettext} macro
3141 around each string that might need translation---like this:
3144 printf (gettext ("Processing file '%s'..."), file);
3148 This permits GNU gettext to replace the string @code{"Processing file
3149 '%s'..."} with a translated version.
3151 Once a program uses gettext, please make a point of writing calls to
3152 @code{gettext} when you add new strings that call for translation.
3154 Using GNU gettext in a package involves specifying a @dfn{text domain
3155 name} for the package. The text domain name is used to separate the
3156 translations for this package from the translations for other packages.
3157 Normally, the text domain name should be the same as the name of the
3158 package---for example, @samp{coreutils} for the GNU core utilities.
3160 @cindex message text, and internationalization
3161 To enable gettext to work well, avoid writing code that makes
3162 assumptions about the structure of words or sentences. When you want
3163 the precise text of a sentence to vary depending on the data, use two or
3164 more alternative string constants each containing a complete sentences,
3165 rather than inserting conditionalized words or phrases into a single
3168 Here is an example of what not to do:
3171 printf ("%s is full", capacity > 5000000 ? "disk" : "floppy disk");
3174 If you apply gettext to all strings, like this,
3177 printf (gettext ("%s is full"),
3178 capacity > 5000000 ? gettext ("disk") : gettext ("floppy disk"));
3182 the translator will hardly know that "disk" and "floppy disk" are meant to
3183 be substituted in the other string. Worse, in some languages (like French)
3184 the construction will not work: the translation of the word "full" depends
3185 on the gender of the first part of the sentence; it happens to be not the
3186 same for "disk" as for "floppy disk".
3188 Complete sentences can be translated without problems:
3191 printf (capacity > 5000000 ? gettext ("disk is full")
3192 : gettext ("floppy disk is full"));
3195 A similar problem appears at the level of sentence structure with this
3199 printf ("# Implicit rule search has%s been done.\n",
3200 f->tried_implicit ? "" : " not");
3204 Adding @code{gettext} calls to this code cannot give correct results for
3205 all languages, because negation in some languages requires adding words
3206 at more than one place in the sentence. By contrast, adding
3207 @code{gettext} calls does the job straightforwardly if the code starts
3211 printf (f->tried_implicit
3212 ? "# Implicit rule search has been done.\n",
3213 : "# Implicit rule search has not been done.\n");
3216 Another example is this one:
3219 printf ("%d file%s processed", nfiles,
3220 nfiles != 1 ? "s" : "");
3224 The problem with this example is that it assumes that plurals are made
3225 by adding `s'. If you apply gettext to the format string, like this,
3228 printf (gettext ("%d file%s processed"), nfiles,
3229 nfiles != 1 ? "s" : "");
3233 the message can use different words, but it will still be forced to use
3234 `s' for the plural. Here is a better way, with gettext being applied to
3235 the two strings independently:
3238 printf ((nfiles != 1 ? gettext ("%d files processed")
3239 : gettext ("%d file processed")),
3244 But this still doesn't work for languages like Polish, which has three
3245 plural forms: one for nfiles == 1, one for nfiles == 2, 3, 4, 22, 23, 24, ...
3246 and one for the rest. The GNU @code{ngettext} function solves this problem:
3249 printf (ngettext ("%d files processed", "%d file processed", nfiles),
3255 @section Character Set
3256 @cindex character set
3258 @cindex ASCII characters
3259 @cindex non-ASCII characters
3261 Sticking to the ASCII character set (plain text, 7-bit characters) is
3262 preferred in GNU source code comments, text documents, and other
3263 contexts, unless there is good reason to do something else because of
3264 the application domain. For example, if source code deals with the
3265 French Revolutionary calendar, it is OK if its literal strings contain
3266 accented characters in month names like ``Flor@'eal''. Also, it is OK
3267 (but not required) to use non-ASCII characters to represent proper
3268 names of contributors in change logs (@pxref{Change Logs}).
3270 If you need to use non-ASCII characters, you should normally stick
3271 with one encoding, certainly within a single file. UTF-8 is likely to
3275 @node Quote Characters
3276 @section Quote Characters
3277 @cindex quote characters
3278 @cindex locale-specific quote characters
3281 @cindex opening quote
3282 @cindex single quote
3283 @cindex double quote
3284 @cindex grave accent
3285 @set txicodequoteundirected
3286 @set txicodequotebacktick
3288 In the C locale, the output of GNU programs should stick to plain
3289 ASCII for quotation characters in messages to users: preferably 0x22
3290 (@samp{"}) or 0x27 (@samp{'}) for both opening and closing quotes.
3291 Although GNU programs traditionally used 0x60 (@samp{`}) for opening
3292 and 0x27 (@samp{'}) for closing quotes, nowadays quotes @samp{`like
3293 this'} are typically rendered asymmetrically, so quoting @samp{"like
3294 this"} or @samp{'like this'} typically looks better.
3296 It is ok, but not required, for GNU programs to generate
3297 locale-specific quotes in non-C locales. For example:
3300 printf (gettext ("Processing file '%s'..."), file);
3304 Here, a French translation might cause @code{gettext} to return the
3305 string @code{"Traitement de fichier
3306 @guilsinglleft{}@tie{}%s@tie{}@guilsinglright{}..."}, yielding quotes
3307 more appropriate for a French locale.
3309 Sometimes a program may need to use opening and closing quotes
3310 directly. By convention, @code{gettext} translates the string
3311 @samp{"`"} to the opening quote and the string @samp{"'"} to the
3312 closing quote, and a program can use these translations. Generally,
3313 though, it is better to translate quote characters in the context of
3316 If the output of your program is ever likely to be parsed by another
3317 program, it is good to provide an option that makes this parsing
3318 reliable. For example, you could escape special characters using
3319 conventions from the C language or the Bourne shell. See for example
3320 the option @option{--quoting-style} of GNU @code{ls}.
3322 @clear txicodequoteundirected
3323 @clear txicodequotebacktick
3330 If you use @code{mmap} to read or write files, don't assume it either
3331 works on all files or fails for all files. It may work on some files
3334 The proper way to use @code{mmap} is to try it on the specific file for
3335 which you want to use it---and if @code{mmap} doesn't work, fall back on
3336 doing the job in another way using @code{read} and @code{write}.
3338 The reason this precaution is needed is that the GNU kernel (the HURD)
3339 provides a user-extensible file system, in which there can be many
3340 different kinds of ``ordinary files''. Many of them support
3341 @code{mmap}, but some do not. It is important to make programs handle
3342 all these kinds of files.
3346 @chapter Documenting Programs
3347 @cindex documentation
3349 A GNU program should ideally come with full free documentation, adequate
3350 for both reference and tutorial purposes. If the package can be
3351 programmed or extended, the documentation should cover programming or
3352 extending it, as well as just using it.
3355 * GNU Manuals:: Writing proper manuals.
3356 * Doc Strings and Manuals:: Compiling doc strings doesn't make a manual.
3357 * Manual Structure Details:: Specific structure conventions.
3358 * License for Manuals:: Writing the distribution terms for a manual.
3359 * Manual Credits:: Giving credit to documentation contributors.
3360 * Printed Manuals:: Mentioning the printed manual.
3361 * NEWS File:: NEWS files supplement manuals.
3362 * Change Logs:: Recording changes.
