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 January 27, 2011
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, 2011
31 Free Software Foundation, Inc.
33 Permission is granted to copy, distribute and/or modify this document
34 under the terms of the GNU Free Documentation License, Version 1.3 or
35 any later version published by the Free Software Foundation; with no
36 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
37 Texts. A copy of the license is included in the section entitled
38 ``GNU Free Documentation License''.
42 @title GNU Coding Standards
43 @author Richard Stallman, et al.
44 @author last updated @value{lastupdate}
46 @vskip 0pt plus 1filll
53 @node Top, Preface, (dir), (dir)
60 * Preface:: About the GNU Coding Standards.
61 * Legal Issues:: Keeping free software free.
62 * Design Advice:: General program design.
63 * Program Behavior:: Program behavior for all programs
64 * Writing C:: Making the best use of C.
65 * Documentation:: Documenting programs.
66 * Managing Releases:: The release process.
67 * References:: Mentioning non-free software or documentation.
68 * GNU Free Documentation License:: Copying and sharing this manual.
74 @chapter About the GNU Coding Standards
76 The GNU Coding Standards were written by Richard Stallman and other GNU
77 Project volunteers. Their purpose is to make the GNU system clean,
78 consistent, and easy to install. This document can also be read as a
79 guide to writing portable, robust and reliable programs. It focuses on
80 programs written in C, but many of the rules and principles are useful
81 even if you write in another programming language. The rules often
82 state reasons for writing in a certain way.
84 @cindex where to obtain @code{standards.texi}
85 @cindex downloading this manual
86 If you did not obtain this file directly from the GNU project and
87 recently, please check for a newer version. You can get the GNU
88 Coding Standards from the GNU web server in many
89 different formats, including the Texinfo source, PDF, HTML, DVI, plain
90 text, and more, at: @uref{http://www.gnu.org/prep/standards/}.
92 If you are maintaining an official GNU package, in addition to this
93 document, please read and follow the GNU maintainer information
94 (@pxref{Top, , Contents, maintain, Information for Maintainers of GNU
97 @cindex @code{gnustandards-commit@@gnu.org} mailing list
98 If you want to receive diffs for every change to these GNU documents,
99 join the mailing list @code{gnustandards-commit@@gnu.org}, via the web
101 @url{http://lists.gnu.org/mailman/listinfo/gnustandards-commit}.
102 Archives are also available there.
104 @cindex @code{bug-standards@@gnu.org} email address
105 @cindex Savannah repository for gnustandards
106 @cindex gnustandards project repository
107 Please send corrections or suggestions for this document to
108 @email{bug-standards@@gnu.org}. If you make a suggestion, please
109 include a suggested new wording for it, to help us consider the
110 suggestion efficiently. We prefer a context diff to the Texinfo
111 source, but if that's difficult for you, you can make a context diff
112 for some other version of this document, or propose it in any way that
113 makes it clear. The source repository for this document can be found
114 at @url{http://savannah.gnu.org/projects/gnustandards}.
116 These standards cover the minimum of what is important when writing a
117 GNU package. Likely, the need for additional standards will come up.
118 Sometimes, you might suggest that such standards be added to this
119 document. If you think your standards would be generally useful, please
122 You should also set standards for your package on many questions not
123 addressed or not firmly specified here. The most important point is to
124 be self-consistent---try to stick to the conventions you pick, and try
125 to document them as much as possible. That way, your program will be
126 more maintainable by others.
128 The GNU Hello program serves as an example of how to follow the GNU
129 coding standards for a trivial program.
130 @uref{http://www.gnu.org/software/hello/hello.html}.
132 This release of the GNU Coding Standards was last updated
137 @chapter Keeping Free Software Free
138 @cindex legal aspects
140 This chapter discusses how you can make sure that GNU software
141 avoids legal difficulties, and other related issues.
144 * Reading Non-Free Code:: Referring to proprietary programs.
145 * Contributions:: Accepting contributions.
146 * Trademarks:: How we deal with trademark issues.
149 @node Reading Non-Free Code
150 @section Referring to Proprietary Programs
151 @cindex proprietary programs
152 @cindex avoiding proprietary code
154 Don't in any circumstances refer to Unix source code for or during
155 your work on GNU! (Or to any other proprietary programs.)
157 If you have a vague recollection of the internals of a Unix program,
158 this does not absolutely mean you can't write an imitation of it, but
159 do try to organize the imitation internally along different lines,
160 because this is likely to make the details of the Unix version
161 irrelevant and dissimilar to your results.
163 For example, Unix utilities were generally optimized to minimize
164 memory use; if you go for speed instead, your program will be very
165 different. You could keep the entire input file in memory and scan it
166 there instead of using stdio. Use a smarter algorithm discovered more
167 recently than the Unix program. Eliminate use of temporary files. Do
168 it in one pass instead of two (we did this in the assembler).
170 Or, on the contrary, emphasize simplicity instead of speed. For some
171 applications, the speed of today's computers makes simpler algorithms
174 Or go for generality. For example, Unix programs often have static
175 tables or fixed-size strings, which make for arbitrary limits; use
176 dynamic allocation instead. Make sure your program handles NULs and
177 other funny characters in the input files. Add a programming language
178 for extensibility and write part of the program in that language.
180 Or turn some parts of the program into independently usable libraries.
181 Or use a simple garbage collector instead of tracking precisely when
182 to free memory, or use a new GNU facility such as obstacks.
185 @section Accepting Contributions
187 @cindex accepting contributions
189 If the program you are working on is copyrighted by the Free Software
190 Foundation, then when someone else sends you a piece of code to add to
191 the program, we need legal papers to use it---just as we asked you to
192 sign papers initially. @emph{Each} person who makes a nontrivial
193 contribution to a program must sign some sort of legal papers in order
194 for us to have clear title to the program; the main author alone is not
197 So, before adding in any contributions from other people, please tell
198 us, so we can arrange to get the papers. Then wait until we tell you
199 that we have received the signed papers, before you actually use the
202 This applies both before you release the program and afterward. If
203 you receive diffs to fix a bug, and they make significant changes, we
204 need legal papers for that change.
206 This also applies to comments and documentation files. For copyright
207 law, comments and code are just text. Copyright applies to all kinds of
208 text, so we need legal papers for all kinds.
210 We know it is frustrating to ask for legal papers; it's frustrating for
211 us as well. But if you don't wait, you are going out on a limb---for
212 example, what if the contributor's employer won't sign a disclaimer?
213 You might have to take that code out again!
215 You don't need papers for changes of a few lines here or there, since
216 they are not significant for copyright purposes. Also, you don't need
217 papers if all you get from the suggestion is some ideas, not actual code
218 which you use. For example, if someone sent you one implementation, but
219 you write a different implementation of the same idea, you don't need to
222 The very worst thing is if you forget to tell us about the other
223 contributor. We could be very embarrassed in court some day as a
226 We have more detailed advice for maintainers of programs; if you have
227 reached the stage of actually maintaining a program for GNU (whether
228 released or not), please ask us for a copy. It is also available
229 online for your perusal: @uref{http://www.gnu.org/prep/maintain/}.
235 Please do not include any trademark acknowledgements in GNU software
236 packages or documentation.
238 Trademark acknowledgements are the statements that such-and-such is a
239 trademark of so-and-so. The GNU Project has no objection to the basic
240 idea of trademarks, but these acknowledgements feel like kowtowing,
241 and there is no legal requirement for them, so we don't use them.
243 What is legally required, as regards other people's trademarks, is to
244 avoid using them in ways which a reader might reasonably understand as
245 naming or labeling our own programs or activities. For example, since
246 ``Objective C'' is (or at least was) a trademark, we made sure to say
247 that we provide a ``compiler for the Objective C language'' rather
248 than an ``Objective C compiler''. The latter would have been meant as
249 a shorter way of saying the former, but it does not explicitly state
250 the relationship, so it could be misinterpreted as using ``Objective
251 C'' as a label for the compiler rather than for the language.
253 Please don't use ``win'' as an abbreviation for Microsoft Windows in
254 GNU software or documentation. In hacker terminology, calling
255 something a ``win'' is a form of praise. If you wish to praise
256 Microsoft Windows when speaking on your own, by all means do so, but
257 not in GNU software. Usually we write the name ``Windows'' in full,
258 but when brevity is very important (as in file names and sometimes
259 symbol names), we abbreviate it to ``w''. For instance, the files and
260 functions in Emacs that deal with Windows start with @samp{w32}.
263 @chapter General Program Design
264 @cindex program design
266 This chapter discusses some of the issues you should take into
267 account when designing your program.
269 @c Standard or ANSI C
271 @c In 1989 the American National Standards Institute (ANSI) standardized
272 @c C as standard X3.159-1989. In December of that year the
273 @c International Standards Organization ISO adopted the ANSI C standard
274 @c making minor changes. In 1990 ANSI then re-adopted ISO standard
275 @c C. This version of C is known as either ANSI C or Standard C.
277 @c A major revision of the C Standard appeared in 1999.
280 * Source Language:: Which languages to use.
281 * Compatibility:: Compatibility with other implementations.
282 * Using Extensions:: Using non-standard features.
283 * Standard C:: Using standard C features.
284 * Conditional Compilation:: Compiling code only if a conditional is true.
287 @node Source Language
288 @section Which Languages to Use
289 @cindex programming languages
291 When you want to use a language that gets compiled and runs at high
292 speed, the best language to use is C. Using another language is like
293 using a non-standard feature: it will cause trouble for users. Even if
294 GCC supports the other language, users may find it inconvenient to have
295 to install the compiler for that other language in order to build your
296 program. For example, if you write your program in C++, people will
297 have to install the GNU C++ compiler in order to compile your program.
299 C has one other advantage over C++ and other compiled languages: more
300 people know C, so more people will find it easy to read and modify the
301 program if it is written in C.
303 So in general it is much better to use C, rather than the
304 comparable alternatives.
306 But there are two exceptions to that conclusion:
310 It is no problem to use another language to write a tool specifically
311 intended for use with that language. That is because the only people
312 who want to build the tool will be those who have installed the other
316 If an application is of interest only to a narrow part of the community,
317 then the question of which language it is written in has less effect on
318 other people, so you may as well please yourself.
321 Many programs are designed to be extensible: they include an interpreter
322 for a language that is higher level than C. Often much of the program
323 is written in that language, too. The Emacs editor pioneered this
327 @cindex GNOME and Guile
328 The standard extensibility interpreter for GNU software is Guile
329 (@uref{http://www.gnu.org/@/software/@/guile/}), which implements the
330 language Scheme (an especially clean and simple dialect of Lisp).
331 Guile also includes bindings for GTK+/GNOME, making it practical to
332 write modern GUI functionality within Guile. We don't reject programs
333 written in other ``scripting languages'' such as Perl and Python, but
334 using Guile is very important for the overall consistency of the GNU
339 @section Compatibility with Other Implementations
340 @cindex compatibility with C and @sc{posix} standards
341 @cindex @sc{posix} compatibility
343 With occasional exceptions, utility programs and libraries for GNU
344 should be upward compatible with those in Berkeley Unix, and upward
345 compatible with Standard C if Standard C specifies their
346 behavior, and upward compatible with @sc{posix} if @sc{posix} specifies
349 When these standards conflict, it is useful to offer compatibility
350 modes for each of them.
352 @cindex options for compatibility
353 Standard C and @sc{posix} prohibit many kinds of extensions. Feel
354 free to make the extensions anyway, and include a @samp{--ansi},
355 @samp{--posix}, or @samp{--compatible} option to turn them off.
356 However, if the extension has a significant chance of breaking any real
357 programs or scripts, then it is not really upward compatible. So you
358 should try to redesign its interface to make it upward compatible.
360 @cindex @code{POSIXLY_CORRECT}, environment variable
361 Many GNU programs suppress extensions that conflict with @sc{posix} if the
362 environment variable @code{POSIXLY_CORRECT} is defined (even if it is
363 defined with a null value). Please make your program recognize this
364 variable if appropriate.
366 When a feature is used only by users (not by programs or command
367 files), and it is done poorly in Unix, feel free to replace it
368 completely with something totally different and better. (For example,
369 @code{vi} is replaced with Emacs.) But it is nice to offer a compatible
370 feature as well. (There is a free @code{vi} clone, so we offer it.)
372 Additional useful features are welcome regardless of whether
373 there is any precedent for them.
375 @node Using Extensions
376 @section Using Non-standard Features
377 @cindex non-standard extensions
379 Many GNU facilities that already exist support a number of convenient
380 extensions over the comparable Unix facilities. Whether to use these
381 extensions in implementing your program is a difficult question.
383 On the one hand, using the extensions can make a cleaner program.
384 On the other hand, people will not be able to build the program
385 unless the other GNU tools are available. This might cause the
386 program to work on fewer kinds of machines.
388 With some extensions, it might be easy to provide both alternatives.
389 For example, you can define functions with a ``keyword'' @code{INLINE}
390 and define that as a macro to expand into either @code{inline} or
391 nothing, depending on the compiler.
393 In general, perhaps it is best not to use the extensions if you can
394 straightforwardly do without them, but to use the extensions if they
395 are a big improvement.
397 An exception to this rule are the large, established programs (such as
398 Emacs) which run on a great variety of systems. Using GNU extensions in
399 such programs would make many users unhappy, so we don't do that.
401 Another exception is for programs that are used as part of compilation:
402 anything that must be compiled with other compilers in order to
403 bootstrap the GNU compilation facilities. If these require the GNU
404 compiler, then no one can compile them without having them installed
405 already. That would be extremely troublesome in certain cases.
408 @section Standard C and Pre-Standard C
409 @cindex @sc{ansi} C standard
411 1989 Standard C is widespread enough now that it is ok to use its
412 features in new programs. There is one exception: do not ever use the
413 ``trigraph'' feature of Standard C.
415 1999 Standard C is not widespread yet, so please do not require its
416 features in programs. It is ok to use its features if they are present.
418 However, it is easy to support pre-standard compilers in most programs,
419 so if you know how to do that, feel free. If a program you are
420 maintaining has such support, you should try to keep it working.
422 @cindex function prototypes
423 To support pre-standard C, instead of writing function definitions in
424 standard prototype form,
433 write the definition in pre-standard style like this,
443 and use a separate declaration to specify the argument prototype:
449 You need such a declaration anyway, in a header file, to get the benefit
450 of prototypes in all the files where the function is called. And once
451 you have the declaration, you normally lose nothing by writing the
452 function definition in the pre-standard style.
454 This technique does not work for integer types narrower than @code{int}.
455 If you think of an argument as being of a type narrower than @code{int},
456 declare it as @code{int} instead.
458 There are a few special cases where this technique is hard to use. For
459 example, if a function argument needs to hold the system type
460 @code{dev_t}, you run into trouble, because @code{dev_t} is shorter than
461 @code{int} on some machines; but you cannot use @code{int} instead,
462 because @code{dev_t} is wider than @code{int} on some machines. There
463 is no type you can safely use on all machines in a non-standard
464 definition. The only way to support non-standard C and pass such an
465 argument is to check the width of @code{dev_t} using Autoconf and choose
466 the argument type accordingly. This may not be worth the trouble.
468 In order to support pre-standard compilers that do not recognize
469 prototypes, you may want to use a preprocessor macro like this:
472 /* Declare the prototype for a general external function. */
473 #if defined (__STDC__) || defined (WINDOWSNT)
474 #define P_(proto) proto
480 @node Conditional Compilation
481 @section Conditional Compilation
483 When supporting configuration options already known when building your
484 program we prefer using @code{if (... )} over conditional compilation,
485 as in the former case the compiler is able to perform more extensive
486 checking of all possible code paths.
488 For example, please write
508 A modern compiler such as GCC will generate exactly the same code in
509 both cases, and we have been using similar techniques with good success
510 in several projects. Of course, the former method assumes that
511 @code{HAS_FOO} is defined as either 0 or 1.
