1 \input texinfo @c -*- texinfo -*-
2 @documentencoding UTF-8
4 @settitle Developer Documentation
6 @center @titlefont{Developer Documentation}
13 @chapter Developers Guide
15 @section Notes for external developers
17 This document is mostly useful for internal FFmpeg developers.
18 External developers who need to use the API in their application should
19 refer to the API doxygen documentation in the public headers, and
20 check the examples in @file{doc/examples} and in the source code to
21 see how the public API is employed.
23 You can use the FFmpeg libraries in your commercial program, but you
24 are encouraged to @emph{publish any patch you make}. In this case the
25 best way to proceed is to send your patches to the ffmpeg-devel
26 mailing list following the guidelines illustrated in the remainder of
29 For more detailed legal information about the use of FFmpeg in
30 external programs read the @file{LICENSE} file in the source tree and
31 consult @url{https://ffmpeg.org/legal.html}.
35 There are 3 ways by which code gets into FFmpeg.
37 @item Submitting patches to the main developer mailing list.
38 See @ref{Submitting patches} for details.
39 @item Directly committing changes to the main tree.
40 @item Committing changes to a git clone, for example on github.com or
41 gitorious.org. And asking us to merge these changes.
44 Whichever way, changes should be reviewed by the maintainer of the code
45 before they are committed. And they should follow the @ref{Coding Rules}.
46 The developer making the commit and the author are responsible for their changes
47 and should try to fix issues their commit causes.
52 @subsection Code formatting conventions
54 There are the following guidelines regarding the indentation in files:
61 The TAB character is forbidden outside of Makefiles as is any
62 form of trailing whitespace. Commits containing either will be
63 rejected by the git repository.
66 You should try to limit your code lines to 80 characters; however, do so if
67 and only if this improves readability.
70 K&R coding style is used.
72 The presentation is one inspired by 'indent -i4 -kr -nut'.
74 The main priority in FFmpeg is simplicity and small code size in order to
75 minimize the bug count.
78 Use the JavaDoc/Doxygen format (see examples below) so that code documentation
79 can be generated automatically. All nontrivial functions should have a comment
80 above them explaining what the function does, even if it is just one sentence.
81 All structures and their member variables should be documented, too.
83 Avoid Qt-style and similar Doxygen syntax with @code{!} in it, i.e. replace
84 @code{//!} with @code{///} and similar. Also @@ syntax should be employed
85 for markup commands, i.e. use @code{@@param} and not @code{\param}.
99 typedef struct Foobar @{
100 int var1; /**< var1 description */
101 int var2; ///< var2 description
102 /** var3 description */
110 * @@param my_parameter description of my_parameter
111 * @@return return value description
113 int myfunc(int my_parameter)
117 @subsection C language features
119 FFmpeg is programmed in the ISO C90 language with a few additional
120 features from ISO C99, namely:
124 the @samp{inline} keyword;
130 designated struct initializers (@samp{struct s x = @{ .i = 17 @};});
133 compound literals (@samp{x = (struct s) @{ 17, 23 @};}).
136 These features are supported by all compilers we care about, so we will not
137 accept patches to remove their use unless they absolutely do not impair
138 clarity and performance.
140 All code must compile with recent versions of GCC and a number of other
141 currently supported compilers. To ensure compatibility, please do not use
142 additional C99 features or GCC extensions. Especially watch out for:
146 mixing statements and declarations;
149 @samp{long long} (use @samp{int64_t} instead);
152 @samp{__attribute__} not protected by @samp{#ifdef __GNUC__} or similar;
155 GCC statement expressions (@samp{(x = (@{ int y = 4; y; @})}).
158 @subsection Naming conventions
159 All names should be composed with underscores (_), not CamelCase. For example,
160 @samp{avfilter_get_video_buffer} is an acceptable function name and
161 @samp{AVFilterGetVideo} is not. The exception from this are type names, like
162 for example structs and enums; they should always be in CamelCase.
164 There are the following conventions for naming variables and functions:
168 For local variables no prefix is required.
171 For file-scope variables and functions declared as @code{static}, no prefix
175 For variables and functions visible outside of file scope, but only used
176 internally by a library, an @code{ff_} prefix should be used,
177 e.g. @samp{ff_w64_demuxer}.