3363 * Man Pages:: Man pages are secondary.
3364 * Reading other Manuals:: How far you can go in learning
3369 @section GNU Manuals
3371 The preferred document format for the GNU system is the Texinfo
3372 formatting language. Every GNU package should (ideally) have
3373 documentation in Texinfo both for reference and for learners. Texinfo
3374 makes it possible to produce a good quality formatted book, using
3375 @TeX{}, and to generate an Info file. It is also possible to generate
3376 HTML output from Texinfo source. See the Texinfo manual, either the
3377 hardcopy, or the on-line version available through @code{info} or the
3378 Emacs Info subsystem (@kbd{C-h i}).
3380 Nowadays some other formats such as Docbook and Sgmltexi can be
3381 converted automatically into Texinfo. It is ok to produce the Texinfo
3382 documentation by conversion this way, as long as it gives good results.
3384 Make sure your manual is clear to a reader who knows nothing about the
3385 topic and reads it straight through. This means covering basic topics
3386 at the beginning, and advanced topics only later. This also means
3387 defining every specialized term when it is first used.
3389 Remember that the audience for a GNU manual (and other GNU
3390 documentation) is global, and that it will be used for years, maybe
3391 decades. This means that the reader could have very different cultural
3392 reference points. Decades from now, all but old folks will have very
3393 different cultural reference points; many things that "everyone knows
3394 about" today may be mostly forgotten.
3396 For this reason, try to avoid writing in a way that depends on
3397 cultural reference points for proper understanding, or that refers to them in
3398 ways that would impede reading for someone that doesn't recognize them.
3400 Likewise, be conservative in your choice of words (aside from technical
3401 terms), linguistic constructs, and spelling: aim to make them
3402 intelligible to readers from ten years ago. In any contest for
3403 trendiness, GNU writing should not even qualify to enter.
3405 It is ok to refer once in a rare while to spatially or temporally
3406 localized reference points or facts, if it is directly pertinent or as
3407 an aside. Changing these few things (which in any case stand out) when
3408 they no longer make sense will not be a lot of work.
3410 By contrast, it is always proper to refer to concepts of GNU and the
3411 free software movement, when they are pertinent. These are a central
3412 part of our message, so we should take advantage of opportunities to
3413 mention them. They are fundamental moral positions, so they will
3414 rarely if ever change.
3416 Programmers tend to carry over the structure of the program as the
3417 structure for its documentation. But this structure is not
3418 necessarily good for explaining how to use the program; it may be
3419 irrelevant and confusing for a user.
3421 Instead, the right way to structure documentation is according to the
3422 concepts and questions that a user will have in mind when reading it.
3423 This principle applies at every level, from the lowest (ordering
3424 sentences in a paragraph) to the highest (ordering of chapter topics
3425 within the manual). Sometimes this structure of ideas matches the
3426 structure of the implementation of the software being documented---but
3427 often they are different. An important part of learning to write good
3428 documentation is to learn to notice when you have unthinkingly
3429 structured the documentation like the implementation, stop yourself,
3430 and look for better alternatives.
3432 For example, each program in the GNU system probably ought to be
3433 documented in one manual; but this does not mean each program should
3434 have its own manual. That would be following the structure of the
3435 implementation, rather than the structure that helps the user
3438 Instead, each manual should cover a coherent @emph{topic}. For example,
3439 instead of a manual for @code{diff} and a manual for @code{diff3}, we
3440 have one manual for ``comparison of files'' which covers both of those
3441 programs, as well as @code{cmp}. By documenting these programs
3442 together, we can make the whole subject clearer.
3444 The manual which discusses a program should certainly document all of
3445 the program's command-line options and all of its commands. It should
3446 give examples of their use. But don't organize the manual as a list
3447 of features. Instead, organize it logically, by subtopics. Address
3448 the questions that a user will ask when thinking about the job that
3449 the program does. Don't just tell the reader what each feature can
3450 do---say what jobs it is good for, and show how to use it for those
3451 jobs. Explain what is recommended usage, and what kinds of usage
3454 In general, a GNU manual should serve both as tutorial and reference.
3455 It should be set up for convenient access to each topic through Info,
3456 and for reading straight through (appendixes aside). A GNU manual
3457 should give a good introduction to a beginner reading through from the
3458 start, and should also provide all the details that hackers want.
3459 The Bison manual is a good example of this---please take a look at it
3460 to see what we mean.
3462 That is not as hard as it first sounds. Arrange each chapter as a
3463 logical breakdown of its topic, but order the sections, and write their
3464 text, so that reading the chapter straight through makes sense. Do
3465 likewise when structuring the book into chapters, and when structuring a
3466 section into paragraphs. The watchword is, @emph{at each point, address
3467 the most fundamental and important issue raised by the preceding text.}
3469 If necessary, add extra chapters at the beginning of the manual which
3470 are purely tutorial and cover the basics of the subject. These provide
3471 the framework for a beginner to understand the rest of the manual. The
3472 Bison manual provides a good example of how to do this.
3474 To serve as a reference, a manual should have an Index that lists all
3475 the functions, variables, options, and important concepts that are
3476 part of the program. One combined Index should do for a short manual,
3477 but sometimes for a complex package it is better to use multiple
3478 indices. The Texinfo manual includes advice on preparing good index
3479 entries, see @ref{Index Entries, , Making Index Entries, texinfo, GNU
3480 Texinfo}, and see @ref{Indexing Commands, , Defining the Entries of an
3481 Index, texinfo, GNU Texinfo}.
3483 Don't use Unix man pages as a model for how to write GNU documentation;
3484 most of them are terse, badly structured, and give inadequate
3485 explanation of the underlying concepts. (There are, of course, some
3486 exceptions.) Also, Unix man pages use a particular format which is
3487 different from what we use in GNU manuals.
3489 Please include an email address in the manual for where to report
3490 bugs @emph{in the text of the manual}.
3492 Please do not use the term ``pathname'' that is used in Unix
3493 documentation; use ``file name'' (two words) instead. We use the term
3494 ``path'' only for search paths, which are lists of directory names.
3496 Please do not use the term ``illegal'' to refer to erroneous input to
3497 a computer program. Please use ``invalid'' for this, and reserve the
3498 term ``illegal'' for activities prohibited by law.
3500 Please do not write @samp{()} after a function name just to indicate
3501 it is a function. @code{foo ()} is not a function, it is a function
3502 call with no arguments.
3504 Whenever possible, please stick to the active voice, avoiding the
3505 passive, and use the present tense, not the future tense. For
3506 instance, write ``The function @code{foo} returns a list containing
3507 @var{a} and @var{b}'' rather than ``A list containing @var{a} and
3508 @var{b} will be returned.'' One advantage of the active voice is it
3509 requires you to state the subject of the sentence; with the passive
3510 voice, you might omit the subject, which leads to vagueness.
3512 It is proper to use the future tense when grammar demands it, as in,
3513 ``If you type @kbd{x}, the computer will self-destruct in 10
3516 @node Doc Strings and Manuals
3517 @section Doc Strings and Manuals
3519 Some programming systems, such as Emacs, provide a documentation string
3520 for each function, command or variable. You may be tempted to write a
3521 reference manual by compiling the documentation strings and writing a
3522 little additional text to go around them---but you must not do it. That
3523 approach is a fundamental mistake. The text of well-written
3524 documentation strings will be entirely wrong for a manual.