513 While this is not a silver bullet solving all portability problems,
514 and is not always appropriate, following this policy would have saved
515 GCC developers many hours, or even days, per year.
517 In the case of function-like macros like @code{REVERSIBLE_CC_MODE} in
518 GCC which cannot be simply used in @code{if (...)} statements, there is
519 an easy workaround. Simply introduce another macro
520 @code{HAS_REVERSIBLE_CC_MODE} as in the following example:
523 #ifdef REVERSIBLE_CC_MODE
524 #define HAS_REVERSIBLE_CC_MODE 1
526 #define HAS_REVERSIBLE_CC_MODE 0
530 @node Program Behavior
531 @chapter Program Behavior for All Programs
533 This chapter describes conventions for writing robust
534 software. It also describes general standards for error messages, the
535 command line interface, and how libraries should behave.
538 * Non-GNU Standards:: We consider standards such as POSIX;
539 we don't "obey" them.
540 * Semantics:: Writing robust programs.
541 * Libraries:: Library behavior.
542 * Errors:: Formatting error messages.
543 * User Interfaces:: Standards about interfaces generally.
544 * Graphical Interfaces:: Standards for graphical interfaces.
545 * Command-Line Interfaces:: Standards for command line interfaces.
546 * Dynamic Plug-In Interfaces:: Standards for dynamic plug-in interfaces.
547 * Option Table:: Table of long options.
548 * OID Allocations:: Table of OID slots for GNU.
549 * Memory Usage:: When and how to care about memory needs.
550 * File Usage:: Which files to use, and where.
553 @node Non-GNU Standards
554 @section Non-GNU Standards
556 The GNU Project regards standards published by other organizations as
557 suggestions, not orders. We consider those standards, but we do not
558 ``obey'' them. In developing a GNU program, you should implement
559 an outside standard's specifications when that makes the GNU system
560 better overall in an objective sense. When it doesn't, you shouldn't.
562 In most cases, following published standards is convenient for
563 users---it means that their programs or scripts will work more
564 portably. For instance, GCC implements nearly all the features of
565 Standard C as specified by that standard. C program developers would
566 be unhappy if it did not. And GNU utilities mostly follow
567 specifications of POSIX.2; shell script writers and users would be
568 unhappy if our programs were incompatible.
570 But we do not follow either of these specifications rigidly, and there
571 are specific points on which we decided not to follow them, so as to
572 make the GNU system better for users.
574 For instance, Standard C says that nearly all extensions to C are
575 prohibited. How silly! GCC implements many extensions, some of which
576 were later adopted as part of the standard. If you want these
577 constructs to give an error message as ``required'' by the standard,
578 you must specify @samp{--pedantic}, which was implemented only so that
579 we can say ``GCC is a 100% implementation of the standard,'' not
580 because there is any reason to actually use it.
582 POSIX.2 specifies that @samp{df} and @samp{du} must output sizes by
583 default in units of 512 bytes. What users want is units of 1k, so
584 that is what we do by default. If you want the ridiculous behavior
585 ``required'' by POSIX, you must set the environment variable
586 @samp{POSIXLY_CORRECT} (which was originally going to be named
587 @samp{POSIX_ME_HARDER}).
589 GNU utilities also depart from the letter of the POSIX.2 specification
590 when they support long-named command-line options, and intermixing
591 options with ordinary arguments. This minor incompatibility with
592 POSIX is never a problem in practice, and it is very useful.
594 In particular, don't reject a new feature, or remove an old one,
595 merely because a standard says it is ``forbidden'' or ``deprecated.''
598 @section Writing Robust Programs
600 @cindex arbitrary limits on data
601 Avoid arbitrary limits on the length or number of @emph{any} data
602 structure, including file names, lines, files, and symbols, by allocating
603 all data structures dynamically. In most Unix utilities, ``long lines
604 are silently truncated''. This is not acceptable in a GNU utility.
606 @cindex @code{NUL} characters
607 Utilities reading files should not drop NUL characters, or any other
608 nonprinting characters @emph{including those with codes above 0177}.
609 The only sensible exceptions would be utilities specifically intended
610 for interface to certain types of terminals or printers
611 that can't handle those characters.
612 Whenever possible, try to make programs work properly with
613 sequences of bytes that represent multibyte characters, using encodings
614 such as UTF-8 and others.
616 @cindex error messages
617 Check every system call for an error return, unless you know you wish to
618 ignore errors. Include the system error text (from @code{perror} or
619 equivalent) in @emph{every} error message resulting from a failing
620 system call, as well as the name of the file if any and the name of the
621 utility. Just ``cannot open foo.c'' or ``stat failed'' is not
624 @cindex @code{malloc} return value
625 @cindex memory allocation failure
626 Check every call to @code{malloc} or @code{realloc} to see if it
627 returned zero. Check @code{realloc} even if you are making the block
628 smaller; in a system that rounds block sizes to a power of 2,
629 @code{realloc} may get a different block if you ask for less space.
631 In Unix, @code{realloc} can destroy the storage block if it returns
632 zero. GNU @code{realloc} does not have this bug: if it fails, the
633 original block is unchanged. Feel free to assume the bug is fixed. If
634 you wish to run your program on Unix, and wish to avoid lossage in this
635 case, you can use the GNU @code{malloc}.
637 You must expect @code{free} to alter the contents of the block that was
638 freed. Anything you want to fetch from the block, you must fetch before
641 If @code{malloc} fails in a noninteractive program, make that a fatal
642 error. In an interactive program (one that reads commands from the
643 user), it is better to abort the command and return to the command
644 reader loop. This allows the user to kill other processes to free up
645 virtual memory, and then try the command again.
647 @cindex command-line arguments, decoding
648 Use @code{getopt_long} to decode arguments, unless the argument syntax
649 makes this unreasonable.
651 When static storage is to be written in during program execution, use
652 explicit C code to initialize it. Reserve C initialized declarations
653 for data that will not be changed.
656 Try to avoid low-level interfaces to obscure Unix data structures (such
657 as file directories, utmp, or the layout of kernel memory), since these
658 are less likely to work compatibly. If you need to find all the files
659 in a directory, use @code{readdir} or some other high-level interface.
660 These are supported compatibly by GNU.
662 @cindex signal handling
663 The preferred signal handling facilities are the BSD variant of
664 @code{signal}, and the @sc{posix} @code{sigaction} function; the
665 alternative USG @code{signal} interface is an inferior design.
667 Nowadays, using the @sc{posix} signal functions may be the easiest way
668 to make a program portable. If you use @code{signal}, then on GNU/Linux
669 systems running GNU libc version 1, you should include
670 @file{bsd/signal.h} instead of @file{signal.h}, so as to get BSD
671 behavior. It is up to you whether to support systems where
672 @code{signal} has only the USG behavior, or give up on them.
674 @cindex impossible conditions
675 In error checks that detect ``impossible'' conditions, just abort.
676 There is usually no point in printing any message. These checks
677 indicate the existence of bugs. Whoever wants to fix the bugs will have
678 to read the source code and run a debugger. So explain the problem with
679 comments in the source. The relevant data will be in variables, which
680 are easy to examine with the debugger, so there is no point moving them
683 Do not use a count of errors as the exit status for a program.
684 @emph{That does not work}, because exit status values are limited to 8
685 bits (0 through 255). A single run of the program might have 256
686 errors; if you try to return 256 as the exit status, the parent process
687 will see 0 as the status, and it will appear that the program succeeded.
689 @cindex temporary files
690 @cindex @code{TMPDIR} environment variable
691 If you make temporary files, check the @code{TMPDIR} environment
692 variable; if that variable is defined, use the specified directory
693 instead of @file{/tmp}.
695 In addition, be aware that there is a possible security problem when
696 creating temporary files in world-writable directories. In C, you can
697 avoid this problem by creating temporary files in this manner:
700 fd = open (filename, O_WRONLY | O_CREAT | O_EXCL, 0600);
704 or by using the @code{mkstemps} function from libiberty.
706 In bash, use @code{set -C} (long name @code{noclobber}) to avoid this
707 problem. In addition, the @code{mktemp} utility is a more general
708 solution for creating temporary files from shell scripts
709 (@pxref{mktemp invocation,,, coreutils, GNU Coreutils}).
713 @section Library Behavior
716 Try to make library functions reentrant. If they need to do dynamic
717 storage allocation, at least try to avoid any nonreentrancy aside from
718 that of @code{malloc} itself.
720 Here are certain name conventions for libraries, to avoid name
723 Choose a name prefix for the library, more than two characters long.
724 All external function and variable names should start with this
725 prefix. In addition, there should only be one of these in any given
726 library member. This usually means putting each one in a separate
729 An exception can be made when two external symbols are always used
730 together, so that no reasonable program could use one without the
731 other; then they can both go in the same file.
733 External symbols that are not documented entry points for the user
734 should have names beginning with @samp{_}. The @samp{_} should be
735 followed by the chosen name prefix for the library, to prevent
736 collisions with other libraries. These can go in the same files with
737 user entry points if you like.
739 Static functions and variables can be used as you like and need not
740 fit any naming convention.
743 @section Formatting Error Messages
744 @cindex formatting error messages
745 @cindex error messages, formatting
747 Error messages from compilers should look like this:
750 @var{source-file-name}:@var{lineno}: @var{message}
754 If you want to mention the column number, use one of these formats:
757 @var{source-file-name}:@var{lineno}:@var{column}: @var{message}
758 @var{source-file-name}:@var{lineno}.@var{column}: @var{message}
763 Line numbers should start from 1 at the beginning of the file, and
764 column numbers should start from 1 at the beginning of the line. (Both
765 of these conventions are chosen for compatibility.) Calculate column
766 numbers assuming that space and all ASCII printing characters have
767 equal width, and assuming tab stops every 8 columns.
769 The error message can also give both the starting and ending positions
770 of the erroneous text. There are several formats so that you can
771 avoid redundant information such as a duplicate line number.
772 Here are the possible formats:
775 @var{source-file-name}:@var{lineno-1}.@var{column-1}-@var{lineno-2}.@var{column-2}: @var{message}
776 @var{source-file-name}:@var{lineno-1}.@var{column-1}-@var{column-2}: @var{message}
777 @var{source-file-name}:@var{lineno-1}-@var{lineno-2}: @var{message}
781 When an error is spread over several files, you can use this format:
784 @var{file-1}:@var{lineno-1}.@var{column-1}-@var{file-2}:@var{lineno-2}.@var{column-2}: @var{message}
787 Error messages from other noninteractive programs should look like this:
790 @var{program}:@var{source-file-name}:@var{lineno}: @var{message}
794 when there is an appropriate source file, or like this:
797 @var{program}: @var{message}
801 when there is no relevant source file.
803 If you want to mention the column number, use this format:
806 @var{program}:@var{source-file-name}:@var{lineno}:@var{column}: @var{message}
809 In an interactive program (one that is reading commands from a
810 terminal), it is better not to include the program name in an error
811 message. The place to indicate which program is running is in the
812 prompt or with the screen layout. (When the same program runs with
813 input from a source other than a terminal, it is not interactive and
814 would do best to print error messages using the noninteractive style.)
816 The string @var{message} should not begin with a capital letter when
817 it follows a program name and/or file name, because that isn't the
818 beginning of a sentence. (The sentence conceptually starts at the
819 beginning of the line.) Also, it should not end with a period.
821 Error messages from interactive programs, and other messages such as
822 usage messages, should start with a capital letter. But they should not
825 @node User Interfaces
826 @section Standards for Interfaces Generally
828 @cindex program name and its behavior
829 @cindex behavior, dependent on program's name
830 Please don't make the behavior of a utility depend on the name used
831 to invoke it. It is useful sometimes to make a link to a utility
832 with a different name, and that should not change what it does.
834 Instead, use a run time option or a compilation switch or both
835 to select among the alternate behaviors.
837 @cindex output device and program's behavior
838 Likewise, please don't make the behavior of the program depend on the
839 type of output device it is used with. Device independence is an
840 important principle of the system's design; do not compromise it merely
841 to save someone from typing an option now and then. (Variation in error
842 message syntax when using a terminal is ok, because that is a side issue
843 that people do not depend on.)
845 If you think one behavior is most useful when the output is to a
846 terminal, and another is most useful when the output is a file or a
847 pipe, then it is usually best to make the default behavior the one that
848 is useful with output to a terminal, and have an option for the other
851 Compatibility requires certain programs to depend on the type of output
852 device. It would be disastrous if @code{ls} or @code{sh} did not do so
853 in the way all users expect. In some of these cases, we supplement the
854 program with a preferred alternate version that does not depend on the
855 output device type. For example, we provide a @code{dir} program much
856 like @code{ls} except that its default output format is always
860 @node Graphical Interfaces
861 @section Standards for Graphical Interfaces
862 @cindex graphical user interface
863 @cindex interface styles
864 @cindex user interface styles
867 When you write a program that provides a graphical user interface,
868 please make it work with the X Window System and the GTK+ toolkit
869 unless the functionality specifically requires some alternative (for
870 example, ``displaying jpeg images while in console mode'').
872 In addition, please provide a command-line interface to control the
873 functionality. (In many cases, the graphical user interface can be a
874 separate program which invokes the command-line program.) This is
875 so that the same jobs can be done from scripts.
880 @cindex keyboard interface
881 @cindex library interface
882 Please also consider providing a D-bus interface for use from other
883 running programs, such as within GNOME. (GNOME used to use CORBA
884 for this, but that is being phased out.) In addition, consider
885 providing a library interface (for use from C), and perhaps a
886 keyboard-driven console interface (for use by users from console
887 mode). Once you are doing the work to provide the functionality and
888 the graphical interface, these won't be much extra work.
891 @node Command-Line Interfaces
892 @section Standards for Command Line Interfaces
893 @cindex command-line interface
896 It is a good idea to follow the @sc{posix} guidelines for the
897 command-line options of a program. The easiest way to do this is to use
898 @code{getopt} to parse them. Note that the GNU version of @code{getopt}
899 will normally permit options anywhere among the arguments unless the
900 special argument @samp{--} is used. This is not what @sc{posix}
901 specifies; it is a GNU extension.
903 @cindex long-named options
904 Please define long-named options that are equivalent to the
905 single-letter Unix-style options. We hope to make GNU more user
906 friendly this way. This is easy to do with the GNU function
909 One of the advantages of long-named options is that they can be
910 consistent from program to program. For example, users should be able
911 to expect the ``verbose'' option of any GNU program which has one, to be
912 spelled precisely @samp{--verbose}. To achieve this uniformity, look at
913 the table of common long-option names when you choose the option names
914 for your program (@pxref{Option Table}).
916 It is usually a good idea for file names given as ordinary arguments to
917 be input files only; any output files would be specified using options
918 (preferably @samp{-o} or @samp{--output}). Even if you allow an output
919 file name as an ordinary argument for compatibility, try to provide an
920 option as another way to specify it. This will lead to more consistency
921 among GNU utilities, and fewer idiosyncrasies for users to remember.
923 @cindex standard command-line options
924 @cindex options, standard command-line
925 @cindex CGI programs, standard options for
926 @cindex PATH_INFO, specifying standard options as
927 All programs should support two standard options: @samp{--version}
928 and @samp{--help}. CGI programs should accept these as command-line
929 options, and also if given as the @env{PATH_INFO}; for instance,
930 visiting @url{http://example.org/p.cgi/--help} in a browser should
931 output the same information as invoking @samp{p.cgi --help} from the
935 * --version:: The standard output for --version.
936 * --help:: The standard output for --help.
940 @subsection @option{--version}
942 @cindex @samp{--version} output
944 The standard @code{--version} option should direct the program to
945 print information about its name, version, origin and legal status,
946 all on standard output, and then exit successfully. Other options and
947 arguments should be ignored once this is seen, and the program should
948 not perform its normal function.
950 @cindex canonical name of a program
951 @cindex program's canonical name
952 The first line is meant to be easy for a program to parse; the version
953 number proper starts after the last space. In addition, it contains
954 the canonical name for this program, in this format:
961 The program's name should be a constant string; @emph{don't} compute it
962 from @code{argv[0]}. The idea is to state the standard or canonical
963 name for the program, not its file name. There are other ways to find
964 out the precise file name where a command is found in @code{PATH}.
966 If the program is a subsidiary part of a larger package, mention the
967 package name in parentheses, like this:
970 emacsserver (GNU Emacs) 19.30
974 If the package has a version number which is different from this
975 program's version number, you can mention the package version number
976 just before the close-parenthesis.
