[Ffmpeg-devel-irc] ffmpeg-devel.log.20120622

burek burek021 at gmail.com
Sat Jun 23 02:05:02 CEST 2012


[00:03] <RobertNagy> why does ff_default_get_video_buffer call memset(data[0], 128, i);?
[00:04] <durandal_1707> to maky it gray?
[00:04] <RobertNagy> and why does it do that?
[00:08] <durandal_1707> it is added by michael when merging qatar code
[00:08] <RobertNagy> qatar code?
[00:09] <durandal_1707> libav
[00:09] <RobertNagy> ic
[00:23] <Daemon404> i dont understand how ami_stuff finds all these weird things
[00:24] <durandal_1707> he have bunch of free time
[00:28] <iive> RobertNagy: U/V planes are signed, so this value makes the image color neutral.
[00:28] <iive> using 0, would make it green.
[00:29] <iive> well, not signed, but with offset
[00:36] <llogan> what's the link to Herve's redesigned FFmpeg logo?
[00:40] <RobertNagy> iive: But why make it neutral at all?
[00:40] <RobertNagy> why not just leave it?
[00:40] <RobertNagy> and let the decoder fill it
[00:47] <llogan> llogan: here you go: http://www.movieconverter-studio.com/_PUBLIC/ffmpeg/logo-new/ffmpeg-logo-src/
[00:50] <Daemon404> llogan, btw: https://github.com/dwbuiten/utvideo
[00:50] <Daemon404> may be relevant to you (i recall you building libutvideo)
[00:52] <llogan> Daemon404: thanks. are your patches often accepted by Takeshi?
[00:52] <Daemon404> parts of it for sure
[00:52] <Daemon404> tehred a windows fix in one part
[00:52] <Daemon404> and teh x86 asm works everywhere
[00:52] <Daemon404> x64 asm is bugged on non-windows
[00:52] <Daemon404> and i suck at debugging asm
[00:53] <Daemon404> esp. when gdb tells me teh error happened on the line where a nasm macro was defined
[01:02] <iive> RobertNagy: well, it could be security risk if memory is not cleared. Neutral image is also better than random junk, if part of the frames end up been displayed.
[01:03] <RobertNagy> is there any case where parts of the frame should be possibly displayed?
[01:03] <iive> mainly errors in the bitstream could cause gaps. also there are cases where decoding starts from prediction frame not keyframe.
[01:04] <iive> aka, error resilience.
[01:06] <llogan> https://ffmpeg.org/trac/ffmpeg/wiki/SubmitALogo
[01:07] <llogan> modify/mutilate/bikeshed as you see fit (especially License section)
[01:17] <llogan> idiot. i keep s/logo/logan
[01:23] <Daemon404> humm... is it possible for c++ and c symbols to share the same lib??
[01:33] <Daemon404> i see the problem.. an invalid def is being generated
[01:36] <teratorn> Daemon404: god i hope not :(
[01:36] <teratorn> you can link to c symbols from c++
[01:36] <teratorn> if you want to live in hell
[02:04] <CIA-119> ffmpeg: 03Lou Logan 07master * rcdb94139cc 10ffmpeg/doc/faq.texi: 
[02:04] <CIA-119> ffmpeg: doc/faq: Update video joining examples
[02:04] <CIA-119> ffmpeg: -same_quant is not designed to convert between quantizer scales, AFAIK.
[02:04] <CIA-119> ffmpeg: Add example using concat protocol.
[02:04] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:04] <CIA-119> ffmpeg: 03Hendrik Leppkes 07master * r5916313967 10ffmpeg/libavformat/utils.c: 
[02:04] <CIA-119> ffmpeg: avformat: offset fps calculation by one to include 60.00 fps
[02:04] <CIA-119> ffmpeg: To ensure the full range of values is still used, also adjust all uses of this function to loop from 0
[02:04] <CIA-119> ffmpeg: instead of 1. This way only 60.00 is added and nothing lost.