180 For variables and functions visible outside of file scope, used internally
181 across multiple libraries, use @code{avpriv_} as prefix, for example,
182 @samp{avpriv_aac_parse_header}.
185 Each library has its own prefix for public symbols, in addition to the
186 commonly used @code{av_} (@code{avformat_} for libavformat,
187 @code{avcodec_} for libavcodec, @code{swr_} for libswresample, etc).
188 Check the existing code and choose names accordingly.
189 Note that some symbols without these prefixes are also exported for
190 retro-compatibility reasons. These exceptions are declared in the
191 @code{lib<name>/lib<name>.v} files.
194 Furthermore, name space reserved for the system should not be invaded.
195 Identifiers ending in @code{_t} are reserved by
196 @url{http://pubs.opengroup.org/onlinepubs/007904975/functions/xsh_chap02_02.html#tag_02_02_02, POSIX}.
197 Also avoid names starting with @code{__} or @code{_} followed by an uppercase
198 letter as they are reserved by the C standard. Names starting with @code{_}
199 are reserved at the file level and may not be used for externally visible
200 symbols. If in doubt, just avoid names starting with @code{_} altogether.
202 @subsection Miscellaneous conventions
206 fprintf and printf are forbidden in libavformat and libavcodec,
207 please use av_log() instead.
210 Casts should be used only when necessary. Unneeded parentheses
211 should also be avoided if they don't make the code easier to understand.
214 @subsection Editor configuration
215 In order to configure Vim to follow FFmpeg formatting conventions, paste
216 the following snippet into your @file{.vimrc}:
218 " indentation rules for FFmpeg: 4 spaces, no tabs
224 " Allow tabs in Makefiles.
225 autocmd FileType make,automake set noexpandtab shiftwidth=8 softtabstop=8
226 " Trailing whitespace and tabs are forbidden, so highlight them.
227 highlight ForbiddenWhitespace ctermbg=red guibg=red
228 match ForbiddenWhitespace /\s\+$\|\t/
229 " Do not highlight spaces at the end of line while typing on that line.
230 autocmd InsertEnter * match ForbiddenWhitespace /\t\|\s\+\%#\@@<!$/
233 For Emacs, add these roughly equivalent lines to your @file{.emacs.d/init.el}:
235 (c-add-style "ffmpeg"
238 (indent-tabs-mode . nil)
239 (show-trailing-whitespace . t)
241 (statement-cont . (c-lineup-assignments +)))
244 (setq c-default-style "ffmpeg")
247 @section Development Policy
251 Contributions should be licensed under the
252 @uref{http://www.gnu.org/licenses/lgpl-2.1.html, LGPL 2.1},
253 including an "or any later version" clause, or, if you prefer
254 a gift-style license, the
255 @uref{http://opensource.org/licenses/isc-license.txt, ISC} or
256 @uref{http://mit-license.org/, MIT} license.
257 @uref{http://www.gnu.org/licenses/gpl-2.0.html, GPL 2} including
258 an "or any later version" clause is also acceptable, but LGPL is
260 If you add a new file, give it a proper license header. Do not copy and
261 paste it from a random place, use an existing file as template.
264 You must not commit code which breaks FFmpeg! (Meaning unfinished but
265 enabled code which breaks compilation or compiles but does not work or
266 breaks the regression tests)
267 You can commit unfinished stuff (for testing etc), but it must be disabled
268 (#ifdef etc) by default so it does not interfere with other developers'
272 The commit message should have a short first line in the form of
273 a @samp{topic: short description} as a header, separated by a newline
274 from the body consisting of an explanation of why the change is necessary.
275 If the commit fixes a known bug on the bug tracker, the commit message
276 should include its bug ID. Referring to the issue on the bug tracker does
277 not exempt you from writing an excerpt of the bug in the commit message.
280 You do not have to over-test things. If it works for you, and you think it
281 should work for others, then commit. If your code has problems
282 (portability, triggers compiler bugs, unusual environment etc) they will be
283 reported and eventually fixed.
286 Do not commit unrelated changes together, split them into self-contained
287 pieces. Also do not forget that if part B depends on part A, but A does not
288 depend on B, then A can and should be committed first and separate from B.