978 If you @emph{need} to mention the version numbers of libraries which
979 are distributed separately from the package which contains this program,
980 you can do so by printing an additional line of version info for each
981 library you want to mention. Use the same format for these lines as for
984 Please do not mention all of the libraries that the program uses ``just
985 for completeness''---that would produce a lot of unhelpful clutter.
986 Please mention library version numbers only if you find in practice that
987 they are very important to you in debugging.
989 The following line, after the version number line or lines, should be a
990 copyright notice. If more than one copyright notice is called for, put
991 each on a separate line.
993 Next should follow a line stating the license, preferably using one of
994 abbrevations below, and a brief statement that the program is free
995 software, and that users are free to copy and change it. Also mention
996 that there is no warranty, to the extent permitted by law. See
997 recommended wording below.
999 It is ok to finish the output with a list of the major authors of the
1000 program, as a way of giving credit.
1002 Here's an example of output that follows these rules:
1006 Copyright (C) 2007 Free Software Foundation, Inc.
1007 License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
1008 This is free software: you are free to change and redistribute it.
1009 There is NO WARRANTY, to the extent permitted by law.
1012 You should adapt this to your program, of course, filling in the proper
1013 year, copyright holder, name of program, and the references to
1014 distribution terms, and changing the rest of the wording as necessary.
1016 This copyright notice only needs to mention the most recent year in
1017 which changes were made---there's no need to list the years for previous
1018 versions' changes. You don't have to mention the name of the program in
1019 these notices, if that is inconvenient, since it appeared in the first
1020 line. (The rules are different for copyright notices in source files;
1021 @pxref{Copyright Notices,,,maintain,Information for GNU Maintainers}.)
1023 Translations of the above lines must preserve the validity of the
1024 copyright notices (@pxref{Internationalization}). If the translation's
1025 character set supports it, the @samp{(C)} should be replaced with the
1026 copyright symbol, as follows:
1029 (the official copyright symbol, which is the letter C in a circle);
1035 Write the word ``Copyright'' exactly like that, in English. Do not
1036 translate it into another language. International treaties recognize
1037 the English word ``Copyright''; translations into other languages do not
1038 have legal significance.
1040 Finally, here is the table of our suggested license abbreviations.
1041 Any abbreviation can be followed by @samp{v@var{version}[+]}, meaning
1042 that particular version, or later versions with the @samp{+}, as shown
1045 In the case of exceptions for extra permissions with the GPL, we use
1046 @samp{/} for a separator; the version number can follow the license
1047 abbreviation as usual, as in the examples below.
1051 GNU General Public License, @url{http://www.gnu.org/@/licenses/@/gpl.html}.
1054 GNU Lesser General Public License, @url{http://www.gnu.org/@/licenses/@/lgpl.html}.
1057 GNU GPL with the exception for Ada.
1060 The Apache Software Foundation license,
1061 @url{http://www.apache.org/@/licenses}.
1064 The Artistic license used for Perl, @url{http://www.perlfoundation.org/@/legal}.
1067 The Expat license, @url{http://www.jclark.com/@/xml/@/copying.txt}.
1070 The Mozilla Public License, @url{http://www.mozilla.org/@/MPL/}.
1073 The original (4-clause) BSD license, incompatible with the GNU GPL
1074 @url{http://www.xfree86.org/@/3.3.6/@/COPYRIGHT2.html#6}.
1077 The license used for PHP, @url{http://www.php.net/@/license/}.
1080 The non-license that is being in the public domain,
1081 @url{http://www.gnu.org/@/licenses/@/license-list.html#PublicDomain}.
1084 The license for Python, @url{http://www.python.org/@/2.0.1/@/license.html}.
1087 The revised (3-clause) BSD, compatible with the GNU GPL,@*
1088 @url{http://www.xfree86.org/@/3.3.6/@/COPYRIGHT2.html#5}.
1091 The simple non-copyleft license used for most versions of the X Window
1092 System, @url{http://www.xfree86.org/@/3.3.6/@/COPYRIGHT2.html#3}.
1095 The license for Zlib, @url{http://www.gzip.org/@/zlib/@/zlib_license.html}.
1099 More information about these licenses and many more are on the GNU
1100 licensing web pages,
1101 @url{http://www.gnu.org/@/licenses/@/license-list.html}.
1105 @subsection @option{--help}
1107 @cindex @samp{--help} output
1109 The standard @code{--help} option should output brief documentation
1110 for how to invoke the program, on standard output, then exit
1111 successfully. Other options and arguments should be ignored once this
1112 is seen, and the program should not perform its normal function.
1114 @cindex address for bug reports
1116 Near the end of the @samp{--help} option's output, please place lines
1117 giving the email address for bug reports, the package's home page
1118 (normally @indicateurl{http://www.gnu.org/software/@var{pkg}}, and the
1119 general page for help using GNU programs. The format should be like this:
1122 Report bugs to: @var{mailing-address}
1123 @var{pkg} home page: <http://www.gnu.org/software/@var{pkg}/>
1124 General help using GNU software: <http://www.gnu.org/gethelp/>
1127 It is ok to mention other appropriate mailing lists and web pages.
1130 @node Dynamic Plug-In Interfaces
1131 @section Standards for Dynamic Plug-in Interfaces
1133 @cindex dynamic plug-ins
1135 Another aspect of keeping free programs free is encouraging
1136 development of free plug-ins, and discouraging development of
1137 proprietary plug-ins. Many GNU programs will not have anything like
1138 plug-ins at all, but those that do should follow these
1141 First, the general plug-in architecture design should closely tie the
1142 plug-in to the original code, such that the plug-in and the base
1143 program are parts of one extended program. For GCC, for example,
1144 plug-ins receive and modify GCC's internal data structures, and so
1145 clearly form an extended program with the base GCC.
1147 @vindex plugin_is_GPL_compatible
1148 Second, you should require plug-in developers to affirm that their
1149 plug-ins are released under an appropriate license. This should be
1150 enforced with a simple programmatic check. For GCC, again for
1151 example, a plug-in must define the global symbol
1152 @code{plugin_is_GPL_compatible}, thus asserting that the plug-in is
1153 released under a GPL-compatible license (@pxref{Plugins,, Plugins,
1154 gccint, GCC Internals}).
1156 By adding this check to your program you are not creating a new legal
1157 requirement. The GPL itself requires plug-ins to be free software,
1158 licensed compatibly. As long as you have followed the first rule above
1159 to keep plug-ins closely tied to your original program, the GPL and AGPL
1160 already require those plug-ins to be released under a compatible
1161 license. The symbol definition in the plug-in---or whatever equivalent
1162 works best in your program---makes it harder for anyone who might
1163 distribute proprietary plug-ins to legally defend themselves. If a case
1164 about this got to court, we can point to that symbol as evidence that
1165 the plug-in developer understood that the license had this requirement.
1169 @section Table of Long Options
1170 @cindex long option names
1171 @cindex table of long options
1173 Here is a table of long options used by GNU programs. It is surely
1174 incomplete, but we aim to list all the options that a new program might
1175 want to be compatible with. If you use names not already in the table,
1176 please send @email{bug-standards@@gnu.org} a list of them, with their
1177 meanings, so we can update the table.
1179 @c Please leave newlines between items in this table; it's much easier
1180 @c to update when it isn't completely squashed together and unreadable.
1181 @c When there is more than one short option for a long option name, put
1182 @c a semicolon between the lists of the programs that use them, not a
1183 @c period. --friedman
1187 @samp{-N} in @code{tar}.
1190 @samp{-a} in @code{du}, @code{ls}, @code{nm}, @code{stty}, @code{uname},
1191 and @code{unexpand}.
1194 @samp{-a} in @code{diff}.
1197 @samp{-A} in @code{ls}.
1200 @samp{-a} in @code{etags}, @code{tee}, @code{time};
1201 @samp{-r} in @code{tar}.
1204 @samp{-a} in @code{cp}.
1207 @samp{-n} in @code{shar}.
1210 @samp{-l} in @code{m4}.
1213 @samp{-a} in @code{diff}.
1216 @samp{-v} in @code{gawk}.
1219 @samp{-W} in @code{make}.
1222 @samp{-o} in @code{make}.
1225 @samp{-a} in @code{recode}.
1228 @samp{-a} in @code{wdiff}.
1230 @item auto-reference
1231 @samp{-A} in @code{ptx}.
1234 @samp{-n} in @code{wdiff}.
1237 For server programs, run in the background.
1239 @item backward-search
1240 @samp{-B} in @code{ctags}.
1243 @samp{-f} in @code{shar}.
1252 @samp{-b} in @code{tac}.
1255 @samp{-b} in @code{cpio} and @code{diff}.
1258 @samp{-b} in @code{shar}.
1261 Used in @code{cpio} and @code{tar}.
1264 @samp{-b} in @code{head} and @code{tail}.
1267 @samp{-b} in @code{ptx}.
1270 Used in various programs to make output shorter.
1273 @samp{-c} in @code{head}, @code{split}, and @code{tail}.
1276 @samp{-C} in @code{etags}.
1279 @samp{-A} in @code{tar}.
1282 Used in various programs to specify the directory to use.
1285 @samp{-c} in @code{chgrp} and @code{chown}.
1288 @samp{-F} in @code{ls}.
1291 @samp{-c} in @code{recode}.
1294 @samp{-c} in @code{su};
1298 @samp{-d} in @code{tar}.
1301 Used in @code{gawk}.
1304 @samp{-Z} in @code{tar} and @code{shar}.
1307 @samp{-A} in @code{tar}.
1310 @samp{-w} in @code{tar}.
1313 Used in @code{diff}.
1316 @samp{-W copyleft} in @code{gawk}.
1319 @samp{-C} in @code{ptx}, @code{recode}, and @code{wdiff};
1320 @samp{-W copyright} in @code{gawk}.
1326 @samp{-q} in @code{who}.
1329 @samp{-l} in @code{du}.
1332 Used in @code{tar} and @code{cpio}.
1335 @samp{-c} in @code{shar}.
1338 @samp{-x} in @code{ctags}.
1341 @samp{-d} in @code{touch}.
1344 @samp{-d} in @code{make} and @code{m4};
1348 @samp{-D} in @code{m4}.
1351 @samp{-d} in Bison and @code{ctags}.
1354 @samp{-D} in @code{tar}.
1357 @samp{-L} in @code{chgrp}, @code{chown}, @code{cpio}, @code{du},
1358 @code{ls}, and @code{tar}.
1360 @item dereference-args
1361 @samp{-D} in @code{du}.
1364 Specify an I/O device (special file name).
1367 @samp{-d} in @code{recode}.
1369 @item dictionary-order
1370 @samp{-d} in @code{look}.
1373 @samp{-d} in @code{tar}.
1376 @samp{-n} in @code{csplit}.
1379 Specify the directory to use, in various programs. In @code{ls}, it
1380 means to show directories themselves rather than their contents. In
1381 @code{rm} and @code{ln}, it means to not treat links to directories
1385 @samp{-x} in @code{strip}.
1387 @item discard-locals
1388 @samp{-X} in @code{strip}.
1391 @samp{-n} in @code{make}.
1394 @samp{-e} in @code{diff}.
1396 @item elide-empty-files
1397 @samp{-z} in @code{csplit}.
1400 @samp{-x} in @code{wdiff}.
1403 @samp{-z} in @code{wdiff}.
1405 @item entire-new-file
1406 @samp{-N} in @code{diff}.
1408 @item environment-overrides
1409 @samp{-e} in @code{make}.
1412 @samp{-e} in @code{xargs}.
1418 Used in @code{makeinfo}.
1421 @samp{-o} in @code{m4}.
1424 @samp{-b} in @code{ls}.
1427 @samp{-X} in @code{tar}.
1433 @samp{-x} in @code{xargs}.
1436 @samp{-e} in @code{unshar}.
1439 @samp{-t} in @code{diff}.
1442 @samp{-e} in @code{sed}.
1445 @samp{-g} in @code{nm}.
1448 @samp{-i} in @code{cpio};
1449 @samp{-x} in @code{tar}.
1452 @samp{-f} in @code{finger}.
1455 @samp{-f} in @code{su}.
1457 @item fatal-warnings
1458 @samp{-E} in @code{m4}.
1461 @samp{-f} in @code{gawk}, @code{info}, @code{make}, @code{mt},
1462 @code{sed}, and @code{tar}.
1464 @item field-separator
1465 @samp{-F} in @code{gawk}.
1471 @samp{-F} in @code{ls}.
1474 @samp{-T} in @code{tar}.
1477 Used in @code{makeinfo}.
1479 @item flag-truncation
1480 @samp{-F} in @code{ptx}.
1482 @item fixed-output-files
1486 @samp{-f} in @code{tail}.
1488 @item footnote-style
1489 Used in @code{makeinfo}.
1492 @samp{-f} in @code{cp}, @code{ln}, @code{mv}, and @code{rm}.
1495 @samp{-F} in @code{shar}.
1498 For server programs, run in the foreground;
1499 in other words, don't do anything special to run the server
1503 Used in @code{ls}, @code{time}, and @code{ptx}.
1506 @samp{-F} in @code{m4}.
1512 @samp{-g} in @code{ptx}.
1515 @samp{-x} in @code{tar}.
1518 @samp{-i} in @code{ul}.
1521 @samp{-g} in @code{recode}.
1524 @samp{-g} in @code{install}.
1527 @samp{-z} in @code{tar} and @code{shar}.
1530 @samp{-H} in @code{m4}.
1533 @samp{-h} in @code{objdump} and @code{recode}
1536 @samp{-H} in @code{who}.
1539 Used to ask for brief usage information.
1541 @item here-delimiter
1542 @samp{-d} in @code{shar}.
1544 @item hide-control-chars
1545 @samp{-q} in @code{ls}.
1548 In @code{makeinfo}, output HTML.
1551 @samp{-u} in @code{who}.
1554 @samp{-D} in @code{diff}.
1557 @samp{-I} in @code{ls};
1558 @samp{-x} in @code{recode}.
1560 @item ignore-all-space
1561 @samp{-w} in @code{diff}.
1563 @item ignore-backups
1564 @samp{-B} in @code{ls}.
1566 @item ignore-blank-lines
1567 @samp{-B} in @code{diff}.
1570 @samp{-f} in @code{look} and @code{ptx};
1571 @samp{-i} in @code{diff} and @code{wdiff}.
1574 @samp{-i} in @code{make}.
1577 @samp{-i} in @code{ptx}.
1579 @item ignore-indentation
1580 @samp{-I} in @code{etags}.
1582 @item ignore-init-file
1585 @item ignore-interrupts
1586 @samp{-i} in @code{tee}.
1588 @item ignore-matching-lines
1589 @samp{-I} in @code{diff}.
1591 @item ignore-space-change
1592 @samp{-b} in @code{diff}.
1595 @samp{-i} in @code{tar}.
1598 @samp{-i} in @code{etags};
1599 @samp{-I} in @code{m4}.
1602 @samp{-I} in @code{make}.
1605 @samp{-G} in @code{tar}.
1608 @samp{-i}, @samp{-l}, and @samp{-m} in Finger.
1611 In some programs, specify the name of the file to read as the user's
1615 @samp{-i} in @code{expand}.
1618 @samp{-T} in @code{diff}.
1621 @samp{-i} in @code{ls}.
1624 @samp{-i} in @code{cp}, @code{ln}, @code{mv}, @code{rm};
1625 @samp{-e} in @code{m4};
1626 @samp{-p} in @code{xargs};
1627 @samp{-w} in @code{tar}.
1630 @samp{-p} in @code{shar}.
1636 @samp{-j} in @code{make}.
1639 @samp{-n} in @code{make}.
1642 @samp{-k} in @code{make}.
1645 @samp{-k} in @code{csplit}.
1648 @samp{-k} in @code{du} and @code{ls}.
1651 @samp{-l} in @code{etags}.
1654 @samp{-l} in @code{wdiff}.
1656 @item level-for-gzip
1657 @samp{-g} in @code{shar}.
1660 @samp{-C} in @code{split}.