3526 A documentation string needs to stand alone---when it appears on the
3527 screen, there will be no other text to introduce or explain it.
3528 Meanwhile, it can be rather informal in style.
3530 The text describing a function or variable in a manual must not stand
3531 alone; it appears in the context of a section or subsection. Other text
3532 at the beginning of the section should explain some of the concepts, and
3533 should often make some general points that apply to several functions or
3534 variables. The previous descriptions of functions and variables in the
3535 section will also have given information about the topic. A description
3536 written to stand alone would repeat some of that information; this
3537 redundancy looks bad. Meanwhile, the informality that is acceptable in
3538 a documentation string is totally unacceptable in a manual.
3540 The only good way to use documentation strings in writing a good manual
3541 is to use them as a source of information for writing good text.
3543 @node Manual Structure Details
3544 @section Manual Structure Details
3545 @cindex manual structure
3547 The title page of the manual should state the version of the programs or
3548 packages documented in the manual. The Top node of the manual should
3549 also contain this information. If the manual is changing more
3550 frequently than or independent of the program, also state a version
3551 number for the manual in both of these places.
3553 Each program documented in the manual should have a node named
3554 @samp{@var{program} Invocation} or @samp{Invoking @var{program}}. This
3555 node (together with its subnodes, if any) should describe the program's
3556 command line arguments and how to run it (the sort of information people
3557 would look for in a man page). Start with an @samp{@@example}
3558 containing a template for all the options and arguments that the program
3561 Alternatively, put a menu item in some menu whose item name fits one of
3562 the above patterns. This identifies the node which that item points to
3563 as the node for this purpose, regardless of the node's actual name.
3565 The @samp{--usage} feature of the Info reader looks for such a node
3566 or menu item in order to find the relevant text, so it is essential
3567 for every Texinfo file to have one.
3569 If one manual describes several programs, it should have such a node for
3570 each program described in the manual.
3572 @node License for Manuals
3573 @section License for Manuals
3574 @cindex license for manuals
3576 Please use the GNU Free Documentation License for all GNU manuals that
3577 are more than a few pages long. Likewise for a collection of short
3578 documents---you only need one copy of the GNU FDL for the whole
3579 collection. For a single short document, you can use a very permissive
3580 non-copyleft license, to avoid taking up space with a long license.
3582 See @uref{https://www.gnu.org/copyleft/fdl-howto.html} for more explanation
3583 of how to employ the GFDL.
3585 Note that it is not obligatory to include a copy of the GNU GPL or GNU
3586 LGPL in a manual whose license is neither the GPL nor the LGPL@. It can
3587 be a good idea to include the program's license in a large manual; in a
3588 short manual, whose size would be increased considerably by including
3589 the program's license, it is probably better not to include it.
3591 @node Manual Credits
3592 @section Manual Credits
3593 @cindex credits for manuals
3595 Please credit the principal human writers of the manual as the authors,
3596 on the title page of the manual. If a company sponsored the work, thank
3597 the company in a suitable place in the manual, but do not cite the
3598 company as an author.
3600 @node Printed Manuals
3601 @section Printed Manuals
3603 The FSF publishes some GNU manuals in printed form. To encourage sales
3604 of these manuals, the on-line versions of the manual should mention at
3605 the very start that the printed manual is available and should point at
3606 information for getting it---for instance, with a link to the page
3607 @url{https://www.gnu.org/order/order.html}. This should not be included
3608 in the printed manual, though, because there it is redundant.
3610 It is also useful to explain in the on-line forms of the manual how the
3611 user can print out the manual from the sources.
3614 @section The NEWS File
3615 @cindex @file{NEWS} file
3617 In addition to its manual, the package should have a file named
3618 @file{NEWS} which contains a list of user-visible changes worth
3619 mentioning. In each new release, add items to the front of the file and
3620 identify the version they pertain to. Don't discard old items; leave
3621 them in the file after the newer items. This way, a user upgrading from
3622 any previous version can see what is new.
3624 If the @file{NEWS} file gets very long, move some of the older items
3625 into a file named @file{ONEWS} and put a note at the end referring the
3629 @section Change Logs
3632 Keep a change log to describe all the changes made to program source
3633 files. The purpose of this is so that people investigating bugs in the
3634 future will know about the changes that might have introduced the bug.
3635 Often a new bug can be found by looking at what was recently changed.
3636 More importantly, change logs can help you eliminate conceptual
3637 inconsistencies between different parts of a program, by giving you a
3638 history of how the conflicting concepts arose, who they came from, and
3639 why the conflicting changes were made.
3641 @cindex software forensics, and change logs
3642 Therefore, change logs should be detailed enough and accurate enough
3643 to provide the information commonly required for such @dfn{software
3644 forensics}. Specifically, change logs should make finding answers to
3645 the following questions easy:
3649 What changes affected a particular source file?
3652 Was a particular source file renamed or moved, and if so, as part of
3656 What changes affected a given function or macro or definition of a
3660 Was a function (or a macro or the definition of a data structure)
3661 renamed or moved from another file, and if so, as part of which
3665 What changes deleted a function (or macro or data structure)?
3668 What was the rationale for a given change, and what were its main
3672 Is there any additional information regarding the change, and if so,
3673 where can it be found?
3677 @cindex version control system, for keeping change logs
3678 Historically, change logs were maintained on specially formatted
3679 files. Nowadays, projects commonly keep their source files under a
3680 @dfn{version control system} (VCS), such as Git,
3681 Subversion, or Mercurial. If the VCS repository is publicly
3682 accessible, and changes are committed to it separately (one commit for
3683 each logical changeset) and record the authors of each change, then
3684 the information recorded by the VCS can be used to produce
3685 the change logs out of VCS logs, and to answer the above
3686 questions by using the suitable VCS commands. (However, the
3687 VCS log messages still need to provide some supporting
3688 information, as described below.) Projects that maintain such
3689 VCS repositories can decide not to maintain separate change
3690 log files, and instead rely on the VCS to keep the change
3693 If you decide not to maintain separate change log files, you should
3694 still consider providing them in the release tarballs, for the benefit
3695 of users who'd like to review the change logs without accessing the
3696 project's VCS repository. Scripts exist that can produce
3697 @file{ChangeLog} files from the VCS logs; for example, the
3698 @file{gitlog-to-changelog} script, which is part of Gnulib, can do
3699 that for Git repositories. In Emacs, the command @kbd{C-x v a}
3700 (@code{vc-update-change-log}) does the job of incrementally updating a
3701 @file{ChangeLog} file from the VCS logs.
3703 If separate change log files @emph{are} maintained, they are normally
3704 called @file{ChangeLog}, and each such file covers an entire
3705 directory. Each directory can have its own change log file, or a
3706 directory can use the change log of its parent directory---it's up to
3710 * Change Log Concepts::
3711 * Style of Change Logs::
3713 * Conditional Changes::
3714 * Indicating the Part Changed::
3717 @node Change Log Concepts
3718 @subsection Change Log Concepts and Conventions
3720 @cindex changeset, in a change log
3721 @cindex batch of changes, in a change log
3722 You can think of the change log as a conceptual ``undo list'' which
3723 states how earlier versions were different from the current version.