289 Keeping changes well split into self-contained parts makes reviewing and
290 understanding them on the commit log mailing list easier. This also helps
291 in case of debugging later on.
292 Also if you have doubts about splitting or not splitting, do not hesitate to
293 ask/discuss it on the developer mailing list.
296 Do not change behavior of the programs (renaming options etc) or public
297 API or ABI without first discussing it on the ffmpeg-devel mailing list.
298 Do not remove functionality from the code. Just improve!
300 Note: Redundant code can be removed.
303 Do not commit changes to the build system (Makefiles, configure script)
304 which change behavior, defaults etc, without asking first. The same
305 applies to compiler warning fixes, trivial looking fixes and to code
306 maintained by other developers. We usually have a reason for doing things
307 the way we do. Send your changes as patches to the ffmpeg-devel mailing
308 list, and if the code maintainers say OK, you may commit. This does not
309 apply to files you wrote and/or maintain.
312 We refuse source indentation and other cosmetic changes if they are mixed
313 with functional changes, such commits will be rejected and removed. Every
314 developer has his own indentation style, you should not change it. Of course
315 if you (re)write something, you can use your own style, even though we would
316 prefer if the indentation throughout FFmpeg was consistent (Many projects
317 force a given indentation style - we do not.). If you really need to make
318 indentation changes (try to avoid this), separate them strictly from real
321 NOTE: If you had to put if()@{ .. @} over a large (> 5 lines) chunk of code,
322 then either do NOT change the indentation of the inner part within (do not
323 move it to the right)! or do so in a separate commit
326 Always fill out the commit log message. Describe in a few lines what you
327 changed and why. You can refer to mailing list postings if you fix a
328 particular bug. Comments such as "fixed!" or "Changed it." are unacceptable.
332 area changed: Short 1 line description
334 details describing what and why and giving references.
338 Make sure the author of the commit is set correctly. (see git commit --author)
339 If you apply a patch, send an
340 answer to ffmpeg-devel (or wherever you got the patch from) saying that
341 you applied the patch.
344 When applying patches that have been discussed (at length) on the mailing
345 list, reference the thread in the log message.
348 Do NOT commit to code actively maintained by others without permission.
349 Send a patch to ffmpeg-devel instead. If no one answers within a reasonable
350 timeframe (12h for build failures and security fixes, 3 days small changes,
351 1 week for big patches) then commit your patch if you think it is OK.
352 Also note, the maintainer can simply ask for more time to review!
355 Subscribe to the ffmpeg-cvslog mailing list. The diffs of all commits
356 are sent there and reviewed by all the other developers. Bugs and possible
357 improvements or general questions regarding commits are discussed there. We
358 expect you to react if problems with your code are uncovered.
361 Update the documentation if you change behavior or add features. If you are
362 unsure how best to do this, send a patch to ffmpeg-devel, the documentation
363 maintainer(s) will review and commit your stuff.
366 Try to keep important discussions and requests (also) on the public
367 developer mailing list, so that all developers can benefit from them.
370 Never write to unallocated memory, never write over the end of arrays,
371 always check values read from some untrusted source before using them
372 as array index or other risky things.
375 Remember to check if you need to bump versions for the specific libav*
376 parts (libavutil, libavcodec, libavformat) you are changing. You need
377 to change the version integer.
378 Incrementing the first component means no backward compatibility to
379 previous versions (e.g. removal of a function from the public API).
380 Incrementing the second component means backward compatible change
381 (e.g. addition of a function to the public API or extension of an
382 existing data structure).
383 Incrementing the third component means a noteworthy binary compatible
384 change (e.g. encoder bug fix that matters for the decoder). The third
385 component always starts at 100 to distinguish FFmpeg from Libav.
388 Compiler warnings indicate potential bugs or code with bad style. If a type of
389 warning always points to correct and clean code, that warning should
390 be disabled, not the code changed.
391 Thus the remaining warnings can either be bugs or correct code.
392 If it is a bug, the bug has to be fixed. If it is not, the code should
393 be changed to not generate a warning unless that causes a slowdown
394 or obfuscates the code.