1663 Used in @code{split}, @code{head}, and @code{tail}.
1666 @samp{-l} in @code{cpio}.
1670 Used in @code{gawk}.
1673 @samp{-t} in @code{cpio};
1674 @samp{-l} in @code{recode}.
1677 @samp{-t} in @code{tar}.
1680 @samp{-N} in @code{ls}.
1683 @samp{-l} in @code{make}.
1689 Used in @code{uname}.
1692 @samp{-M} in @code{ptx}.
1695 @samp{-m} in @code{hello} and @code{uname}.
1697 @item make-directories
1698 @samp{-d} in @code{cpio}.
1701 @samp{-f} in @code{make}.
1707 @samp{-n} in @code{xargs}.
1710 @samp{-n} in @code{xargs}.
1713 @samp{-l} in @code{xargs}.
1716 @samp{-l} in @code{make}.
1719 @samp{-P} in @code{xargs}.
1722 @samp{-T} in @code{who}.
1725 @samp{-T} in @code{who}.
1728 @samp{-d} in @code{diff}.
1730 @item mixed-uuencode
1731 @samp{-M} in @code{shar}.
1734 @samp{-m} in @code{install}, @code{mkdir}, and @code{mkfifo}.
1736 @item modification-time
1737 @samp{-m} in @code{tar}.
1740 @samp{-M} in @code{tar}.
1746 @samp{-L} in @code{m4}.
1749 @samp{-a} in @code{shar}.
1752 @samp{-W} in @code{make}.
1754 @item no-builtin-rules
1755 @samp{-r} in @code{make}.
1757 @item no-character-count
1758 @samp{-w} in @code{shar}.
1760 @item no-check-existing
1761 @samp{-x} in @code{shar}.
1764 @samp{-3} in @code{wdiff}.
1767 @samp{-c} in @code{touch}.
1770 @samp{-D} in @code{etags}.
1773 @samp{-1} in @code{wdiff}.
1775 @item no-dereference
1776 @samp{-d} in @code{cp}.
1779 @samp{-2} in @code{wdiff}.
1782 @samp{-S} in @code{make}.
1788 @samp{-P} in @code{shar}.
1791 @samp{-e} in @code{gprof}.
1794 @samp{-R} in @code{etags}.
1797 @samp{-p} in @code{nm}.
1800 Don't print a startup splash screen.
1803 Used in @code{makeinfo}.
1806 @samp{-a} in @code{gprof}.
1809 @samp{-E} in @code{gprof}.
1812 @samp{-m} in @code{shar}.
1815 Used in @code{makeinfo}.
1818 Used in @code{emacsclient}.
1821 Used in various programs to inhibit warnings.
1824 @samp{-n} in @code{info}.
1827 @samp{-n} in @code{uname}.
1830 @samp{-f} in @code{cpio}.
1833 @samp{-n} in @code{objdump}.
1836 @samp{-0} in @code{xargs}.
1839 @samp{-n} in @code{cat}.
1841 @item number-nonblank
1842 @samp{-b} in @code{cat}.
1845 @samp{-n} in @code{nm}.
1847 @item numeric-uid-gid
1848 @samp{-n} in @code{cpio} and @code{ls}.
1854 @samp{-o} in @code{tar}.
1857 @samp{-o} in @code{make}.
1859 @item one-file-system
1860 @samp{-l} in @code{tar}, @code{cp}, and @code{du}.
1863 @samp{-o} in @code{ptx}.
1866 @samp{-f} in @code{gprof}.
1869 @samp{-F} in @code{gprof}.
1872 @samp{-o} in @code{getopt}, @code{fdlist}, @code{fdmount},
1873 @code{fdmountd}, and @code{fdumount}.
1876 In various programs, specify the output file name.
1879 @samp{-o} in @code{shar}.
1882 @samp{-o} in @code{rm}.
1885 @samp{-c} in @code{unshar}.
1888 @samp{-o} in @code{install}.
1891 @samp{-l} in @code{diff}.
1893 @item paragraph-indent
1894 Used in @code{makeinfo}.
1897 @samp{-p} in @code{mkdir} and @code{rmdir}.
1900 @samp{-p} in @code{ul}.
1903 @samp{-p} in @code{cpio}.
1906 @samp{-P} in @code{finger}.
1909 @samp{-c} in @code{cpio} and @code{tar}.
1912 Used in @code{gawk}.
1914 @item prefix-builtins
1915 @samp{-P} in @code{m4}.
1918 @samp{-f} in @code{csplit}.
1921 Used in @code{tar} and @code{cp}.
1923 @item preserve-environment
1924 @samp{-p} in @code{su}.
1926 @item preserve-modification-time
1927 @samp{-m} in @code{cpio}.
1929 @item preserve-order
1930 @samp{-s} in @code{tar}.
1932 @item preserve-permissions
1933 @samp{-p} in @code{tar}.
1936 @samp{-l} in @code{diff}.
1939 @samp{-L} in @code{cmp}.
1941 @item print-data-base
1942 @samp{-p} in @code{make}.
1944 @item print-directory
1945 @samp{-w} in @code{make}.
1947 @item print-file-name
1948 @samp{-o} in @code{nm}.
1951 @samp{-s} in @code{nm}.
1954 @samp{-p} in @code{wdiff}.
1957 @samp{-p} in @code{ed}.
1960 Specify an HTTP proxy.
1963 @samp{-X} in @code{shar}.
1966 @samp{-q} in @code{make}.
1969 Used in many programs to inhibit the usual output. Every
1970 program accepting @samp{--quiet} should accept @samp{--silent} as a
1974 @samp{-Q} in @code{shar}
1977 @samp{-Q} in @code{ls}.
1980 @samp{-n} in @code{diff}.
1983 Used in @code{gawk}.
1985 @item read-full-blocks
1986 @samp{-B} in @code{tar}.
1992 @samp{-n} in @code{make}.
1995 @samp{-R} in @code{tar}.
1998 Used in @code{chgrp}, @code{chown}, @code{cp}, @code{ls}, @code{diff},
2002 @samp{-r} in @code{touch}.
2005 @samp{-r} in @code{ptx}.
2008 @samp{-r} in @code{tac} and @code{etags}.
2011 @samp{-r} in @code{uname}.
2014 @samp{-R} in @code{m4}.
2017 @samp{-r} in @code{objdump}.
2020 @samp{-r} in @code{cpio}.
2023 @samp{-i} in @code{xargs}.
2025 @item report-identical-files
2026 @samp{-s} in @code{diff}.
2028 @item reset-access-time
2029 @samp{-a} in @code{cpio}.
2032 @samp{-r} in @code{ls} and @code{nm}.
2035 @samp{-f} in @code{diff}.
2037 @item right-side-defs
2038 @samp{-R} in @code{ptx}.
2041 @samp{-s} in @code{tar}.
2043 @item same-permissions
2044 @samp{-p} in @code{tar}.
2047 @samp{-g} in @code{stty}.
2052 @item sentence-regexp
2053 @samp{-S} in @code{ptx}.
2056 @samp{-S} in @code{du}.
2059 @samp{-s} in @code{tac}.
2062 Used by @code{recode} to chose files or pipes for sequencing passes.
2065 @samp{-s} in @code{su}.
2068 @samp{-A} in @code{cat}.
2070 @item show-c-function
2071 @samp{-p} in @code{diff}.
2074 @samp{-E} in @code{cat}.
2076 @item show-function-line
2077 @samp{-F} in @code{diff}.
2080 @samp{-T} in @code{cat}.
2083 Used in many programs to inhibit the usual output.
2084 Every program accepting
2085 @samp{--silent} should accept @samp{--quiet} as a synonym.
2088 @samp{-s} in @code{ls}.
2091 Specify a file descriptor for a network server to use for its socket,
2092 instead of opening and binding a new socket. This provides a way to
2093 run, in a non-privileged process, a server that normally needs a
2094 reserved port number.
2100 @samp{-W source} in @code{gawk}.
2103 @samp{-S} in @code{tar}.
2105 @item speed-large-files
2106 @samp{-H} in @code{diff}.
2109 @samp{-E} in @code{unshar}.
2111 @item split-size-limit
2112 @samp{-L} in @code{shar}.
2115 @samp{-s} in @code{cat}.
2118 @samp{-w} in @code{wdiff}.
2121 @samp{-y} in @code{wdiff}.
2124 Used in @code{tar} and @code{diff} to specify which file within
2125 a directory to start processing with.
2128 @samp{-s} in @code{wdiff}.
2130 @item stdin-file-list
2131 @samp{-S} in @code{shar}.
2134 @samp{-S} in @code{make}.
2137 @samp{-s} in @code{recode}.
2140 @samp{-s} in @code{install}.
2143 @samp{-s} in @code{strip}.
2146 @samp{-S} in @code{strip}.
2149 @samp{-s} in @code{shar}.
2152 @samp{-S} in @code{cp}, @code{ln}, @code{mv}.
2155 @samp{-b} in @code{csplit}.
2158 @samp{-s} in @code{gprof}.
2161 @samp{-s} in @code{du}.
2164 @samp{-s} in @code{ln}.
2167 Used in GDB and @code{objdump}.
2170 @samp{-s} in @code{m4}.
2173 @samp{-s} in @code{uname}.
2176 @samp{-t} in @code{expand} and @code{unexpand}.
2179 @samp{-T} in @code{ls}.
2182 @samp{-T} in @code{tput} and @code{ul}.
2183 @samp{-t} in @code{wdiff}.
2186 @samp{-a} in @code{diff}.
2189 @samp{-T} in @code{shar}.
2192 Used in @code{ls} and @code{touch}.
2195 Specify how long to wait before giving up on some operation.
2198 @samp{-O} in @code{tar}.
2201 @samp{-c} in @code{du}.
2204 @samp{-t} in @code{make}, @code{ranlib}, and @code{recode}.
2207 @samp{-t} in @code{m4}.
2210 @samp{-t} in @code{hello};
2211 @samp{-W traditional} in @code{gawk};
2212 @samp{-G} in @code{ed}, @code{m4}, and @code{ptx}.
2218 @samp{-t} in @code{ctags}.
2220 @item typedefs-and-c++
2221 @samp{-T} in @code{ctags}.
2224 @samp{-t} in @code{ptx}.
2227 @samp{-z} in @code{tar}.
2230 @samp{-u} in @code{cpio}.
2233 @samp{-U} in @code{m4}.
2235 @item undefined-only
2236 @samp{-u} in @code{nm}.
2239 @samp{-u} in @code{cp}, @code{ctags}, @code{mv}, @code{tar}.
2242 Used in @code{gawk}; same as @samp{--help}.
2245 @samp{-B} in @code{shar}.
2247 @item vanilla-operation
2248 @samp{-V} in @code{shar}.
2251 Print more information about progress. Many programs support this.
2254 @samp{-W} in @code{tar}.
2257 Print the version number.
2259 @item version-control
2260 @samp{-V} in @code{cp}, @code{ln}, @code{mv}.
2263 @samp{-v} in @code{ctags}.
2266 @samp{-V} in @code{tar}.
2269 @samp{-W} in @code{make}.
2271 @item whole-size-limit
2272 @samp{-l} in @code{shar}.
2275 @samp{-w} in @code{ls} and @code{ptx}.
2278 @samp{-W} in @code{ptx}.
2281 @samp{-T} in @code{who}.
2284 @samp{-z} in @code{gprof}.
2287 @node OID Allocations
2288 @section OID Allocations
2289 @cindex OID allocations for GNU
2294 The OID (object identifier) 1.3.6.1.4.1.11591 has been assigned to the
2295 GNU Project (thanks to Werner Koch). These are used for SNMP, LDAP,
2296 X.509 certificates, and so on. The web site
2297 @url{http://www.alvestrand.no/objectid} has a (voluntary) listing of
2298 many OID assignments.
2300 If you need a new slot for your GNU package, write
2301 @email{maintainers@@gnu.org}. Here is a list of arcs currently
2305 @include gnu-oids.texi
2310 @section Memory Usage
2311 @cindex memory usage
2313 If a program typically uses just a few meg of memory, don't bother making any
2314 effort to reduce memory usage. For example, if it is impractical for
2315 other reasons to operate on files more than a few meg long, it is
2316 reasonable to read entire input files into memory to operate on them.
2318 However, for programs such as @code{cat} or @code{tail}, that can
2319 usefully operate on very large files, it is important to avoid using a
2320 technique that would artificially limit the size of files it can handle.
2321 If a program works by lines and could be applied to arbitrary
2322 user-supplied input files, it should keep only a line in memory, because
2323 this is not very hard and users will want to be able to operate on input
2324 files that are bigger than will fit in memory all at once.
2326 If your program creates complicated data structures, just make them in
2327 memory and give a fatal error if @code{malloc} returns zero.
2331 Memory leak detectors such as @command{valgrind} can be useful, but
2332 don't complicate a program merely to avoid their false alarms. For
2333 example, if memory is used until just before a process exits, don't
2334 free it simply to silence a leak detector.
2340 Programs should be prepared to operate when @file{/usr} and @file{/etc}
2341 are read-only file systems. Thus, if the program manages log files,
2342 lock files, backup files, score files, or any other files which are
2343 modified for internal purposes, these files should not be stored in
2344 @file{/usr} or @file{/etc}.
2346 There are two exceptions. @file{/etc} is used to store system
2347 configuration information; it is reasonable for a program to modify
2348 files in @file{/etc} when its job is to update the system configuration.
2349 Also, if the user explicitly asks to modify one file in a directory, it
2350 is reasonable for the program to store other files in the same
2354 @chapter Making The Best Use of C
2356 This chapter provides advice on how best to use the C language
2357 when writing GNU software.
2360 * Formatting:: Formatting your source code.
2361 * Comments:: Commenting your work.
2362 * Syntactic Conventions:: Clean use of C constructs.
2363 * Names:: Naming variables, functions, and files.
2364 * System Portability:: Portability among different operating systems.
2365 * CPU Portability:: Supporting the range of CPU types.
2366 * System Functions:: Portability and ``standard'' library functions.
2367 * Internationalization:: Techniques for internationalization.
2368 * Character Set:: Use ASCII by default.
2369 * Quote Characters:: Use `...' in the C locale.
2370 * Mmap:: How you can safely use @code{mmap}.
2374 @section Formatting Your Source Code
2375 @cindex formatting source code
2378 @cindex braces, in C source
2379 It is important to put the open-brace that starts the body of a C
2380 function in column one, so that they will start a defun. Several
2381 tools look for open-braces in column one to find the beginnings of C
2382 functions. These tools will not work on code not formatted that way.
2384 Avoid putting open-brace, open-parenthesis or open-bracket in column
2385 one when they are inside a function, so that they won't start a defun.
2386 The open-brace that starts a @code{struct} body can go in column one
2387 if you find it useful to treat that definition as a defun.
2389 It is also important for function definitions to start the name of the
2390 function in column one. This helps people to search for function
2391 definitions, and may also help certain tools recognize them. Thus,
2392 using Standard C syntax, the format is this:
2396 concat (char *s1, char *s2)
2403 or, if you want to use traditional C syntax, format the definition like
2408 concat (s1, s2) /* Name starts in column one here */
2410 @{ /* Open brace in column one here */
2415 In Standard C, if the arguments don't fit nicely on one line,
2420 lots_of_args (int an_integer, long a_long, short a_short,
2421 double a_double, float a_float)
2425 The rest of this section gives our recommendations for other aspects of
2426 C formatting style, which is also the default style of the @code{indent}
2427 program in version 1.2 and newer. It corresponds to the options
2430 -nbad -bap -nbc -bbo -bl -bli2 -bls -ncdb -nce -cp1 -cs -di2
2431 -ndj -nfc1 -nfca -hnl -i2 -ip5 -lp -pcs -psl -nsc -nsob
2434 We don't think of these recommendations as requirements, because it
2435 causes no problems for users if two different programs have different
2438 But whatever style you use, please use it consistently, since a mixture
2439 of styles within one program tends to look ugly. If you are
2440 contributing changes to an existing program, please follow the style of
2443 For the body of the function, our recommended style looks like this:
2455 return ++x + bar ();
2459 @cindex spaces before open-paren
2460 We find it easier to read a program when it has spaces before the
2461 open-parentheses and after the commas. Especially after the commas.