3724 People can see the current version; they don't need the change log to
3725 tell them what is in it. What they want from a change log is a clear
3726 explanation of how the earlier version differed. Each @dfn{entry} in
3727 a change log describes either an individual change or the smallest
3728 batch of changes that belong together, also known as a @dfn{changeset}.
3730 @cindex title, change log entry
3731 @cindex header line, change log entry
3732 It is a good idea to start the change log entry with a @dfn{header
3733 line}: a single line that is a complete sentence which summarizes the
3734 changeset. If you keep the change log in a VCS, this
3735 should be a requirement, as VCS commands that show the
3736 change log in abbreviated form, such as @kbd{git log --oneline}, treat
3737 the header line specially. (In a @file{ChangeLog} file, the header
3738 line follows a line that says who was the author of the change and
3739 when it was installed.)
3741 @cindex description, change log entry
3742 Follow the change log entry's header line with a description of the
3743 overall change. This should be as long as needed to give a clear
3744 description. Pay special attention to aspects of the changeset not
3745 easily gleaned from the diffs or from the names of modified files and
3746 functions: the overall idea of the change and the need for it, and the
3747 relations, if any, between changes made to different files/functions.
3748 If the change or its reasons were discussed on some public forum, such
3749 as the project's issue tracker or mailing list, it is a good idea to
3750 summarize the main points of that discussion in the change's
3751 description, and include a pointer to that discussion or the issue ID
3752 for those who'd like to read it in full.
3754 The best place to explain how parts of the new code work with other code
3755 is in comments in the code, not in the change log.
3757 If you think that a change calls for explanation of @emph{why} the
3758 change was needed---that is, what problem the old code had such that
3759 it required this change---you're probably right. Please put the
3760 explanation in comments in the code, where people will see it whenever
3761 they see the code. An example of such an explanation is, ``This
3762 function used to be iterative, but that failed when MUMBLE was a
3763 tree.'' (Though such a simple reason would not need this kind of
3766 The best place for other kinds of explanation of the change is in the
3767 change log entry. In particular, comments usually will not say why
3768 some code was deleted or moved to another place---that belongs to the
3769 description of the change which did that.
3771 Following the free-text description of the change, it is a good idea
3772 to give a list of names of the entities or definitions that you
3773 changed, according to the files they are in, and what was changed in
3774 each one. @xref{Style of Change Logs}. If a project uses a modern
3775 VCS to keep the change log information, as described in
3776 @ref{Change Logs}, explicitly listing the files and functions that
3777 were changed is not strictly necessary, and in some cases (like
3778 identical mechanical changes in many places) even tedious. It is up
3779 to you to decide whether to allow your project's developers to omit
3780 the list of changed files and functions from the log entries, and
3781 whether to allow such omissions under some specific conditions.
3782 However, while making this decision, please consider the following
3783 benefits of providing the list of changed entities with each change:
3787 Generation of useful @file{ChangeLog} files from VCS logs
3788 becomes more difficult if the change log entries don't list the
3789 modified functions/macros, because VCS commands cannot
3790 reliably reproduce their names from the commit information alone. For
3791 example, when there is a change in the header part of a function
3792 definition, the heading of the diff hunk as shown in the VCS log
3793 commands will name the wrong function as being modified (usually, the
3794 function defined before the one being modified), so using those diffs
3795 to glean the names of the modified functions will produce inaccurate
3796 results. You will need to use specialized scripts, such as gnulib's
3797 @file{vcs-to-changelog.py}, mentioned below, to solve these
3798 difficulties, and make sure it supports the source languages used by
3802 While modern VCS commands, such as Git's @kbd{git log -L}
3803 and @kbd{git log -G}, provide powerful means for finding changes that
3804 affected a certain function or macro or data structure (and thus might
3805 make @file{ChangeLog} files unnecessary if you have the repository
3806 available), they can sometimes fail. For example, @kbd{git log -L}
3807 doesn't support syntax of some programming languages out of the box.
3808 Mentioning the modified functions/macros explicitly allows finding the
3809 related changes simply and reliably.
3812 Some VCS commands have difficulties or limitations when
3813 tracking changes across file moves or renames. Again, if the entities
3814 are mentioned explicitly, those difficulties can be overcome.
3817 Users that review changes using the generated @file{ChangeLog} files
3818 may not have the repository and the VCS commands available
3819 to them. Naming the modified entities alleviates that problem.
3823 For these reasons, providing lists of modified files and functions
3824 with each change makes the change logs more useful, and we therefore
3825 recommend to include them whenever possible and practical.
3827 It is also possible to generate the lists naming the modified entities
3828 by running a script. One such script is @file{mklog.py} (written in
3829 Python 3); it is used by the @code{GCC} project. Gnulib provides
3830 another variant of such a script, called @file{vcs-to-changelog.py},
3831 part of the @code{vcs-to-changelog} module. Note that these scripts
3832 currently support fewer programming languages than the manual commands
3833 provided by Emacs (@pxref{Style of Change Logs}). Therefore, the
3834 above mentioned method of generating the @code{ChangeLog} file from
3835 the VCS commit history, for instance via the
3836 @code{gitlog-to-changelog} script, usually gives better
3837 results---provided that the contributors stick to providing good
3840 @node Style of Change Logs
3841 @subsection Style of Change Logs
3842 @cindex change logs, style
3844 Here are some simple examples of change log entries, starting with the
3845 header line that says who made the change and when it was installed,
3846 followed by descriptions of specific changes. (These examples are
3847 drawn from Emacs.) Keep in mind that the line which shows the date of
3848 the change and the author's name and email address is needed only in a
3849 separate @file{ChangeLog} file, not when the change logs are kept in a
3853 2019-08-29 Noam Postavsky <npostavs@@gmail.com>
3855 Handle completely undecoded input in term (Bug#29918)
3857 * lisp/term.el (term-emulate-terminal): Avoid errors if the whole
3858 decoded string is eight-bit characters. Don't attempt to save the
3859 string for next iteration in that case.
3860 * test/lisp/term-tests.el (term-decode-partial)
3861 (term-undecodable-input): New tests.
3863 2019-06-15 Paul Eggert <eggert@@cs.ucla.edu>
3865 Port to platforms where tputs is in libtinfow
3867 * configure.ac (tputs_library): Also try tinfow, ncursesw (Bug#33977).
3869 2019-02-08 Eli Zaretskii <eliz@@gnu.org>
3871 Improve documentation of 'date-to-time' and 'parse-time-string'
3873 * doc/lispref/os.texi (Time Parsing): Document
3874 'parse-time-string', and refer to it for the description of
3875 the argument of 'date-to-time'.
3877 * lisp/calendar/time-date.el (date-to-time): Refer in the doc
3878 string to 'parse-time-string' for more information about the
3879 format of the DATE argument. (Bug#34303)
3882 If you mention the names of the modified functions or variables, it's
3883 important to name them in full. Don't abbreviate function or variable
3884 names, and don't combine them. Subsequent maintainers will often
3885 search for a function name to find all the change log entries that
3886 pertain to it; if you abbreviate the name, they won't find it when
3889 For example, some people are tempted to abbreviate groups of function
3890 names by writing @samp{* register.el (@{insert,jump-to@}-register)};
3891 this is not a good idea, since searching for @code{jump-to-register} or
3892 @code{insert-register} would not find that entry.