397 Make sure that no parts of the codebase that you maintain are missing from the
398 @file{MAINTAINERS} file. If something that you want to maintain is missing add it with
400 If at some point you no longer want to maintain some code, then please help in
401 finding a new maintainer and also don't forget to update the @file{MAINTAINERS} file.
404 We think our rules are not too hard. If you have comments, contact us.
406 @anchor{Submitting patches}
407 @section Submitting patches
409 First, read the @ref{Coding Rules} above if you did not yet, in particular
410 the rules regarding patch submission.
412 When you submit your patch, please use @code{git format-patch} or
413 @code{git send-email}. We cannot read other diffs :-).
415 Also please do not submit a patch which contains several unrelated changes.
416 Split it into separate, self-contained pieces. This does not mean splitting
417 file by file. Instead, make the patch as small as possible while still
418 keeping it as a logical unit that contains an individual change, even
419 if it spans multiple files. This makes reviewing your patches much easier
420 for us and greatly increases your chances of getting your patch applied.
422 Use the patcheck tool of FFmpeg to check your patch.
423 The tool is located in the tools directory.
425 Run the @ref{Regression tests} before submitting a patch in order to verify
426 it does not cause unexpected problems.
428 It also helps quite a bit if you tell us what the patch does (for example
429 'replaces lrint by lrintf'), and why (for example '*BSD isn't C99 compliant
432 Also please if you send several patches, send each patch as a separate mail,
433 do not attach several unrelated patches to the same mail.
435 Patches should be posted to the
436 @uref{https://lists.ffmpeg.org/mailman/listinfo/ffmpeg-devel, ffmpeg-devel}
437 mailing list. Use @code{git send-email} when possible since it will properly
438 send patches without requiring extra care. If you cannot, then send patches
439 as base64-encoded attachments, so your patch is not trashed during
442 Your patch will be reviewed on the mailing list. You will likely be asked
443 to make some changes and are expected to send in an improved version that
444 incorporates the requests from the review. This process may go through
445 several iterations. Once your patch is deemed good enough, some developer
446 will pick it up and commit it to the official FFmpeg tree.
448 Give us a few days to react. But if some time passes without reaction,
449 send a reminder by email. Your patch should eventually be dealt with.
452 @section New codecs or formats checklist
456 Did you use av_cold for codec initialization and close functions?
459 Did you add a long_name under NULL_IF_CONFIG_SMALL to the AVCodec or
460 AVInputFormat/AVOutputFormat struct?
463 Did you bump the minor version number (and reset the micro version
464 number) in @file{libavcodec/version.h} or @file{libavformat/version.h}?
467 Did you register it in @file{allcodecs.c} or @file{allformats.c}?
470 Did you add the AVCodecID to @file{avcodec.h}?
471 When adding new codec IDs, also add an entry to the codec descriptor
472 list in @file{libavcodec/codec_desc.c}.
475 If it has a FourCC, did you add it to @file{libavformat/riff.c},
476 even if it is only a decoder?
479 Did you add a rule to compile the appropriate files in the Makefile?
480 Remember to do this even if you're just adding a format to a file that is
481 already being compiled by some other rule, like a raw demuxer.
484 Did you add an entry to the table of supported formats or codecs in
485 @file{doc/general.texi}?
488 Did you add an entry in the Changelog?
491 If it depends on a parser or a library, did you add that dependency in
495 Did you @code{git add} the appropriate files before committing?
498 Did you make sure it compiles standalone, i.e. with
499 @code{configure --disable-everything --enable-decoder=foo}
500 (or @code{--enable-demuxer} or whatever your component is)?
504 @section patch submission checklist
508 Does @code{make fate} pass with the patch applied?
511 Was the patch generated with git format-patch or send-email?
514 Did you sign off your patch? (git commit -s)
515 See @url{http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=blob_plain;f=Documentation/SubmittingPatches} for the meaning
519 Did you provide a clear git commit log message?
522 Is the patch against latest FFmpeg git master branch?
525 Are you subscribed to ffmpeg-devel?
526 (the list is subscribers only due to spam)
529 Have you checked that the changes are minimal, so that the same cannot be
530 achieved with a smaller patch and/or simpler final code?