2463 When you split an expression into multiple lines, split it
2464 before an operator, not after one. Here is the right way:
2466 @cindex expressions, splitting
2468 if (foo_this_is_long && bar > win (x, y, z)
2469 && remaining_condition)
2472 Try to avoid having two operators of different precedence at the same
2473 level of indentation. For example, don't write this:
2476 mode = (inmode[j] == VOIDmode
2477 || GET_MODE_SIZE (outmode[j]) > GET_MODE_SIZE (inmode[j])
2478 ? outmode[j] : inmode[j]);
2481 Instead, use extra parentheses so that the indentation shows the nesting:
2484 mode = ((inmode[j] == VOIDmode
2485 || (GET_MODE_SIZE (outmode[j]) > GET_MODE_SIZE (inmode[j])))
2486 ? outmode[j] : inmode[j]);
2489 Insert extra parentheses so that Emacs will indent the code properly.
2490 For example, the following indentation looks nice if you do it by hand,
2493 v = rup->ru_utime.tv_sec*1000 + rup->ru_utime.tv_usec/1000
2494 + rup->ru_stime.tv_sec*1000 + rup->ru_stime.tv_usec/1000;
2498 but Emacs would alter it. Adding a set of parentheses produces
2499 something that looks equally nice, and which Emacs will preserve:
2502 v = (rup->ru_utime.tv_sec*1000 + rup->ru_utime.tv_usec/1000
2503 + rup->ru_stime.tv_sec*1000 + rup->ru_stime.tv_usec/1000);
2506 Format do-while statements like this:
2518 Please use formfeed characters (control-L) to divide the program into
2519 pages at logical places (but not within a function). It does not matter
2520 just how long the pages are, since they do not have to fit on a printed
2521 page. The formfeeds should appear alone on lines by themselves.
2524 @section Commenting Your Work
2527 Every program should start with a comment saying briefly what it is for.
2528 Example: @samp{fmt - filter for simple filling of text}. This comment
2529 should be at the top of the source file containing the @samp{main}
2530 function of the program.
2532 Also, please write a brief comment at the start of each source file,
2533 with the file name and a line or two about the overall purpose of the
2536 Please write the comments in a GNU program in English, because English
2537 is the one language that nearly all programmers in all countries can
2538 read. If you do not write English well, please write comments in
2539 English as well as you can, then ask other people to help rewrite them.
2540 If you can't write comments in English, please find someone to work with
2541 you and translate your comments into English.
2543 Please put a comment on each function saying what the function does,
2544 what sorts of arguments it gets, and what the possible values of
2545 arguments mean and are used for. It is not necessary to duplicate in
2546 words the meaning of the C argument declarations, if a C type is being
2547 used in its customary fashion. If there is anything nonstandard about
2548 its use (such as an argument of type @code{char *} which is really the
2549 address of the second character of a string, not the first), or any
2550 possible values that would not work the way one would expect (such as,
2551 that strings containing newlines are not guaranteed to work), be sure
2554 Also explain the significance of the return value, if there is one.
2556 Please put two spaces after the end of a sentence in your comments, so
2557 that the Emacs sentence commands will work. Also, please write
2558 complete sentences and capitalize the first word. If a lower-case
2559 identifier comes at the beginning of a sentence, don't capitalize it!
2560 Changing the spelling makes it a different identifier. If you don't
2561 like starting a sentence with a lower case letter, write the sentence
2562 differently (e.g., ``The identifier lower-case is @dots{}'').
2564 The comment on a function is much clearer if you use the argument
2565 names to speak about the argument values. The variable name itself
2566 should be lower case, but write it in upper case when you are speaking
2567 about the value rather than the variable itself. Thus, ``the inode
2568 number NODE_NUM'' rather than ``an inode''.
2570 There is usually no purpose in restating the name of the function in
2571 the comment before it, because the reader can see that for himself.
2572 There might be an exception when the comment is so long that the function
2573 itself would be off the bottom of the screen.
2575 There should be a comment on each static variable as well, like this:
2578 /* Nonzero means truncate lines in the display;
2579 zero means continue them. */
2583 @cindex conditionals, comments for
2584 @cindex @code{#endif}, commenting
2585 Every @samp{#endif} should have a comment, except in the case of short
2586 conditionals (just a few lines) that are not nested. The comment should
2587 state the condition of the conditional that is ending, @emph{including
2588 its sense}. @samp{#else} should have a comment describing the condition
2589 @emph{and sense} of the code that follows. For example:
2597 #endif /* not foo */
2607 but, by contrast, write the comments this way for a @samp{#ifndef}:
2620 #endif /* not foo */
2624 @node Syntactic Conventions
2625 @section Clean Use of C Constructs
2626 @cindex syntactic conventions
2628 @cindex implicit @code{int}
2629 @cindex function argument, declaring
2630 Please explicitly declare the types of all objects. For example, you
2631 should explicitly declare all arguments to functions, and you should
2632 declare functions to return @code{int} rather than omitting the
2635 @cindex compiler warnings
2636 @cindex @samp{-Wall} compiler option
2637 Some programmers like to use the GCC @samp{-Wall} option, and change the
2638 code whenever it issues a warning. If you want to do this, then do.
2639 Other programmers prefer not to use @samp{-Wall}, because it gives
2640 warnings for valid and legitimate code which they do not want to change.
2641 If you want to do this, then do. The compiler should be your servant,
2646 Don't make the program ugly just to placate static analysis tools such
2647 as @command{lint}, @command{clang}, and GCC with extra warnings
2648 options such as @option{-Wconversion} and @option{-Wundef}. These
2649 tools can help find bugs and unclear code, but they can also generate
2650 so many false alarms that that it hurts readability to silence them
2651 with unnecessary casts, wrappers, and other complications. For
2652 example, please don't insert casts to @code{void} or calls to
2653 do-nothing functions merely to pacify a lint checker.
2655 Declarations of external functions and functions to appear later in the
2656 source file should all go in one place near the beginning of the file
2657 (somewhere before the first function definition in the file), or else
2658 should go in a header file. Don't put @code{extern} declarations inside
2661 @cindex temporary variables
2662 It used to be common practice to use the same local variables (with
2663 names like @code{tem}) over and over for different values within one
2664 function. Instead of doing this, it is better to declare a separate local
2665 variable for each distinct purpose, and give it a name which is
2666 meaningful. This not only makes programs easier to understand, it also
2667 facilitates optimization by good compilers. You can also move the
2668 declaration of each local variable into the smallest scope that includes
2669 all its uses. This makes the program even cleaner.
2671 Don't use local variables or parameters that shadow global identifiers.
2672 GCC's @samp{-Wshadow} option can detect this problem.
2674 @cindex multiple variables in a line
2675 Don't declare multiple variables in one declaration that spans lines.
2676 Start a new declaration on each line, instead. For example, instead
2702 (If they are global variables, each should have a comment preceding it
2705 When you have an @code{if}-@code{else} statement nested in another
2706 @code{if} statement, always put braces around the @code{if}-@code{else}.
2707 Thus, never write like this:
2730 If you have an @code{if} statement nested inside of an @code{else}
2731 statement, either write @code{else if} on one line, like this,
2741 with its @code{then}-part indented like the preceding @code{then}-part,
2742 or write the nested @code{if} within braces like this:
2754 Don't declare both a structure tag and variables or typedefs in the
2755 same declaration. Instead, declare the structure tag separately
2756 and then use it to declare the variables or typedefs.
2758 Try to avoid assignments inside @code{if}-conditions (assignments
2759 inside @code{while}-conditions are ok). For example, don't write
2763 if ((foo = (char *) malloc (sizeof *foo)) == 0)
2764 fatal ("virtual memory exhausted");
2768 instead, write this:
2771 foo = (char *) malloc (sizeof *foo);
2773 fatal ("virtual memory exhausted");
2776 This example uses zero without a cast as a null pointer constant.
2777 This is perfectly fine, except that a cast is needed when calling a
2778 varargs function or when using @code{sizeof}.
2781 @section Naming Variables, Functions, and Files
2783 @cindex names of variables, functions, and files
2784 The names of global variables and functions in a program serve as
2785 comments of a sort. So don't choose terse names---instead, look for
2786 names that give useful information about the meaning of the variable or
2787 function. In a GNU program, names should be English, like other
2790 Local variable names can be shorter, because they are used only within
2791 one context, where (presumably) comments explain their purpose.
2793 Try to limit your use of abbreviations in symbol names. It is ok to
2794 make a few abbreviations, explain what they mean, and then use them
2795 frequently, but don't use lots of obscure abbreviations.
2797 Please use underscores to separate words in a name, so that the Emacs
2798 word commands can be useful within them. Stick to lower case; reserve
2799 upper case for macros and @code{enum} constants, and for name-prefixes
2800 that follow a uniform convention.
2802 For example, you should use names like @code{ignore_space_change_flag};
2803 don't use names like @code{iCantReadThis}.
2805 Variables that indicate whether command-line options have been
2806 specified should be named after the meaning of the option, not after
2807 the option-letter. A comment should state both the exact meaning of
2808 the option and its letter. For example,
2812 /* Ignore changes in horizontal whitespace (-b). */
2813 int ignore_space_change_flag;
2817 When you want to define names with constant integer values, use
2818 @code{enum} rather than @samp{#define}. GDB knows about enumeration
2821 @cindex file-name limitations
2823 You might want to make sure that none of the file names would conflict
2824 if the files were loaded onto an MS-DOS file system which shortens the
2825 names. You can use the program @code{doschk} to test for this.
2827 Some GNU programs were designed to limit themselves to file names of 14
2828 characters or less, to avoid file name conflicts if they are read into
2829 older System V systems. Please preserve this feature in the existing
2830 GNU programs that have it, but there is no need to do this in new GNU
2831 programs. @code{doschk} also reports file names longer than 14
2834 @node System Portability
2835 @section Portability between System Types
2836 @cindex portability, between system types
2838 In the Unix world, ``portability'' refers to porting to different Unix
2839 versions. For a GNU program, this kind of portability is desirable, but
2842 The primary purpose of GNU software is to run on top of the GNU kernel,
2843 compiled with the GNU C compiler, on various types of @sc{cpu}. So the
2844 kinds of portability that are absolutely necessary are quite limited.
2845 But it is important to support Linux-based GNU systems, since they
2846 are the form of GNU that is popular.
2848 Beyond that, it is good to support the other free operating systems
2849 (*BSD), and it is nice to support other Unix-like systems if you want
2850 to. Supporting a variety of Unix-like systems is desirable, although
2851 not paramount. It is usually not too hard, so you may as well do it.
2852 But you don't have to consider it an obligation, if it does turn out to
2856 The easiest way to achieve portability to most Unix-like systems is to
2857 use Autoconf. It's unlikely that your program needs to know more
2858 information about the host platform than Autoconf can provide, simply
2859 because most of the programs that need such knowledge have already been
2862 Avoid using the format of semi-internal data bases (e.g., directories)
2863 when there is a higher-level alternative (@code{readdir}).
2865 @cindex non-@sc{posix} systems, and portability
2866 As for systems that are not like Unix, such as MSDOS, Windows, VMS, MVS,
2867 and older Macintosh systems, supporting them is often a lot of work.
2868 When that is the case, it is better to spend your time adding features
2869 that will be useful on GNU and GNU/Linux, rather than on supporting
2870 other incompatible systems.
2872 If you do support Windows, please do not abbreviate it as ``win''. In
2873 hacker terminology, calling something a ``win'' is a form of praise.
2874 You're free to praise Microsoft Windows on your own if you want, but
2875 please don't do this in GNU packages. Instead of abbreviating
2876 ``Windows'' to ``win'', you can write it in full or abbreviate it to
2877 ``woe'' or ``w''. In GNU Emacs, for instance, we use @samp{w32} in
2878 file names of Windows-specific files, but the macro for Windows
2879 conditionals is called @code{WINDOWSNT}.
2881 It is a good idea to define the ``feature test macro''
2882 @code{_GNU_SOURCE} when compiling your C files. When you compile on GNU
2883 or GNU/Linux, this will enable the declarations of GNU library extension
2884 functions, and that will usually give you a compiler error message if
2885 you define the same function names in some other way in your program.
2886 (You don't have to actually @emph{use} these functions, if you prefer
2887 to make the program more portable to other systems.)
2889 But whether or not you use these GNU extensions, you should avoid
2890 using their names for any other meanings. Doing so would make it hard
2891 to move your code into other GNU programs.
2893 @node CPU Portability
2894 @section Portability between @sc{cpu}s
2896 @cindex data types, and portability
2897 @cindex portability, and data types
2898 Even GNU systems will differ because of differences among @sc{cpu}
2899 types---for example, difference in byte ordering and alignment
2900 requirements. It is absolutely essential to handle these differences.
2901 However, don't make any effort to cater to the possibility that an
2902 @code{int} will be less than 32 bits. We don't support 16-bit machines
2905 Similarly, don't make any effort to cater to the possibility that
2906 @code{long} will be smaller than predefined types like @code{size_t}.
2907 For example, the following code is ok:
2910 printf ("size = %lu\n", (unsigned long) sizeof array);
2911 printf ("diff = %ld\n", (long) (pointer2 - pointer1));
2914 1989 Standard C requires this to work, and we know of only one
2915 counterexample: 64-bit programs on Microsoft Windows. We will
2916 leave it to those who want to port GNU programs to that environment
2917 to figure out how to do it.
2919 Predefined file-size types like @code{off_t} are an exception: they are
2920 longer than @code{long} on many platforms, so code like the above won't
2921 work with them. One way to print an @code{off_t} value portably is to
2922 print its digits yourself, one by one.
2924 Don't assume that the address of an @code{int} object is also the
2925 address of its least-significant byte. This is false on big-endian
2926 machines. Thus, don't make the following mistake:
2931 while ((c = getchar ()) != EOF)
2932 write (file_descriptor, &c, 1);
2935 @noindent Instead, use @code{unsigned char} as follows. (The @code{unsigned}
2936 is for portability to unusual systems where @code{char} is signed and
2937 where there is integer overflow checking.)
2941 while ((c = getchar ()) != EOF)
2943 unsigned char u = c;
2944 write (file_descriptor, &u, 1);
2948 It used to be ok to not worry about the difference between pointers
2949 and integers when passing arguments to functions. However, on most
2950 modern 64-bit machines pointers are wider than @code{int}.
2951 Conversely, integer types like @code{long long int} and @code{off_t}
2952 are wider than pointers on most modern 32-bit machines. Hence it's
2953 often better nowadays to use prototypes to define functions whose
2954 argument types are not trivial.
2956 In particular, if functions accept varying argument counts or types
2957 they should be declared using prototypes containing @samp{...} and
2958 defined using @file{stdarg.h}. For an example of this, please see the
2959 @uref{http://www.gnu.org/software/gnulib/, Gnulib} error module, which
2960 declares and defines the following function:
2963 /* Print a message with `fprintf (stderr, FORMAT, ...)';
2964 if ERRNUM is nonzero, follow it with ": " and strerror (ERRNUM).
2965 If STATUS is nonzero, terminate the program with `exit (STATUS)'. */
2967 void error (int status, int errnum, const char *format, ...);
2970 A simple way to use the Gnulib error module is to obtain the two
2971 source files @file{error.c} and @file{error.h} from the Gnulib library
2972 source code repository at
2973 @uref{http://git.savannah.gnu.org/@/gitweb/@/?p=gnulib.git}.
2974 Here's a sample use:
2981 char *program_name = "myprogram";
2984 xfopen (char const *name)
2986 FILE *fp = fopen (name, "r");
2988 error (1, errno, "cannot read %s", name);
2993 @cindex casting pointers to integers
2994 Avoid casting pointers to integers if you can. Such casts greatly
2995 reduce portability, and in most programs they are easy to avoid. In the
2996 cases where casting pointers to integers is essential---such as, a Lisp
2997 interpreter which stores type information as well as an address in one
2998 word---you'll have to make explicit provisions to handle different word
2999 sizes. You will also need to make provision for systems in which the
3000 normal range of addresses you can get from @code{malloc} starts far away
3003 @node System Functions
3004 @section Calling System Functions
3005 @cindex library functions, and portability
3006 @cindex portability, and library functions
3008 C implementations differ substantially. Standard C reduces but does
3009 not eliminate the incompatibilities; meanwhile, many GNU packages still
3010 support pre-standard compilers because this is not hard to do. This
3011 chapter gives recommendations for how to use the more-or-less standard C
3012 library functions to avoid unnecessary loss of portability.