3894 Separate unrelated change log entries with blank lines. Don't put
3895 blank lines between individual changes of an entry. You can omit the
3896 file name and the asterisk when successive individual changes are in
3899 Break long lists of function names by closing continued lines with
3900 @samp{)}, rather than @samp{,}, and opening the continuation with
3901 @samp{(}. This makes highlighting in Emacs work better.
3905 * src/keyboard.c (menu_bar_items, tool_bar_items)
3906 (Fexecute_extended_command): Deal with 'keymap' property.
3909 The easiest way to add an entry to @file{ChangeLog} is with the Emacs
3910 command @kbd{M-x add-change-log-entry}, or its variant @kbd{C-x 4 a}
3911 (@code{add-change-log-entry-other-window}). This automatically
3912 collects the name of the changed file and the changed function or
3913 variable, and formats a change log entry according to the conventions
3914 described above, leaving it up to you to describe the changes you made
3915 to that function or variable.
3917 When you install someone else's changes, put the contributor's name in
3918 the change log entry rather than in the text of the entry. In other
3922 2002-07-14 John Doe <jdoe@@gnu.org>
3924 * sewing.c: Make it sew.
3931 2002-07-14 Usual Maintainer <usual@@gnu.org>
3933 * sewing.c: Make it sew. Patch by jdoe@@gnu.org.
3936 When committing someone else's changes into a VCS, use the
3937 VCS features to specify the author. For example, with Git,
3938 use @kbd{git commit --author=@var{author}}.
3940 As for the date, that should be the date you applied the change.
3941 (With a VCS, use the appropriate command-line switches,
3942 e.g., @kbd{git commit --date=@var{date}}.)
3944 Modern VCS have commands to apply changes sent via email
3945 (e.g., Git has @kbd{git am}); in that case the author of the changeset
3946 and the date it was made will be automatically gleaned from the email
3947 message and recorded in the repository. If the patches are prepared
3948 with suitable VCS commands, such as @kbd{git format-patch},
3949 the email message body will also have the original author of the
3950 changeset, so resending or forwarding the message will not interfere
3951 with attributing the changes to their author. Thus, we recommend that
3952 you request your contributors to use commands such as @kbd{git
3953 format-patch} to prepare the patches.
3955 @node Simple Changes
3956 @subsection Simple Changes
3958 Certain simple kinds of changes don't need much detail in the change
3961 If the description of the change is short enough, it can serve as its
3965 2019-08-29 Eli Zaretskii <eliz@@gnu.org>
3967 * lisp/simple.el (kill-do-not-save-duplicates): Doc fix. (Bug#36827)
3970 When you change the calling sequence of a function in a simple fashion,
3971 and you change all the callers of the function to use the new calling
3972 sequence, there is no need to make individual entries for all the
3973 callers that you changed. Just write in the entry for the function
3974 being called, ``All callers changed''---like this:
3977 * keyboard.c (Fcommand_execute): New arg SPECIAL.
3978 All callers changed.
3981 When you change just comments or doc strings, it is enough to write an
3982 entry for the file, without mentioning the functions. Just ``Doc
3983 fixes'' is enough for the change log.
3985 When you make changes in many files that follow mechanically from one
3986 underlying change, it is enough to describe the underlying change.
3987 Here's an example of a change that affects all of the files in the
3991 2019-01-07 Paul Eggert <eggert@@cs.ucla.edu>
3993 Update copyright year to 2019
3995 Run 'TZ=UTC0 admin/update-copyright $(git ls-files)'.
3998 Test suite files are part of the software, so we recommend treating
3999 them as code for change-log purposes.
4001 There's no technical need to make change log entries for non-software
4002 files (manuals, help files, media files, etc.). This is because they
4003 are not susceptible to bugs that are hard to understand. To correct
4004 an error, you need not know the history of the erroneous passage; it
4005 is enough to compare what the file says with the actual facts.
4007 However, you should keep change logs for non-software files when the
4008 project gets copyright assignments from its contributors, so as to
4009 make the records of authorship more accurate. For that reason, we
4010 recommend to keep change logs for Texinfo sources of your project's
4013 @node Conditional Changes
4014 @subsection Conditional Changes
4015 @cindex conditional changes, and change logs
4016 @cindex change logs, conditional changes
4018 Source files can often contain code that is conditional to build-time
4019 or static conditions. For example, C programs can contain
4020 compile-time @code{#if} conditionals; programs implemented in
4021 interpreted languages can contain module imports of function
4022 definitions that are only performed for certain versions of the
4023 interpreter; and Automake @file{Makefile.am} files can contain
4024 variable definitions or target declarations that are only to be
4025 considered if a configure-time Automake conditional is true.
4027 Many changes are conditional as well: sometimes you add a new variable,
4028 or function, or even a new program or library, which is entirely
4029 dependent on a build-time condition. It is useful to indicate
4030 in the change log the conditions for which a change applies.
4032 Our convention for indicating conditional changes is to use
4033 @emph{square brackets around the name of the condition}.
4035 Conditional changes can happen in numerous scenarios and with many
4036 variations, so here are some examples to help clarify. This first
4037 example describes changes in C, Perl, and Python files which are
4038 conditional but do not have an associated function or entity name:
4041 * xterm.c [SOLARIS2]: Include <string.h>.
4042 * FilePath.pm [$^O eq 'VMS']: Import the VMS::Feature module.
4043 * framework.py [sys.version_info < (2, 6)]: Make "with" statement
4044 available by importing it from __future__,
4045 to support also python 2.5.
4048 Our other examples will for simplicity be limited to C, as the minor
4049 changes necessary to adapt them to other languages should be
4052 Next, here is an entry describing a new definition which is entirely
4053 conditional: the C macro @code{FRAME_WINDOW_P} is defined (and used)
4054 only when the macro @code{HAVE_X_WINDOWS} is defined:
4057 * frame.h [HAVE_X_WINDOWS] (FRAME_WINDOW_P): Macro defined.
4060 Next, an entry for a change within the function @code{init_display},
4061 whose definition as a whole is unconditional, but the changes
4062 themselves are contained in a @samp{#ifdef HAVE_LIBNCURSES}
4066 * dispnew.c (init_display) [HAVE_LIBNCURSES]: If X, call tgetent.
4069 Finally, here is an entry for a change that takes effect only when
4070 a certain macro is @emph{not} defined:
4073 * host.c (gethostname) [!HAVE_SOCKETS]: Replace with winsock version.
4077 @node Indicating the Part Changed
4078 @subsection Indicating the Part Changed
4080 Indicate the part of a function which changed by using angle brackets
4081 enclosing an indication of what the changed part does. Here is an entry
4082 for a change in the part of the function @code{sh-while-getopts} that
4083 deals with @code{sh} commands:
4086 * progmodes/sh-script.el (sh-while-getopts) <sh>: Handle case that
4087 user-specified option string is empty.
4095 In the GNU project, man pages are secondary. It is not necessary or
4096 expected for every GNU program to have a man page, but some of them do.
4097 It's your choice whether to include a man page in your program.
4099 When you make this decision, consider that supporting a man page
4100 requires continual effort each time the program is changed. The time
4101 you spend on the man page is time taken away from more useful work.