533 If the change is to speed critical code, did you benchmark it?
536 If you did any benchmarks, did you provide them in the mail?
539 Have you checked that the patch does not introduce buffer overflows or
540 other security issues?
543 Did you test your decoder or demuxer against damaged data? If no, see
544 tools/trasher, the noise bitstream filter, and
545 @uref{http://caca.zoy.org/wiki/zzuf, zzuf}. Your decoder or demuxer
546 should not crash, end in a (near) infinite loop, or allocate ridiculous
547 amounts of memory when fed damaged data.
550 Did you test your decoder or demuxer against sample files?
551 Samples may be obtained at @url{https://samples.ffmpeg.org}.
554 Does the patch not mix functional and cosmetic changes?
557 Did you add tabs or trailing whitespace to the code? Both are forbidden.
560 Is the patch attached to the email you send?
563 Is the mime type of the patch correct? It should be text/x-diff or
564 text/x-patch or at least text/plain and not application/octet-stream.
567 If the patch fixes a bug, did you provide a verbose analysis of the bug?
570 If the patch fixes a bug, did you provide enough information, including
571 a sample, so the bug can be reproduced and the fix can be verified?
572 Note please do not attach samples >100k to mails but rather provide a
573 URL, you can upload to ftp://upload.ffmpeg.org.
576 Did you provide a verbose summary about what the patch does change?
579 Did you provide a verbose explanation why it changes things like it does?
582 Did you provide a verbose summary of the user visible advantages and
583 disadvantages if the patch is applied?
586 Did you provide an example so we can verify the new feature added by the
590 If you added a new file, did you insert a license header? It should be
591 taken from FFmpeg, not randomly copied and pasted from somewhere else.
594 You should maintain alphabetical order in alphabetically ordered lists as
595 long as doing so does not break API/ABI compatibility.
598 Lines with similar content should be aligned vertically when doing so
599 improves readability.
602 Consider adding a regression test for your code.
605 If you added YASM code please check that things still work with --disable-yasm.
608 Make sure you check the return values of function and return appropriate
609 error codes. Especially memory allocation functions like @code{av_malloc()}
610 are notoriously left unchecked, which is a serious problem.
613 Test your code with valgrind and or Address Sanitizer to ensure it's free
614 of leaks, out of array accesses, etc.
617 @section Patch review process
619 All patches posted to ffmpeg-devel will be reviewed, unless they contain a
620 clear note that the patch is not for the git master branch.
621 Reviews and comments will be posted as replies to the patch on the
622 mailing list. The patch submitter then has to take care of every comment,
623 that can be by resubmitting a changed patch or by discussion. Resubmitted
624 patches will themselves be reviewed like any other patch. If at some point
625 a patch passes review with no comments then it is approved, that can for
626 simple and small patches happen immediately while large patches will generally
627 have to be changed and reviewed many times before they are approved.
628 After a patch is approved it will be committed to the repository.
630 We will review all submitted patches, but sometimes we are quite busy so
631 especially for large patches this can take several weeks.
633 If you feel that the review process is too slow and you are willing to try to
634 take over maintainership of the area of code you change then just clone
635 git master and maintain the area of code there. We will merge each area from
636 where its best maintained.
638 When resubmitting patches, please do not make any significant changes
639 not related to the comments received during review. Such patches will
640 be rejected. Instead, submit significant changes or new features as
643 Everyone is welcome to review patches. Also if you are waiting for your patch
644 to be reviewed, please consider helping to review other patches, that is a great
645 way to get everyone's patches reviewed sooner.
647 @anchor{Regression tests}
648 @section Regression tests
650 Before submitting a patch (or committing to the repository), you should at least
651 test that you did not break anything.
653 Running 'make fate' accomplishes this, please see @url{fate.html} for details.
655 [Of course, some patches may change the results of the regression tests. In
656 this case, the reference results of the regression tests shall be modified
659 @subsection Adding files to the fate-suite dataset
661 When there is no muxer or encoder available to generate test media for a
662 specific test then the media has to be included in the fate-suite.
663 First please make sure that the sample file is as small as possible to test the
664 respective decoder or demuxer sufficiently. Large files increase network
665 bandwidth and disk space requirements.