3016 Don't use the return value of @code{sprintf}. It returns the number of
3017 characters written on some systems, but not on all systems.
3020 Be aware that @code{vfprintf} is not always available.
3023 @code{main} should be declared to return type @code{int}. It should
3024 terminate either by calling @code{exit} or by returning the integer
3025 status code; make sure it cannot ever return an undefined value.
3027 @cindex declaration for system functions
3029 Don't declare system functions explicitly.
3031 Almost any declaration for a system function is wrong on some system.
3032 To minimize conflicts, leave it to the system header files to declare
3033 system functions. If the headers don't declare a function, let it
3036 While it may seem unclean to use a function without declaring it, in
3037 practice this works fine for most system library functions on the
3038 systems where this really happens; thus, the disadvantage is only
3039 theoretical. By contrast, actual declarations have frequently caused
3043 If you must declare a system function, don't specify the argument types.
3044 Use an old-style declaration, not a Standard C prototype. The more you
3045 specify about the function, the more likely a conflict.
3048 In particular, don't unconditionally declare @code{malloc} or
3051 Most GNU programs use those functions just once, in functions
3052 conventionally named @code{xmalloc} and @code{xrealloc}. These
3053 functions call @code{malloc} and @code{realloc}, respectively, and
3056 Because @code{xmalloc} and @code{xrealloc} are defined in your program,
3057 you can declare them in other files without any risk of type conflict.
3059 On most systems, @code{int} is the same length as a pointer; thus, the
3060 calls to @code{malloc} and @code{realloc} work fine. For the few
3061 exceptional systems (mostly 64-bit machines), you can use
3062 @strong{conditionalized} declarations of @code{malloc} and
3063 @code{realloc}---or put these declarations in configuration files
3064 specific to those systems.
3066 @cindex string library functions
3068 The string functions require special treatment. Some Unix systems have
3069 a header file @file{string.h}; others have @file{strings.h}. Neither
3070 file name is portable. There are two things you can do: use Autoconf to
3071 figure out which file to include, or don't include either file.
3074 If you don't include either strings file, you can't get declarations for
3075 the string functions from the header file in the usual way.
3077 That causes less of a problem than you might think. The newer standard
3078 string functions should be avoided anyway because many systems still
3079 don't support them. The string functions you can use are these:
3082 strcpy strncpy strcat strncat
3083 strlen strcmp strncmp
3087 The copy and concatenate functions work fine without a declaration as
3088 long as you don't use their values. Using their values without a
3089 declaration fails on systems where the width of a pointer differs from
3090 the width of @code{int}, and perhaps in other cases. It is trivial to
3091 avoid using their values, so do that.
3093 The compare functions and @code{strlen} work fine without a declaration
3094 on most systems, possibly all the ones that GNU software runs on.
3095 You may find it necessary to declare them @strong{conditionally} on a
3098 The search functions must be declared to return @code{char *}. Luckily,
3099 there is no variation in the data type they return. But there is
3100 variation in their names. Some systems give these functions the names
3101 @code{index} and @code{rindex}; other systems use the names
3102 @code{strchr} and @code{strrchr}. Some systems support both pairs of
3103 names, but neither pair works on all systems.
3105 You should pick a single pair of names and use it throughout your
3106 program. (Nowadays, it is better to choose @code{strchr} and
3107 @code{strrchr} for new programs, since those are the standard
3108 names.) Declare both of those names as functions returning @code{char
3109 *}. On systems which don't support those names, define them as macros
3110 in terms of the other pair. For example, here is what to put at the
3111 beginning of your file (or in a header) if you want to use the names
3112 @code{strchr} and @code{strrchr} throughout:
3116 #define strchr index
3118 #ifndef HAVE_STRRCHR
3119 #define strrchr rindex
3127 Here we assume that @code{HAVE_STRCHR} and @code{HAVE_STRRCHR} are
3128 macros defined in systems where the corresponding functions exist.
3129 One way to get them properly defined is to use Autoconf.
3131 @node Internationalization
3132 @section Internationalization
3133 @cindex internationalization
3136 GNU has a library called GNU gettext that makes it easy to translate the
3137 messages in a program into various languages. You should use this
3138 library in every program. Use English for the messages as they appear
3139 in the program, and let gettext provide the way to translate them into
3142 Using GNU gettext involves putting a call to the @code{gettext} macro
3143 around each string that might need translation---like this:
3146 printf (gettext ("Processing file `%s'..."));
3150 This permits GNU gettext to replace the string @code{"Processing file
3151 `%s'..."} with a translated version.
3153 Once a program uses gettext, please make a point of writing calls to
3154 @code{gettext} when you add new strings that call for translation.
3156 Using GNU gettext in a package involves specifying a @dfn{text domain
3157 name} for the package. The text domain name is used to separate the
3158 translations for this package from the translations for other packages.
3159 Normally, the text domain name should be the same as the name of the
3160 package---for example, @samp{coreutils} for the GNU core utilities.
3162 @cindex message text, and internationalization
3163 To enable gettext to work well, avoid writing code that makes
3164 assumptions about the structure of words or sentences. When you want
3165 the precise text of a sentence to vary depending on the data, use two or
3166 more alternative string constants each containing a complete sentences,
3167 rather than inserting conditionalized words or phrases into a single
3170 Here is an example of what not to do:
3173 printf ("%s is full", capacity > 5000000 ? "disk" : "floppy disk");
3176 If you apply gettext to all strings, like this,
3179 printf (gettext ("%s is full"),
3180 capacity > 5000000 ? gettext ("disk") : gettext ("floppy disk"));
3184 the translator will hardly know that "disk" and "floppy disk" are meant to
3185 be substituted in the other string. Worse, in some languages (like French)
3186 the construction will not work: the translation of the word "full" depends
3187 on the gender of the first part of the sentence; it happens to be not the
3188 same for "disk" as for "floppy disk".
3190 Complete sentences can be translated without problems:
3193 printf (capacity > 5000000 ? gettext ("disk is full")
3194 : gettext ("floppy disk is full"));
3197 A similar problem appears at the level of sentence structure with this
3201 printf ("# Implicit rule search has%s been done.\n",
3202 f->tried_implicit ? "" : " not");
3206 Adding @code{gettext} calls to this code cannot give correct results for
3207 all languages, because negation in some languages requires adding words
3208 at more than one place in the sentence. By contrast, adding
3209 @code{gettext} calls does the job straightforwardly if the code starts
3213 printf (f->tried_implicit
3214 ? "# Implicit rule search has been done.\n",
3215 : "# Implicit rule search has not been done.\n");
3218 Another example is this one:
3221 printf ("%d file%s processed", nfiles,
3222 nfiles != 1 ? "s" : "");
3226 The problem with this example is that it assumes that plurals are made
3227 by adding `s'. If you apply gettext to the format string, like this,
3230 printf (gettext ("%d file%s processed"), nfiles,
3231 nfiles != 1 ? "s" : "");
3235 the message can use different words, but it will still be forced to use
3236 `s' for the plural. Here is a better way, with gettext being applied to
3237 the two strings independently:
3240 printf ((nfiles != 1 ? gettext ("%d files processed")
3241 : gettext ("%d file processed")),
3246 But this still doesn't work for languages like Polish, which has three
3247 plural forms: one for nfiles == 1, one for nfiles == 2, 3, 4, 22, 23, 24, ...
3248 and one for the rest. The GNU @code{ngettext} function solves this problem:
3251 printf (ngettext ("%d files processed", "%d file processed", nfiles),
3257 @section Character Set
3258 @cindex character set
3260 @cindex ASCII characters
3261 @cindex non-ASCII characters
3263 Sticking to the ASCII character set (plain text, 7-bit characters) is
3264 preferred in GNU source code comments, text documents, and other
3265 contexts, unless there is good reason to do something else because of
3266 the application domain. For example, if source code deals with the
3267 French Revolutionary calendar, it is OK if its literal strings contain
3268 accented characters in month names like ``Flor@'eal''. Also, it is OK
3269 to use non-ASCII characters to represent proper names of contributors in
3270 change logs (@pxref{Change Logs}).
3272 If you need to use non-ASCII characters, you should normally stick with
3273 one encoding, as one cannot in general mix encodings reliably.
3276 @node Quote Characters
3277 @section Quote Characters
3278 @cindex quote characters
3279 @cindex locale-specific quote characters
3281 @cindex grave accent
3283 In the C locale, GNU programs should stick to plain ASCII for quotation
3284 characters in messages to users: preferably 0x60 (@samp{`}) for left
3285 quotes and 0x27 (@samp{'}) for right quotes. It is ok, but not
3286 required, to use locale-specific quotes in other locales.
3288 The @uref{http://www.gnu.org/software/gnulib/, Gnulib} @code{quote} and
3289 @code{quotearg} modules provide a reasonably straightforward way to
3290 support locale-specific quote characters, as well as taking care of
3291 other issues, such as quoting a filename that itself contains a quote
3292 character. See the Gnulib documentation for usage details.
3294 In any case, the documentation for your program should clearly specify
3295 how it does quoting, if different than the preferred method of @samp{`}
3296 and @samp{'}. This is especially important if the output of your
3297 program is ever likely to be parsed by another program.
3299 Quotation characters are a difficult area in the computing world at
3300 this time: there are no true left or right quote characters in Latin1;
3301 the @samp{`} character we use was standardized there as a grave
3302 accent. Moreover, Latin1 is still not universally usable.
3304 Unicode contains the unambiguous quote characters required, and its
3305 common encoding UTF-8 is upward compatible with Latin1. However,
3306 Unicode and UTF-8 are not universally well-supported, either.
3308 This may change over the next few years, and then we will revisit
3316 Don't assume that @code{mmap} either works on all files or fails
3317 for all files. It may work on some files and fail on others.
3319 The proper way to use @code{mmap} is to try it on the specific file for
3320 which you want to use it---and if @code{mmap} doesn't work, fall back on
3321 doing the job in another way using @code{read} and @code{write}.
3323 The reason this precaution is needed is that the GNU kernel (the HURD)
3324 provides a user-extensible file system, in which there can be many
3325 different kinds of ``ordinary files.'' Many of them support
3326 @code{mmap}, but some do not. It is important to make programs handle
3327 all these kinds of files.
3330 @chapter Documenting Programs
3331 @cindex documentation
3333 A GNU program should ideally come with full free documentation, adequate
3334 for both reference and tutorial purposes. If the package can be
3335 programmed or extended, the documentation should cover programming or
3336 extending it, as well as just using it.
3339 * GNU Manuals:: Writing proper manuals.
3340 * Doc Strings and Manuals:: Compiling doc strings doesn't make a manual.
3341 * Manual Structure Details:: Specific structure conventions.
3342 * License for Manuals:: Writing the distribution terms for a manual.
3343 * Manual Credits:: Giving credit to documentation contributors.
3344 * Printed Manuals:: Mentioning the printed manual.
3345 * NEWS File:: NEWS files supplement manuals.
3346 * Change Logs:: Recording changes.
3347 * Man Pages:: Man pages are secondary.
3348 * Reading other Manuals:: How far you can go in learning
3353 @section GNU Manuals
3355 The preferred document format for the GNU system is the Texinfo
3356 formatting language. Every GNU package should (ideally) have
3357 documentation in Texinfo both for reference and for learners. Texinfo
3358 makes it possible to produce a good quality formatted book, using
3359 @TeX{}, and to generate an Info file. It is also possible to generate
3360 HTML output from Texinfo source. See the Texinfo manual, either the
3361 hardcopy, or the on-line version available through @code{info} or the
3362 Emacs Info subsystem (@kbd{C-h i}).
3364 Nowadays some other formats such as Docbook and Sgmltexi can be
3365 converted automatically into Texinfo. It is ok to produce the Texinfo
3366 documentation by conversion this way, as long as it gives good results.
3368 Make sure your manual is clear to a reader who knows nothing about the
3369 topic and reads it straight through. This means covering basic topics
3370 at the beginning, and advanced topics only later. This also means
3371 defining every specialized term when it is first used.
3373 Programmers tend to carry over the structure of the program as the
3374 structure for its documentation. But this structure is not
3375 necessarily good for explaining how to use the program; it may be
3376 irrelevant and confusing for a user.
3378 Instead, the right way to structure documentation is according to the
3379 concepts and questions that a user will have in mind when reading it.
3380 This principle applies at every level, from the lowest (ordering
3381 sentences in a paragraph) to the highest (ordering of chapter topics
3382 within the manual). Sometimes this structure of ideas matches the
3383 structure of the implementation of the software being documented---but
3384 often they are different. An important part of learning to write good
3385 documentation is to learn to notice when you have unthinkingly
3386 structured the documentation like the implementation, stop yourself,
3387 and look for better alternatives.
3389 For example, each program in the GNU system probably ought to be
3390 documented in one manual; but this does not mean each program should
3391 have its own manual. That would be following the structure of the
3392 implementation, rather than the structure that helps the user
3395 Instead, each manual should cover a coherent @emph{topic}. For example,
3396 instead of a manual for @code{diff} and a manual for @code{diff3}, we
3397 have one manual for ``comparison of files'' which covers both of those
3398 programs, as well as @code{cmp}. By documenting these programs
3399 together, we can make the whole subject clearer.
3401 The manual which discusses a program should certainly document all of
3402 the program's command-line options and all of its commands. It should
3403 give examples of their use. But don't organize the manual as a list
3404 of features. Instead, organize it logically, by subtopics. Address
3405 the questions that a user will ask when thinking about the job that
3406 the program does. Don't just tell the reader what each feature can
3407 do---say what jobs it is good for, and show how to use it for those
3408 jobs. Explain what is recommended usage, and what kinds of usage
3411 In general, a GNU manual should serve both as tutorial and reference.
3412 It should be set up for convenient access to each topic through Info,
3413 and for reading straight through (appendixes aside). A GNU manual
3414 should give a good introduction to a beginner reading through from the
3415 start, and should also provide all the details that hackers want.
3416 The Bison manual is a good example of this---please take a look at it
3417 to see what we mean.
3419 That is not as hard as it first sounds. Arrange each chapter as a
3420 logical breakdown of its topic, but order the sections, and write their
3421 text, so that reading the chapter straight through makes sense. Do
3422 likewise when structuring the book into chapters, and when structuring a
3423 section into paragraphs. The watchword is, @emph{at each point, address
3424 the most fundamental and important issue raised by the preceding text.}
3426 If necessary, add extra chapters at the beginning of the manual which
3427 are purely tutorial and cover the basics of the subject. These provide
3428 the framework for a beginner to understand the rest of the manual. The
3429 Bison manual provides a good example of how to do this.
3431 To serve as a reference, a manual should have an Index that list all the
3432 functions, variables, options, and important concepts that are part of
3433 the program. One combined Index should do for a short manual, but
3434 sometimes for a complex package it is better to use multiple indices.
3435 The Texinfo manual includes advice on preparing good index entries, see
3436 @ref{Index Entries, , Making Index Entries, texinfo, GNU Texinfo}, and
3437 see @ref{Indexing Commands, , Defining the Entries of an
3438 Index, texinfo, GNU Texinfo}.
3440 Don't use Unix man pages as a model for how to write GNU documentation;
3441 most of them are terse, badly structured, and give inadequate
3442 explanation of the underlying concepts. (There are, of course, some
3443 exceptions.) Also, Unix man pages use a particular format which is
3444 different from what we use in GNU manuals.
3446 Please include an email address in the manual for where to report
3447 bugs @emph{in the text of the manual}.
3449 Please do not use the term ``pathname'' that is used in Unix
3450 documentation; use ``file name'' (two words) instead. We use the term
3451 ``path'' only for search paths, which are lists of directory names.