4103 For a simple program which changes little, updating the man page may be
4104 a small job. Then there is little reason not to include a man page, if
4107 For a large program that changes a great deal, updating a man page may
4108 be a substantial burden. If a user offers to donate a man page, you may
4109 find this gift costly to accept. It may be better to refuse the man
4110 page unless the same person agrees to take full responsibility for
4111 maintaining it---so that you can wash your hands of it entirely. If
4112 this volunteer later ceases to do the job, then don't feel obliged to
4113 pick it up yourself; it may be better to withdraw the man page from the
4114 distribution until someone else agrees to update it.
4116 When a program changes only a little, you may feel that the
4117 discrepancies are small enough that the man page remains useful without
4118 updating. If so, put a prominent note near the beginning of the man
4119 page stating that you don't maintain it and that the Texinfo manual
4120 is more authoritative. The note should say how to access the Texinfo
4123 Be sure that man pages include a copyright statement and free license.
4124 The simple all-permissive license is appropriate for simple man pages
4125 (@pxref{License Notices for Other Files,,,maintain,Information for GNU
4128 For long man pages, with enough explanation and documentation that
4129 they can be considered true manuals, use the GFDL (@pxref{License for
4132 Finally, the GNU help2man program
4133 (@uref{https://www.gnu.org/software/help2man/}) is one way to automate
4134 generation of a man page, in this case from @option{--help} output.
4135 This is sufficient in many cases.
4137 @node Reading other Manuals
4138 @section Reading other Manuals
4140 There may be non-free books or documentation files that describe the
4141 program you are documenting.
4143 It is ok to use these documents for reference, just as the author of a
4144 new algebra textbook can read other books on algebra. A large portion
4145 of any non-fiction book consists of facts, in this case facts about how
4146 a certain program works, and these facts are necessarily the same for
4147 everyone who writes about the subject. But be careful not to copy your
4148 outline structure, wording, tables or examples from preexisting non-free
4149 documentation. Copying from free documentation may be ok; please check
4150 with the FSF about the individual case.
4152 @node Managing Releases
4153 @chapter The Release Process
4156 Making a release is more than just bundling up your source files in a
4157 tar file and putting it up for FTP@. You should set up your software so
4158 that it can be configured to run on a variety of systems. Your Makefile
4159 should conform to the GNU standards described below, and your directory
4160 layout should also conform to the standards discussed below. Doing so
4161 makes it easy to include your package into the larger framework of
4165 * Configuration:: How configuration of GNU packages should work.
4166 * Makefile Conventions:: Makefile conventions.
4167 * Releases:: Making releases
4171 @section How Configuration Should Work
4172 @cindex program configuration
4175 Each GNU distribution should come with a shell script named
4176 @code{configure}. This script is given arguments which describe the
4177 kind of machine and system you want to compile the program for.
4178 The @code{configure} script must record the configuration options so
4179 that they affect compilation.
4181 The description here is the specification of the interface for the
4182 @code{configure} script in GNU packages. Many packages implement it
4183 using GNU Autoconf (@pxref{Top,, Introduction, autoconf, Autoconf})
4184 and/or GNU Automake (@pxref{Top,, Introduction, automake, Automake}),
4185 but you do not have to use these tools. You can implement it any way
4186 you like; for instance, by making @code{configure} be a wrapper around
4187 a completely different configuration system.
4189 Another way for the @code{configure} script to operate is to make a
4190 link from a standard name such as @file{config.h} to the proper
4191 configuration file for the chosen system. If you use this technique,
4192 the distribution should @emph{not} contain a file named
4193 @file{config.h}. This is so that people won't be able to build the
4194 program without configuring it first.
4196 Another thing that @code{configure} can do is to edit the Makefile. If
4197 you do this, the distribution should @emph{not} contain a file named
4198 @file{Makefile}. Instead, it should include a file @file{Makefile.in} which
4199 contains the input used for editing. Once again, this is so that people
4200 won't be able to build the program without configuring it first.
4202 If @code{configure} does write the @file{Makefile}, then @file{Makefile}
4203 should have a target named @file{Makefile} which causes @code{configure}
4204 to be rerun, setting up the same configuration that was set up last
4205 time. The files that @code{configure} reads should be listed as
4206 dependencies of @file{Makefile}.
4208 All the files which are output from the @code{configure} script should
4209 have comments at the beginning stating that they were generated
4210 automatically using @code{configure}. This is so that users won't think
4211 of trying to edit them by hand.
4213 The @code{configure} script should write a file named @file{config.status}
4214 which describes which configuration options were specified when the
4215 program was last configured. This file should be a shell script which,
4216 if run, will recreate the same configuration.
4218 The @code{configure} script should accept an option of the form
4219 @samp{--srcdir=@var{dirname}} to specify the directory where sources are found
4220 (if it is not the current directory). This makes it possible to build
4221 the program in a separate directory, so that the actual source directory
4224 If the user does not specify @samp{--srcdir}, then @code{configure} should
4225 check both @file{.} and @file{..} to see if it can find the sources. If
4226 it finds the sources in one of these places, it should use them from
4227 there. Otherwise, it should report that it cannot find the sources, and
4228 should exit with nonzero status.
4230 Usually the easy way to support @samp{--srcdir} is by editing a
4231 definition of @code{VPATH} into the Makefile. Some rules may need to
4232 refer explicitly to the specified source directory. To make this
4233 possible, @code{configure} can add to the Makefile a variable named
4234 @code{srcdir} whose value is precisely the specified directory.
4236 In addition, the @samp{configure} script should take options
4237 corresponding to most of the standard directory variables
4238 (@pxref{Directory Variables}). Here is the list:
4241 --prefix --exec-prefix --bindir --sbindir --libexecdir --sysconfdir
4242 --sharedstatedir --localstatedir --runstatedir
4243 --libdir --includedir --oldincludedir
4244 --datarootdir --datadir --infodir --localedir --mandir --docdir
4245 --htmldir --dvidir --pdfdir --psdir
4248 The @code{configure} script should also take an argument which specifies the
4249 type of system to build the program for. This argument should look like
4253 @var{cpu}-@var{company}-@var{system}
4256 For example, an Athlon-based GNU/Linux system might be
4257 @samp{i686-pc-linux-gnu}.
4259 The @code{configure} script needs to be able to decode all plausible
4260 alternatives for how to describe a machine. Thus,
4261 @samp{athlon-pc-gnu/linux} would be a valid alias. There is a shell
4263 @uref{https://git.savannah.gnu.org/@/gitweb/@/?p=config.git;a=blob_plain;f=config.sub;hb=HEAD,
4264 @file{config.sub}} that you can use as a subroutine to validate system
4265 types and canonicalize aliases.
4267 The @code{configure} script should also take the option
4268 @option{--build=@var{buildtype}}, which should be equivalent to a
4269 plain @var{buildtype} argument. For example, @samp{configure
4270 --build=i686-pc-linux-gnu} is equivalent to @samp{configure
4271 i686-pc-linux-gnu}. When the build type is not specified by an option
4272 or argument, the @code{configure} script should normally guess it using
4274 @uref{https://git.savannah.gnu.org/@/gitweb/@/?p=config.git;a=blob_plain;f=config.guess;hb=HEAD,
4275 @file{config.guess}}.