666 Once you have a working fate test and fate sample, provide in the commit
667 message or introductory message for the patch series that you post to
668 the ffmpeg-devel mailing list, a direct link to download the sample media.
670 @subsection Visualizing Test Coverage
672 The FFmpeg build system allows visualizing the test coverage in an easy
673 manner with the coverage tools @code{gcov}/@code{lcov}. This involves
678 Configure to compile with instrumentation enabled:
679 @code{configure --toolchain=gcov}.
682 Run your test case, either manually or via FATE. This can be either
683 the full FATE regression suite, or any arbitrary invocation of any
684 front-end tool provided by FFmpeg, in any combination.
687 Run @code{make lcov} to generate coverage data in HTML format.
690 View @code{lcov/index.html} in your preferred HTML viewer.
693 You can use the command @code{make lcov-reset} to reset the coverage
694 measurements. You will need to rerun @code{make lcov} after running a
697 @subsection Using Valgrind
699 The configure script provides a shortcut for using valgrind to spot bugs
700 related to memory handling. Just add the option
701 @code{--toolchain=valgrind-memcheck} or @code{--toolchain=valgrind-massif}
702 to your configure line, and reasonable defaults will be set for running
703 FATE under the supervision of either the @strong{memcheck} or the
704 @strong{massif} tool of the valgrind suite.
706 In case you need finer control over how valgrind is invoked, use the
707 @code{--target-exec='valgrind <your_custom_valgrind_options>} option in
708 your configure line instead.
710 @anchor{Release process}
711 @section Release process
713 FFmpeg maintains a set of @strong{release branches}, which are the
714 recommended deliverable for system integrators and distributors (such as
715 Linux distributions, etc.). At regular times, a @strong{release
716 manager} prepares, tests and publishes tarballs on the
717 @url{https://ffmpeg.org} website.
719 There are two kinds of releases:
723 @strong{Major releases} always include the latest and greatest
724 features and functionality.
727 @strong{Point releases} are cut from @strong{release} branches,
728 which are named @code{release/X}, with @code{X} being the release
732 Note that we promise to our users that shared libraries from any FFmpeg
733 release never break programs that have been @strong{compiled} against
734 previous versions of @strong{the same release series} in any case!
736 However, from time to time, we do make API changes that require adaptations
737 in applications. Such changes are only allowed in (new) major releases and
738 require further steps such as bumping library version numbers and/or
739 adjustments to the symbol versioning file. Please discuss such changes
740 on the @strong{ffmpeg-devel} mailing list in time to allow forward planning.
742 @anchor{Criteria for Point Releases}
743 @subsection Criteria for Point Releases
745 Changes that match the following criteria are valid candidates for
746 inclusion into a point release:
750 Fixes a security issue, preferably identified by a @strong{CVE
751 number} issued by @url{http://cve.mitre.org/}.
754 Fixes a documented bug in @url{https://trac.ffmpeg.org}.
757 Improves the included documentation.
760 Retains both source code and binary compatibility with previous
761 point releases of the same release branch.
764 The order for checking the rules is (1 OR 2 OR 3) AND 4.
767 @subsection Release Checklist
769 The release process involves the following steps:
773 Ensure that the @file{RELEASE} file contains the version number for
774 the upcoming release.
777 Add the release at @url{https://trac.ffmpeg.org/admin/ticket/versions}.
780 Announce the intent to do a release to the mailing list.
783 Make sure all relevant security fixes have been backported. See
784 @url{https://ffmpeg.org/security.html}.
787 Ensure that the FATE regression suite still passes in the release
788 branch on at least @strong{i386} and @strong{amd64}
789 (cf. @ref{Regression tests}).
792 Prepare the release tarballs in @code{bz2} and @code{gz} formats, and
793 supplementing files that contain @code{gpg} signatures
796 Publish the tarballs at @url{https://ffmpeg.org/releases}. Create and
797 push an annotated tag in the form @code{nX}, with @code{X}
798 containing the version number.
801 Propose and send a patch to the @strong{ffmpeg-devel} mailing list
802 with a news entry for the website.
805 Publish the news entry.
808 Send an announcement to the mailing list.