3453 Please do not use the term ``illegal'' to refer to erroneous input to
3454 a computer program. Please use ``invalid'' for this, and reserve the
3455 term ``illegal'' for activities prohibited by law.
3457 Please do not write @samp{()} after a function name just to indicate
3458 it is a function. @code{foo ()} is not a function, it is a function
3459 call with no arguments.
3461 @node Doc Strings and Manuals
3462 @section Doc Strings and Manuals
3464 Some programming systems, such as Emacs, provide a documentation string
3465 for each function, command or variable. You may be tempted to write a
3466 reference manual by compiling the documentation strings and writing a
3467 little additional text to go around them---but you must not do it. That
3468 approach is a fundamental mistake. The text of well-written
3469 documentation strings will be entirely wrong for a manual.
3471 A documentation string needs to stand alone---when it appears on the
3472 screen, there will be no other text to introduce or explain it.
3473 Meanwhile, it can be rather informal in style.
3475 The text describing a function or variable in a manual must not stand
3476 alone; it appears in the context of a section or subsection. Other text
3477 at the beginning of the section should explain some of the concepts, and
3478 should often make some general points that apply to several functions or
3479 variables. The previous descriptions of functions and variables in the
3480 section will also have given information about the topic. A description
3481 written to stand alone would repeat some of that information; this
3482 redundancy looks bad. Meanwhile, the informality that is acceptable in
3483 a documentation string is totally unacceptable in a manual.
3485 The only good way to use documentation strings in writing a good manual
3486 is to use them as a source of information for writing good text.
3488 @node Manual Structure Details
3489 @section Manual Structure Details
3490 @cindex manual structure
3492 The title page of the manual should state the version of the programs or
3493 packages documented in the manual. The Top node of the manual should
3494 also contain this information. If the manual is changing more
3495 frequently than or independent of the program, also state a version
3496 number for the manual in both of these places.
3498 Each program documented in the manual should have a node named
3499 @samp{@var{program} Invocation} or @samp{Invoking @var{program}}. This
3500 node (together with its subnodes, if any) should describe the program's
3501 command line arguments and how to run it (the sort of information people
3502 would look for in a man page). Start with an @samp{@@example}
3503 containing a template for all the options and arguments that the program
3506 Alternatively, put a menu item in some menu whose item name fits one of
3507 the above patterns. This identifies the node which that item points to
3508 as the node for this purpose, regardless of the node's actual name.
3510 The @samp{--usage} feature of the Info reader looks for such a node
3511 or menu item in order to find the relevant text, so it is essential
3512 for every Texinfo file to have one.
3514 If one manual describes several programs, it should have such a node for
3515 each program described in the manual.
3517 @node License for Manuals
3518 @section License for Manuals
3519 @cindex license for manuals
3521 Please use the GNU Free Documentation License for all GNU manuals that
3522 are more than a few pages long. Likewise for a collection of short
3523 documents---you only need one copy of the GNU FDL for the whole
3524 collection. For a single short document, you can use a very permissive
3525 non-copyleft license, to avoid taking up space with a long license.
3527 See @uref{http://www.gnu.org/copyleft/fdl-howto.html} for more explanation
3528 of how to employ the GFDL.
3530 Note that it is not obligatory to include a copy of the GNU GPL or GNU
3531 LGPL in a manual whose license is neither the GPL nor the LGPL. It can
3532 be a good idea to include the program's license in a large manual; in a
3533 short manual, whose size would be increased considerably by including
3534 the program's license, it is probably better not to include it.
3536 @node Manual Credits
3537 @section Manual Credits
3538 @cindex credits for manuals
3540 Please credit the principal human writers of the manual as the authors,
3541 on the title page of the manual. If a company sponsored the work, thank
3542 the company in a suitable place in the manual, but do not cite the
3543 company as an author.
3545 @node Printed Manuals
3546 @section Printed Manuals
3548 The FSF publishes some GNU manuals in printed form. To encourage sales
3549 of these manuals, the on-line versions of the manual should mention at
3550 the very start that the printed manual is available and should point at
3551 information for getting it---for instance, with a link to the page
3552 @url{http://www.gnu.org/order/order.html}. This should not be included
3553 in the printed manual, though, because there it is redundant.
3555 It is also useful to explain in the on-line forms of the manual how the
3556 user can print out the manual from the sources.
3559 @section The NEWS File
3560 @cindex @file{NEWS} file
3562 In addition to its manual, the package should have a file named
3563 @file{NEWS} which contains a list of user-visible changes worth
3564 mentioning. In each new release, add items to the front of the file and
3565 identify the version they pertain to. Don't discard old items; leave
3566 them in the file after the newer items. This way, a user upgrading from
3567 any previous version can see what is new.
3569 If the @file{NEWS} file gets very long, move some of the older items
3570 into a file named @file{ONEWS} and put a note at the end referring the
3574 @section Change Logs
3577 Keep a change log to describe all the changes made to program source
3578 files. The purpose of this is so that people investigating bugs in the
3579 future will know about the changes that might have introduced the bug.
3580 Often a new bug can be found by looking at what was recently changed.
3581 More importantly, change logs can help you eliminate conceptual
3582 inconsistencies between different parts of a program, by giving you a
3583 history of how the conflicting concepts arose and who they came from.
3586 * Change Log Concepts::
3587 * Style of Change Logs::
3589 * Conditional Changes::
3590 * Indicating the Part Changed::
3593 @node Change Log Concepts
3594 @subsection Change Log Concepts
3596 You can think of the change log as a conceptual ``undo list'' which
3597 explains how earlier versions were different from the current version.
3598 People can see the current version; they don't need the change log
3599 to tell them what is in it. What they want from a change log is a
3600 clear explanation of how the earlier version differed.
3602 The change log file is normally called @file{ChangeLog} and covers an
3603 entire directory. Each directory can have its own change log, or a
3604 directory can use the change log of its parent directory---it's up to
3607 Another alternative is to record change log information with a version
3608 control system such as RCS or CVS. This can be converted automatically
3609 to a @file{ChangeLog} file using @code{rcs2log}; in Emacs, the command
3610 @kbd{C-x v a} (@code{vc-update-change-log}) does the job.
3612 There's no need to describe the full purpose of the changes or how
3613 they work together. However, sometimes it is useful to write one line
3614 to describe the overall purpose of a change or a batch of changes. If
3615 you think that a change calls for explanation, you're probably right.
3616 Please do explain it---but please put the full explanation in comments
3617 in the code, where people will see it whenever they see the code. For
3618 example, ``New function'' is enough for the change log when you add a
3619 function, because there should be a comment before the function
3620 definition to explain what it does.
3622 In the past, we recommended not mentioning changes in non-software
3623 files (manuals, help files, etc.) in change logs. However, we've been
3624 advised that it is a good idea to include them, for the sake of
3627 The easiest way to add an entry to @file{ChangeLog} is with the Emacs
3628 command @kbd{M-x add-change-log-entry}. An entry should have an
3629 asterisk, the name of the changed file, and then in parentheses the name
3630 of the changed functions, variables or whatever, followed by a colon.
3631 Then describe the changes you made to that function or variable.
3633 @node Style of Change Logs
3634 @subsection Style of Change Logs
3635 @cindex change logs, style
3637 Here are some simple examples of change log entries, starting with the
3638 header line that says who made the change and when it was installed,
3639 followed by descriptions of specific changes. (These examples are
3640 drawn from Emacs and GCC.)
3643 1998-08-17 Richard Stallman <rms@@gnu.org>
3645 * register.el (insert-register): Return nil.
3646 (jump-to-register): Likewise.
3648 * sort.el (sort-subr): Return nil.
3650 * tex-mode.el (tex-bibtex-file, tex-file, tex-region):
3651 Restart the tex shell if process is gone or stopped.
3652 (tex-shell-running): New function.
3654 * expr.c (store_one_arg): Round size up for move_block_to_reg.
3655 (expand_call): Round up when emitting USE insns.
3656 * stmt.c (assign_parms): Round size up for move_block_from_reg.
3659 It's important to name the changed function or variable in full. Don't
3660 abbreviate function or variable names, and don't combine them.
3661 Subsequent maintainers will often search for a function name to find all
3662 the change log entries that pertain to it; if you abbreviate the name,
3663 they won't find it when they search.
3665 For example, some people are tempted to abbreviate groups of function
3666 names by writing @samp{* register.el (@{insert,jump-to@}-register)};
3667 this is not a good idea, since searching for @code{jump-to-register} or
3668 @code{insert-register} would not find that entry.
3670 Separate unrelated change log entries with blank lines. When two
3671 entries represent parts of the same change, so that they work together,
3672 then don't put blank lines between them. Then you can omit the file
3673 name and the asterisk when successive entries are in the same file.
3675 Break long lists of function names by closing continued lines with
3676 @samp{)}, rather than @samp{,}, and opening the continuation with
3677 @samp{(} as in this example:
3680 * keyboard.c (menu_bar_items, tool_bar_items)
3681 (Fexecute_extended_command): Deal with `keymap' property.
3684 When you install someone else's changes, put the contributor's name in
3685 the change log entry rather than in the text of the entry. In other
3689 2002-07-14 John Doe <jdoe@@gnu.org>
3691 * sewing.c: Make it sew.
3698 2002-07-14 Usual Maintainer <usual@@gnu.org>
3700 * sewing.c: Make it sew. Patch by jdoe@@gnu.org.
3703 As for the date, that should be the date you applied the change.
3705 @node Simple Changes
3706 @subsection Simple Changes
3708 Certain simple kinds of changes don't need much detail in the change
3711 When you change the calling sequence of a function in a simple fashion,
3712 and you change all the callers of the function to use the new calling
3713 sequence, there is no need to make individual entries for all the
3714 callers that you changed. Just write in the entry for the function
3715 being called, ``All callers changed''---like this:
3718 * keyboard.c (Fcommand_execute): New arg SPECIAL.
3719 All callers changed.
3722 When you change just comments or doc strings, it is enough to write an
3723 entry for the file, without mentioning the functions. Just ``Doc
3724 fixes'' is enough for the change log.
3726 There's no technical need to make change log entries for documentation
3727 files. This is because documentation is not susceptible to bugs that
3728 are hard to fix. Documentation does not consist of parts that must
3729 interact in a precisely engineered fashion. To correct an error, you
3730 need not know the history of the erroneous passage; it is enough to
3731 compare what the documentation says with the way the program actually
3734 However, you should keep change logs for documentation files when the
3735 project gets copyright assignments from its contributors, so as to
3736 make the records of authorship more accurate.
3738 @node Conditional Changes
3739 @subsection Conditional Changes
3740 @cindex conditional changes, and change logs
3741 @cindex change logs, conditional changes
3743 C programs often contain compile-time @code{#if} conditionals. Many
3744 changes are conditional; sometimes you add a new definition which is
3745 entirely contained in a conditional. It is very useful to indicate in
3746 the change log the conditions for which the change applies.
3748 Our convention for indicating conditional changes is to use square
3749 brackets around the name of the condition.
3751 Here is a simple example, describing a change which is conditional but
3752 does not have a function or entity name associated with it:
3755 * xterm.c [SOLARIS2]: Include string.h.
3758 Here is an entry describing a new definition which is entirely
3759 conditional. This new definition for the macro @code{FRAME_WINDOW_P} is
3760 used only when @code{HAVE_X_WINDOWS} is defined:
3763 * frame.h [HAVE_X_WINDOWS] (FRAME_WINDOW_P): Macro defined.
3766 Here is an entry for a change within the function @code{init_display},
3767 whose definition as a whole is unconditional, but the changes themselves
3768 are contained in a @samp{#ifdef HAVE_LIBNCURSES} conditional:
3771 * dispnew.c (init_display) [HAVE_LIBNCURSES]: If X, call tgetent.
3774 Here is an entry for a change that takes affect only when
3775 a certain macro is @emph{not} defined:
3778 (gethostname) [!HAVE_SOCKETS]: Replace with winsock version.
3781 @node Indicating the Part Changed
3782 @subsection Indicating the Part Changed
3784 Indicate the part of a function which changed by using angle brackets
3785 enclosing an indication of what the changed part does. Here is an entry
3786 for a change in the part of the function @code{sh-while-getopts} that
3787 deals with @code{sh} commands:
3790 * progmodes/sh-script.el (sh-while-getopts) <sh>: Handle case that
3791 user-specified option string is empty.
3799 In the GNU project, man pages are secondary. It is not necessary or
3800 expected for every GNU program to have a man page, but some of them do.
3801 It's your choice whether to include a man page in your program.
3803 When you make this decision, consider that supporting a man page
3804 requires continual effort each time the program is changed. The time
3805 you spend on the man page is time taken away from more useful work.
3807 For a simple program which changes little, updating the man page may be
3808 a small job. Then there is little reason not to include a man page, if
3811 For a large program that changes a great deal, updating a man page may
3812 be a substantial burden. If a user offers to donate a man page, you may
3813 find this gift costly to accept. It may be better to refuse the man
3814 page unless the same person agrees to take full responsibility for
3815 maintaining it---so that you can wash your hands of it entirely. If
3816 this volunteer later ceases to do the job, then don't feel obliged to
3817 pick it up yourself; it may be better to withdraw the man page from the
3818 distribution until someone else agrees to update it.
3820 When a program changes only a little, you may feel that the
3821 discrepancies are small enough that the man page remains useful without
3822 updating. If so, put a prominent note near the beginning of the man
3823 page explaining that you don't maintain it and that the Texinfo manual
3824 is more authoritative. The note should say how to access the Texinfo
3827 Be sure that man pages include a copyright statement and free license.
3828 The simple all-permissive license is appropriate for simple man pages
3829 (@pxref{License Notices for Other Files,,,maintain,Information for GNU
3832 For long man pages, with enough explanation and documentation that
3833 they can be considered true manuals, use the GFDL (@pxref{License for
3836 Finally, the GNU help2man program
3837 (@uref{http://www.gnu.org/software/help2man/}) is one way to automate
3838 generation of a man page, in this case from @option{--help} output.
3839 This is sufficient in many cases.
3841 @node Reading other Manuals
3842 @section Reading other Manuals
3844 There may be non-free books or documentation files that describe the
3845 program you are documenting.
3847 It is ok to use these documents for reference, just as the author of a
3848 new algebra textbook can read other books on algebra. A large portion
3849 of any non-fiction book consists of facts, in this case facts about how
3850 a certain program works, and these facts are necessarily the same for
3851 everyone who writes about the subject. But be careful not to copy your
3852 outline structure, wording, tables or examples from preexisting non-free
3853 documentation. Copying from free documentation may be ok; please check
3854 with the FSF about the individual case.
3856 @node Managing Releases
3857 @chapter The Release Process
3860 Making a release is more than just bundling up your source files in a
3861 tar file and putting it up for FTP. You should set up your software so
3862 that it can be configured to run on a variety of systems. Your Makefile
3863 should conform to the GNU standards described below, and your directory
3864 layout should also conform to the standards discussed below. Doing so
3865 makes it easy to include your package into the larger framework of
3869 * Configuration:: How configuration of GNU packages should work.
3870 * Makefile Conventions:: Makefile conventions.
3871 * Releases:: Making releases
3875 @section How Configuration Should Work
3876 @cindex program configuration
3879 Each GNU distribution should come with a shell script named
3880 @code{configure}. This script is given arguments which describe the
3881 kind of machine and system you want to compile the program for.
3882 The @code{configure} script must record the configuration options so
3883 that they affect compilation.
3885 The description here is the specification of the interface for the
3886 @code{configure} script in GNU packages. Many packages implement it
3887 using GNU Autoconf (@pxref{Top,, Introduction, autoconf, Autoconf})
3888 and/or GNU Automake (@pxref{Top,, Introduction, automake, Automake}),
3889 but you do not have to use these tools. You can implement it any way
3890 you like; for instance, by making @code{configure} be a wrapper around
3891 a completely different configuration system.
3893 Another way for the @code{configure} script to operate is to make a
3894 link from a standard name such as @file{config.h} to the proper
3895 configuration file for the chosen system. If you use this technique,
3896 the distribution should @emph{not} contain a file named
3897 @file{config.h}. This is so that people won't be able to build the
3898 program without configuring it first.