4277 @cindex optional features, configure-time
4278 Other options are permitted to specify in more detail the software
4279 or hardware present on the machine, to include or exclude optional parts
4280 of the package, or to adjust the name of some tools or arguments to them:
4283 @item --enable-@var{feature}@r{[}=@var{parameter}@r{]}
4284 Configure the package to build and install an optional user-level
4285 facility called @var{feature}. This allows users to choose which
4286 optional features to include. Giving an optional @var{parameter} of
4287 @samp{no} should omit @var{feature}, if it is built by default.
4289 No @samp{--enable} option should @strong{ever} cause one feature to
4290 replace another. No @samp{--enable} option should ever substitute one
4291 useful behavior for another useful behavior. The only proper use for
4292 @samp{--enable} is for questions of whether to build part of the program
4295 @item --with-@var{package}
4296 @c @r{[}=@var{parameter}@r{]}
4297 The package @var{package} will be installed, so configure this package
4298 to work with @var{package}.
4300 @c Giving an optional @var{parameter} of
4301 @c @samp{no} should omit @var{package}, if it is used by default.
4303 Possible values of @var{package} include
4304 @samp{gnu-as} (or @samp{gas}), @samp{gnu-ld}, @samp{gnu-libc},
4310 Do not use a @samp{--with} option to specify the file name to use to
4311 find certain files. That is outside the scope of what @samp{--with}
4314 @item @var{variable}=@var{value}
4315 Set the value of the variable @var{variable} to @var{value}. This is
4316 used to override the default values of commands or arguments in the
4317 build process. For example, the user could issue @samp{configure
4318 CFLAGS=-g CXXFLAGS=-g} to build with debugging information and without
4319 the default optimization.
4321 Specifying variables as arguments to @code{configure}, like this:
4325 is preferable to setting them in environment variables:
4329 as it helps to recreate the same configuration later with
4330 @file{config.status}. However, both methods should be supported.
4333 All @code{configure} scripts should accept all of the ``detail''
4334 options and the variable settings, whether or not they make any
4335 difference to the particular package at hand. In particular, they
4336 should accept any option that starts with @samp{--with-} or
4337 @samp{--enable-}. This is so users will be able to configure an
4338 entire GNU source tree at once with a single set of options.
4340 You will note that the categories @samp{--with-} and @samp{--enable-}
4341 are narrow: they @strong{do not} provide a place for any sort of option
4342 you might think of. That is deliberate. We want to limit the possible
4343 configuration options in GNU software. We do not want GNU programs to
4344 have idiosyncratic configuration options.
4346 Packages that perform part of the compilation process may support
4347 cross-compilation. In such a case, the host and target machines for the
4348 program may be different.
4350 The @code{configure} script should normally treat the specified type of
4351 system as both the host and the target, thus producing a program which
4352 works for the same type of machine that it runs on.
4354 To compile a program to run on a host type that differs from the build
4355 type, use the configure option @option{--host=@var{hosttype}}, where
4356 @var{hosttype} uses the same syntax as @var{buildtype}. The host type
4357 normally defaults to the build type.
4359 To configure a cross-compiler, cross-assembler, or what have you, you
4360 should specify a target different from the host, using the configure
4361 option @samp{--target=@var{targettype}}. The syntax for
4362 @var{targettype} is the same as for the host type. So the command would
4366 ./configure --host=@var{hosttype} --target=@var{targettype}
4369 The target type normally defaults to the host type.
4370 Programs for which cross-operation is not meaningful need not accept the
4371 @samp{--target} option, because configuring an entire operating system for
4372 cross-operation is not a meaningful operation.
4374 Some programs have ways of configuring themselves automatically. If
4375 your program is set up to do this, your @code{configure} script can simply
4376 ignore most of its arguments.
4378 @comment The makefile standards are in a separate file that is also
4379 @comment included by make.texinfo. Done by roland@gnu.ai.mit.edu on 1/6/93.
4380 @comment For this document, turn chapters into sections, etc.
4382 @include make-stds.texi
4386 @section Making Releases
4389 @cindex version numbers, for releases
4390 You should identify each release with a pair of version numbers, a
4391 major version and a minor. We have no objection to using more than
4392 two numbers, but it is very unlikely that you really need them.
4394 Package the distribution of @code{Foo version 69.96} up in a gzipped tar
4395 file with the name @file{foo-69.96.tar.gz}. It should unpack into a
4396 subdirectory named @file{foo-69.96}.
4398 Building and installing the program should never modify any of the files
4399 contained in the distribution. This means that all the files that form
4400 part of the program in any way must be classified into @dfn{source
4401 files} and @dfn{non-source files}. Source files are written by humans
4402 and never changed automatically; non-source files are produced from
4403 source files by programs under the control of the Makefile.
4405 @cindex @file{README} file
4406 The distribution should contain a file named @file{README} with a
4407 general overview of the package:
4410 @item the name of the package;
4412 @item the version number of the package, or refer to where in the
4413 package the version can be found;
4415 @item a general description of what the package does;
4417 @item a reference to the file @file{INSTALL}, which
4418 should in turn contain an explanation of the installation procedure;
4420 @item a brief explanation of any unusual top-level directories or
4421 files, or other hints for readers to find their way around the source;
4423 @item a reference to the file which contains the copying conditions.
4424 The GNU GPL, if used, should be in a file called @file{COPYING}. If
4425 the GNU LGPL is used, it should be in a file called
4426 @file{COPYING.LESSER}.
4429 Naturally, all the source files must be in the distribution. It is
4430 okay to include non-source files in the distribution along with the
4431 source files they are generated from, provided they are up-to-date
4432 with the source they are made from, and machine-independent, so that
4433 normal building of the distribution will never modify them. We
4434 commonly include non-source files produced by Autoconf, Automake,
4435 Bison, @code{flex}, @TeX{}, and @code{makeinfo}; this helps avoid
4436 unnecessary dependencies between our distributions, so that users can
4437 install whichever versions of whichever packages they like. Do not
4438 induce new dependencies on other software lightly.
4440 Non-source files that might actually be modified by building and
4441 installing the program should @strong{never} be included in the
4442 distribution. So if you do distribute non-source files, always make
4443 sure they are up to date when you make a new distribution.
4445 Make sure that all the files in the distribution are world-readable, and
4446 that directories are world-readable and world-searchable (octal mode 755).
4447 We used to recommend that all directories in the distribution also be
4448 world-writable (octal mode 777), because ancient versions of @code{tar}
4449 would otherwise not cope when extracting the archive as an unprivileged
4450 user. That can easily lead to security issues when creating the archive,
4451 however, so now we recommend against that.
4453 Don't include any symbolic links in the distribution itself. If the tar
4454 file contains symbolic links, then people cannot even unpack it on
4455 systems that don't support symbolic links. Also, don't use multiple
4456 names for one file in different directories, because certain file
4457 systems cannot handle this and that prevents unpacking the
4460 Try to make sure that all the file names will be unique on MS-DOS@. A
4461 name on MS-DOS consists of up to 8 characters, optionally followed by a
4462 period and up to three characters. MS-DOS will truncate extra
4463 characters both before and after the period. Thus,
4464 @file{foobarhacker.c} and @file{foobarhacker.o} are not ambiguous; they
4465 are truncated to @file{foobarha.c} and @file{foobarha.o}, which are
4468 @cindex @file{texinfo.tex}, in a distribution
4469 Include in your distribution a copy of the @file{texinfo.tex} you used
4470 to test print any @file{*.texinfo} or @file{*.texi} files.
4472 Likewise, if your program uses small GNU software packages like regex,
4473 getopt, obstack, or termcap, include them in the distribution file.
4474 Leaving them out would make the distribution file a little smaller at
4475 the expense of possible inconvenience to a user who doesn't know what
4479 @chapter References to Non-Free Software and Documentation
4480 @cindex references to non-free material
4482 A GNU program should not recommend, promote, or grant legitimacy to
4483 the use of any non-free program. Proprietary software is a social and
4484 ethical problem, and our aim is to put an end to that problem. We
4485 can't stop some people from writing proprietary programs, or stop
4486 other people from using them, but we can and should refuse to
4487 advertise them to new potential customers, or to give the public the
4488 impression that their existence is legitimate.
4490 The GNU definition of free software is found on the GNU web site at
4491 @url{https://www.gnu.org/@/philosophy/@/free-sw.html}, and the definition
4492 of free documentation is found at
4493 @url{https://www.gnu.org/@/philosophy/@/free-doc.html}. The terms ``free''
4494 and ``non-free'', used in this document, refer to those definitions.
4496 A list of important licenses and whether they qualify as free is in
4497 @url{https://www.gnu.org/@/licenses/@/license-list.html}. If it is not
4498 clear whether a license qualifies as free, please ask the GNU Project
4499 by writing to @email{licensing@@gnu.org}. We will answer, and if the
4500 license is an important one, we will add it to the list.
4502 When a non-free program or system is well known, you can mention it in
4503 passing---that is harmless, since users who might want to use it
4504 probably already know about it. For instance, it is fine to explain
4505 how to build your package on top of some widely used non-free
4506 operating system, or how to use it together with some widely used
4507 non-free program, after first explaining how to use it on the GNU
4510 However, you should give only the necessary information to help those
4511 who already use the non-free program to use your program with
4512 it---don't give, or refer to, any further information about the
4513 proprietary program, and don't imply that the proprietary program
4514 enhances your program, or that its existence is in any way a good
4515 thing. The goal should be that people already using the proprietary
4516 program will get the advice they need about how to use your free
4517 program with it, while people who don't already use the proprietary
4518 program will not see anything likely to lead them to take an interest
4521 You shouldn't recommend any non-free add-ons for the non-free program,
4522 but it is ok to mention free add-ons that help it to work with your
4523 program, and how to install the free add-ons even if that requires
4524 running some non-free program.
4526 If a non-free program or system is obscure in your program's domain,
4527 your program should not mention or support it at all, since doing so
4528 would tend to popularize the non-free program more than it popularizes
4529 your program. (You cannot hope to find many additional users for your
4530 program among the users of Foobar, if the existence of Foobar is not
4531 generally known among people who might want to use your program.)
4533 Sometimes a program is free software in itself but depends on a
4534 non-free platform in order to run. For instance, it used to be the
4535 case that many Java programs depended on some non-free Java libraries.
4536 (See @uref{https://www.gnu.org/philosophy/java-trap.html}.)
4537 To recommend or promote such a program is to promote the other
4538 programs it needs; therefore, judge mentions of the former as if they
4539 were mentions of the latter. For this reason, we were careful about
4540 listing Java programs in the Free Software Directory: we wanted to
4541 avoid promoting the non-free Java libraries.
4543 Java no longer has this problem, but the general principle will remain
4544 the same: don't recommend, promote or legitimize programs that depend
4545 on non-free software to run.
4547 Some free programs strongly encourage the use of non-free software. A
4548 typical example is @command{mplayer}. It is free software in itself,
4549 and the free code can handle some kinds of files. However,
4550 @command{mplayer} recommends use of non-free codecs for other kinds of
4551 files, and users that install @command{mplayer} are very likely to
4552 install those codecs along with it. To recommend @command{mplayer}
4553 is, in effect, to promote use of the non-free codecs.
4555 Thus, you should not recommend programs that strongly encourage the
4556 use of non-free software. This is why we do not list
4557 @command{mplayer} in the Free Software Directory.
4559 A GNU package should not refer the user to any non-free documentation
4560 for free software. Free documentation that can be included in free
4561 operating systems is essential for completing the GNU system, or any
4562 free operating system, so encouraging it is a priority; to recommend
4563 use of documentation that we are not allowed to include undermines the
4564 impetus for the community to produce documentation that we can
4565 include. So GNU packages should never recommend non-free
4568 By contrast, it is ok to refer to journal articles and textbooks in
4569 the comments of a program for explanation of how it functions, even
4570 though they are non-free. This is because we don't include such
4571 things in the GNU system even if they are free---they are outside the
4572 scope of what a software distribution needs to include.
4574 Referring to a web site that describes or recommends a non-free
4575 program is promoting that program, so please do not make links to (or
4576 mention by name) web sites that contain such material. This policy is
4577 relevant particularly for the web pages for a GNU package.
4579 What about chains of links? Following links from nearly any web site
4580 can lead eventually to promotion of non-free software; this is
4581 inherent in the nature of the web. Here's how we treat that.
4583 You should not refer to AT&T's web site if that recommends AT&T's
4584 non-free software packages; you should not refer to a page @var{p}
4585 that links to AT&T's site presenting it as a place to get some
4586 non-free program, because that part of the page @var{p} itself
4587 recommends and legitimizes the non-free program.
4589 However, if @var{p} contains a link to AT&T's web site for some other
4590 purpose (such as long-distance telephone service), that is no reason
4591 you should not link to @var{p}.
4593 A web page recommends a program in an implicit but particularly strong
4594 way if it requires users to run that program in order to use the page.
4595 Many pages contain Javascript code which they recommend in this way.
4596 This Javascript code may be free or nonfree, but nonfree is the usual
4599 If the purpose for which you would refer to the page cannot be carried
4600 out without running nonfree Javascript code, then you should not refer
4601 to it. Thus, if the purpose of referring to the page is for people to
4602 view a video, or subscribing to a mailing list, and the viewing or
4603 subscribing fail to work if the user's browser blocks the nonfree
4604 Javascript code, then don't refer to that page.
4606 The extreme case is that of web sites which depend on nonfree
4607 Javascript code even to @emph{see} the contents of the pages. Any
4608 site hosted on @indicateurl{wix.com} has this problem, and so do some
4609 other sites. Referring people to such pages to read their contents
4610 is, in effect, urging them to run those nonfree programs---so please
4611 don't refer to those pages. (Such pages also break the Web, so they
4612 deserve condemnation for two reasons.)
4614 Instead, please quote excerpts from the page to make your point,
4615 or find another place to refer to that information.
4617 @node GNU Free Documentation License
4618 @appendix GNU Free Documentation License
4620 @cindex FDL, GNU Free Documentation License
4630 eval: (add-hook 'before-save-hook 'time-stamp)
4631 time-stamp-start: "@set lastupdate "
4633 time-stamp-format: "%:b %:d, %:y"
4634 compile-command: "cd work.s && make"