3900 Another thing that @code{configure} can do is to edit the Makefile. If
3901 you do this, the distribution should @emph{not} contain a file named
3902 @file{Makefile}. Instead, it should include a file @file{Makefile.in} which
3903 contains the input used for editing. Once again, this is so that people
3904 won't be able to build the program without configuring it first.
3906 If @code{configure} does write the @file{Makefile}, then @file{Makefile}
3907 should have a target named @file{Makefile} which causes @code{configure}
3908 to be rerun, setting up the same configuration that was set up last
3909 time. The files that @code{configure} reads should be listed as
3910 dependencies of @file{Makefile}.
3912 All the files which are output from the @code{configure} script should
3913 have comments at the beginning explaining that they were generated
3914 automatically using @code{configure}. This is so that users won't think
3915 of trying to edit them by hand.
3917 The @code{configure} script should write a file named @file{config.status}
3918 which describes which configuration options were specified when the
3919 program was last configured. This file should be a shell script which,
3920 if run, will recreate the same configuration.
3922 The @code{configure} script should accept an option of the form
3923 @samp{--srcdir=@var{dirname}} to specify the directory where sources are found
3924 (if it is not the current directory). This makes it possible to build
3925 the program in a separate directory, so that the actual source directory
3928 If the user does not specify @samp{--srcdir}, then @code{configure} should
3929 check both @file{.} and @file{..} to see if it can find the sources. If
3930 it finds the sources in one of these places, it should use them from
3931 there. Otherwise, it should report that it cannot find the sources, and
3932 should exit with nonzero status.
3934 Usually the easy way to support @samp{--srcdir} is by editing a
3935 definition of @code{VPATH} into the Makefile. Some rules may need to
3936 refer explicitly to the specified source directory. To make this
3937 possible, @code{configure} can add to the Makefile a variable named
3938 @code{srcdir} whose value is precisely the specified directory.
3940 In addition, the @samp{configure} script should take options
3941 corresponding to most of the standard directory variables
3942 (@pxref{Directory Variables}). Here is the list:
3945 --prefix --exec-prefix --bindir --sbindir --libexecdir --sysconfdir
3946 --sharedstatedir --localstatedir --libdir --includedir --oldincludedir
3947 --datarootdir --datadir --infodir --localedir --mandir --docdir
3948 --htmldir --dvidir --pdfdir --psdir
3951 The @code{configure} script should also take an argument which specifies the
3952 type of system to build the program for. This argument should look like
3956 @var{cpu}-@var{company}-@var{system}
3959 For example, an Athlon-based GNU/Linux system might be
3960 @samp{i686-pc-linux-gnu}.
3962 The @code{configure} script needs to be able to decode all plausible
3963 alternatives for how to describe a machine. Thus,
3964 @samp{athlon-pc-gnu/linux} would be a valid alias. There is a shell
3966 @uref{http://git.savannah.gnu.org/@/gitweb/@/?p=config.git;a=blob_plain;f=config.sub;hb=HEAD,
3967 @file{config.sub}} that you can use as a subroutine to validate system
3968 types and canonicalize aliases.
3970 The @code{configure} script should also take the option
3971 @option{--build=@var{buildtype}}, which should be equivalent to a
3972 plain @var{buildtype} argument. For example, @samp{configure
3973 --build=i686-pc-linux-gnu} is equivalent to @samp{configure
3974 i686-pc-linux-gnu}. When the build type is not specified by an option
3975 or argument, the @code{configure} script should normally guess it using
3977 @uref{http://git.savannah.gnu.org/@/gitweb/@/?p=config.git;a=blob_plain;f=config.guess;hb=HEAD,
3978 @file{config.guess}}.
3980 @cindex optional features, configure-time
3981 Other options are permitted to specify in more detail the software
3982 or hardware present on the machine, to include or exclude optional parts
3983 of the package, or to adjust the name of some tools or arguments to them:
3986 @item --enable-@var{feature}@r{[}=@var{parameter}@r{]}
3987 Configure the package to build and install an optional user-level
3988 facility called @var{feature}. This allows users to choose which
3989 optional features to include. Giving an optional @var{parameter} of
3990 @samp{no} should omit @var{feature}, if it is built by default.
3992 No @samp{--enable} option should @strong{ever} cause one feature to
3993 replace another. No @samp{--enable} option should ever substitute one
3994 useful behavior for another useful behavior. The only proper use for
3995 @samp{--enable} is for questions of whether to build part of the program
3998 @item --with-@var{package}
3999 @c @r{[}=@var{parameter}@r{]}
4000 The package @var{package} will be installed, so configure this package
4001 to work with @var{package}.
4003 @c Giving an optional @var{parameter} of
4004 @c @samp{no} should omit @var{package}, if it is used by default.
4006 Possible values of @var{package} include
4007 @samp{gnu-as} (or @samp{gas}), @samp{gnu-ld}, @samp{gnu-libc},
4013 Do not use a @samp{--with} option to specify the file name to use to
4014 find certain files. That is outside the scope of what @samp{--with}
4017 @item @var{variable}=@var{value}
4018 Set the value of the variable @var{variable} to @var{value}. This is
4019 used to override the default values of commands or arguments in the
4020 build process. For example, the user could issue @samp{configure
4021 CFLAGS=-g CXXFLAGS=-g} to build with debugging information and without
4022 the default optimization.
4024 Specifying variables as arguments to @code{configure}, like this:
4028 is preferable to setting them in environment variables:
4032 as it helps to recreate the same configuration later with
4033 @file{config.status}. However, both methods should be supported.
4036 All @code{configure} scripts should accept all of the ``detail''
4037 options and the variable settings, whether or not they make any
4038 difference to the particular package at hand. In particular, they
4039 should accept any option that starts with @samp{--with-} or
4040 @samp{--enable-}. This is so users will be able to configure an
4041 entire GNU source tree at once with a single set of options.
4043 You will note that the categories @samp{--with-} and @samp{--enable-}
4044 are narrow: they @strong{do not} provide a place for any sort of option
4045 you might think of. That is deliberate. We want to limit the possible
4046 configuration options in GNU software. We do not want GNU programs to
4047 have idiosyncratic configuration options.
4049 Packages that perform part of the compilation process may support
4050 cross-compilation. In such a case, the host and target machines for the
4051 program may be different.
4053 The @code{configure} script should normally treat the specified type of
4054 system as both the host and the target, thus producing a program which
4055 works for the same type of machine that it runs on.
4057 To compile a program to run on a host type that differs from the build
4058 type, use the configure option @option{--host=@var{hosttype}}, where
4059 @var{hosttype} uses the same syntax as @var{buildtype}. The host type
4060 normally defaults to the build type.
4062 To configure a cross-compiler, cross-assembler, or what have you, you
4063 should specify a target different from the host, using the configure
4064 option @samp{--target=@var{targettype}}. The syntax for
4065 @var{targettype} is the same as for the host type. So the command would
4069 ./configure --host=@var{hosttype} --target=@var{targettype}
4072 The target type normally defaults to the host type.
4073 Programs for which cross-operation is not meaningful need not accept the
4074 @samp{--target} option, because configuring an entire operating system for
4075 cross-operation is not a meaningful operation.
4077 Some programs have ways of configuring themselves automatically. If
4078 your program is set up to do this, your @code{configure} script can simply
4079 ignore most of its arguments.
4081 @comment The makefile standards are in a separate file that is also
4082 @comment included by make.texinfo. Done by roland@gnu.ai.mit.edu on 1/6/93.
4083 @comment For this document, turn chapters into sections, etc.
4085 @include make-stds.texi
4089 @section Making Releases
4092 You should identify each release with a pair of version numbers, a
4093 major version and a minor. We have no objection to using more than
4094 two numbers, but it is very unlikely that you really need them.
4096 Package the distribution of @code{Foo version 69.96} up in a gzipped tar
4097 file with the name @file{foo-69.96.tar.gz}. It should unpack into a
4098 subdirectory named @file{foo-69.96}.
4100 Building and installing the program should never modify any of the files
4101 contained in the distribution. This means that all the files that form
4102 part of the program in any way must be classified into @dfn{source
4103 files} and @dfn{non-source files}. Source files are written by humans
4104 and never changed automatically; non-source files are produced from
4105 source files by programs under the control of the Makefile.
4107 @cindex @file{README} file
4108 The distribution should contain a file named @file{README} which gives
4109 the name of the package, and a general description of what it does. It
4110 is also good to explain the purpose of each of the first-level
4111 subdirectories in the package, if there are any. The @file{README} file
4112 should either state the version number of the package, or refer to where
4113 in the package it can be found.
4115 The @file{README} file should refer to the file @file{INSTALL}, which
4116 should contain an explanation of the installation procedure.
4118 The @file{README} file should also refer to the file which contains the
4119 copying conditions. The GNU GPL, if used, should be in a file called
4120 @file{COPYING}. If the GNU LGPL is used, it should be in a file called
4121 @file{COPYING.LESSER}.
4123 Naturally, all the source files must be in the distribution. It is okay
4124 to include non-source files in the distribution, provided they are
4125 up-to-date and machine-independent, so that building the distribution
4126 normally will never modify them. We commonly include non-source files
4127 produced by Bison, @code{lex}, @TeX{}, and @code{makeinfo}; this helps avoid
4128 unnecessary dependencies between our distributions, so that users can
4129 install whichever packages they want to install.
4131 Non-source files that might actually be modified by building and
4132 installing the program should @strong{never} be included in the
4133 distribution. So if you do distribute non-source files, always make
4134 sure they are up to date when you make a new distribution.
4136 Make sure that all the files in the distribution are world-readable, and
4137 that directories are world-readable and world-searchable (octal mode 755).
4138 We used to recommend that all directories in the distribution also be
4139 world-writable (octal mode 777), because ancient versions of @code{tar}
4140 would otherwise not cope when extracting the archive as an unprivileged
4141 user. That can easily lead to security issues when creating the archive,
4142 however, so now we recommend against that.
4144 Don't include any symbolic links in the distribution itself. If the tar
4145 file contains symbolic links, then people cannot even unpack it on
4146 systems that don't support symbolic links. Also, don't use multiple
4147 names for one file in different directories, because certain file
4148 systems cannot handle this and that prevents unpacking the
4151 Try to make sure that all the file names will be unique on MS-DOS. A
4152 name on MS-DOS consists of up to 8 characters, optionally followed by a
4153 period and up to three characters. MS-DOS will truncate extra
4154 characters both before and after the period. Thus,
4155 @file{foobarhacker.c} and @file{foobarhacker.o} are not ambiguous; they
4156 are truncated to @file{foobarha.c} and @file{foobarha.o}, which are
4159 @cindex @file{texinfo.tex}, in a distribution
4160 Include in your distribution a copy of the @file{texinfo.tex} you used
4161 to test print any @file{*.texinfo} or @file{*.texi} files.
4163 Likewise, if your program uses small GNU software packages like regex,
4164 getopt, obstack, or termcap, include them in the distribution file.
4165 Leaving them out would make the distribution file a little smaller at
4166 the expense of possible inconvenience to a user who doesn't know what
4170 @chapter References to Non-Free Software and Documentation
4171 @cindex references to non-free material
4173 A GNU program should not recommend, promote, or grant legitimacy to
4174 the use of any non-free program. Proprietary software is a social and
4175 ethical problem, and our aim is to put an end to that problem. We
4176 can't stop some people from writing proprietary programs, or stop
4177 other people from using them, but we can and should refuse to
4178 advertise them to new potential customers, or to give the public the
4179 idea that their existence is ethical.
4181 The GNU definition of free software is found on the GNU web site at
4182 @url{http://www.gnu.org/@/philosophy/@/free-sw.html}, and the definition
4183 of free documentation is found at
4184 @url{http://www.gnu.org/@/philosophy/@/free-doc.html}. The terms ``free''
4185 and ``non-free'', used in this document, refer to those definitions.
4187 A list of important licenses and whether they qualify as free is in
4188 @url{http://www.gnu.org/@/licenses/@/license-list.html}. If it is not
4189 clear whether a license qualifies as free, please ask the GNU Project
4190 by writing to @email{licensing@@gnu.org}. We will answer, and if the
4191 license is an important one, we will add it to the list.
4193 When a non-free program or system is well known, you can mention it in
4194 passing---that is harmless, since users who might want to use it
4195 probably already know about it. For instance, it is fine to explain
4196 how to build your package on top of some widely used non-free
4197 operating system, or how to use it together with some widely used
4200 However, you should give only the necessary information to help those
4201 who already use the non-free program to use your program with
4202 it---don't give, or refer to, any further information about the
4203 proprietary program, and don't imply that the proprietary program
4204 enhances your program, or that its existence is in any way a good
4205 thing. The goal should be that people already using the proprietary
4206 program will get the advice they need about how to use your free
4207 program with it, while people who don't already use the proprietary
4208 program will not see anything likely to lead them to take an interest
4211 If a non-free program or system is obscure in your program's domain,
4212 your program should not mention or support it at all, since doing so
4213 would tend to popularize the non-free program more than it popularizes
4214 your program. (You cannot hope to find many additional users for your
4215 program among the users of Foobar, if the existence of Foobar is not
4216 generally known among people who might want to use your program.)
4218 Sometimes a program is free software in itself but depends on a
4219 non-free platform in order to run. For instance, many Java programs
4220 depend on some non-free Java libraries. To recommend or promote such
4221 a program is to promote the other programs it needs. This is why we
4222 are careful about listing Java programs in the Free Software
4223 Directory: we don't want to promote the non-free Java libraries.
4225 We hope this particular problem with Java will be gone by and by, as
4226 we replace the remaining non-free standard Java libraries with free
4227 software, but the general principle will remain the same: don't
4228 recommend, promote or legitimize programs that depend on non-free
4231 Some free programs strongly encourage the use of non-free software. A
4232 typical example is @command{mplayer}. It is free software in itself,
4233 and the free code can handle some kinds of files. However,
4234 @command{mplayer} recommends use of non-free codecs for other kinds of
4235 files, and users that install @command{mplayer} are very likely to
4236 install those codecs along with it. To recommend @command{mplayer}
4237 is, in effect, to promote use of the non-free codecs.
4239 Thus, you should not recommend programs that strongly encourage the
4240 use of non-free software. This is why we do not list
4241 @command{mplayer} in the Free Software Directory.
4243 A GNU package should not refer the user to any non-free documentation
4244 for free software. Free documentation that can be included in free
4245 operating systems is essential for completing the GNU system, or any
4246 free operating system, so encouraging it is a priority; to recommend
4247 use of documentation that we are not allowed to include undermines the
4248 impetus for the community to produce documentation that we can
4249 include. So GNU packages should never recommend non-free
4252 By contrast, it is ok to refer to journal articles and textbooks in
4253 the comments of a program for explanation of how it functions, even
4254 though they are non-free. This is because we don't include such
4255 things in the GNU system even if they are free---they are outside the
4256 scope of what a software distribution needs to include.
4258 Referring to a web site that describes or recommends a non-free
4259 program is promoting that program, so please do not make links (or
4260 mention by name) web sites that contain such material. This policy is
4261 relevant particularly for the web pages for a GNU package.
4263 Following links from nearly any web site can lead eventually to
4264 non-free software; this is inherent in the nature of the web. So it
4265 makes no sense to criticize a site for having such links. As long as
4266 the site does not itself recommend a non-free program, there is no
4267 need to consider the question of the sites that it links to for other
4270 Thus, for example, you should not refer to AT&T's web site if that
4271 recommends AT&T's non-free software packages; you should not refer to
4272 a site that links to AT&T's site presenting it as a place to get some
4273 non-free program, because that link recommends and legitimizes the
4274 non-free program. However, that a site contains a link to AT&T's web
4275 site for some other purpose (such as long-distance telephone service)
4276 is not an objection against it.
4278 @node GNU Free Documentation License
4279 @appendix GNU Free Documentation License
4281 @cindex FDL, GNU Free Documentation License
4291 eval: (add-hook 'write-file-hooks 'time-stamp)
4292 time-stamp-start: "@set lastupdate "
4294 time-stamp-format: "%:b %:d, %:y"
4295 compile-command: "cd work.s && make"