[02:04] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:09] <durandal_1707> hmm, why snow supports only 420 yuv?
[02:16] <CIA-119> ffmpeg: 03Paul B Mahol 07master * r4cdff58206 10ffmpeg/tests/ (fate/audio.mak fate/video.mak ref/fate/sanm ref/fate/vima): 
[02:16] <CIA-119> ffmpeg: fate: add tests for SANM and VIMA
[02:16] <CIA-119> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[03:12] <durandal_1707> libavfilter should really have generic options so that ffmpeg -help can show them
[03:13] <durandal_1707> michaelni: libswsresample does not provide options like libavresample
[03:13] <durandal_1707> though avconv doesn't use them
[03:26] <michaelni> durandal_1707, swresample has options and they worked via ffmpeg command line
[03:27] <durandal_1707> but are not shown with -help
[03:29] <durandal_1707> libavresample have api to access AVClass...
[03:29] <durandal_1707> but avconv is not using it at all
[03:44] <durandal_1707> michaelni: qatar is going to bump mayor for lavfi anyway so it may be good time to add priv_class to AVFilter?
[03:46] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r2fa5d8f234 10ffmpeg/ffmpeg.c: 
[03:46] <CIA-119> ffmpeg: ffmpeg: print swr options in --help too
[03:46] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:00] <durandal_1707> michaelni: now do same for lavfi :evil
[04:03] <durandal_1707> michaelni: libavresample have some stuff (matrix_encoding) do you plan to port that to swr?
[04:06] <michaelni> matrix_encoding ?
[04:06] <durandal_1707> libavresample/options.c
[04:07] <michaelni> ahh the dolby stuff
[04:07] <michaelni> yes i already had planed to port this
[04:07] <michaelni> its trivial
[04:11] <durandal_1707> crap, i cant get lavfi options to show up i keep getting segv
[04:12] <michaelni> about snow, there is some code (and not so little) that assumes h == v subsampling
[04:12] <michaelni> so i might get 444 and gray working but 422 seems harder
[04:36] <CIA-119> ffmpeg: 03Paul B Mahol 07master * r81416c2f64 10ffmpeg/libswscale/options.c: 
[04:36] <CIA-119> ffmpeg: swscale: use designated initializers for AVClass
[04:36] <CIA-119> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[04:53] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rb4b7471ec4 10ffmpeg/libavcodec/snowenc.c: 
[04:53] <CIA-119> ffmpeg: snowenc: set chroma subsample factors correctly for non 420
[04:53] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:53] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r13ba872a2a 10ffmpeg/libavcodec/snowenc.c: 
[04:53] <CIA-119> ffmpeg: snowenc: update pix_sum for rectangles, needed for non 420
[04:53] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[12:26] <saste> ouch, git docs are even worse than ffmpeg's
[12:29] <iive> at least git cli is better than ffmpeg's, but not by much.
[12:36] <cbsrobot> I always thought the gim man pages are exlusively on stackoverflow
[12:36] <cbsrobot> *git
[12:59] <Compn> iive : yes, i think having multiple seperate binaries for each git command is brilliant...
[12:59] <Compn> or maybe i'm thinking of something else
[12:59] <Compn> nevermind
[15:01] <ubitux> saste: any plans on the audio testing?
[15:02] <saste> ubitux: not yet
[15:02] <ubitux> ok
[15:02] <ubitux> it would be nice to have the test in fate/lavfi.mak if possible
[15:02] <ubitux> instead of the script
[15:03] <saste> http://gitorious.org/~saste/ffmpeg/sastes-ffmpeg/commit/a14986d48c3daef018be059c0322434e82e36499
[15:04] <saste> this is my ancient patch, don't know if it still works (most likely not)
[15:04] <saste> it was relying on the possibility of having an explicit list of the supported channel layouts
[15:08] <saste> bbl, bye
[15:08] <ubitux> bye :)
[15:35] <ubitux> http://documentation.apple.com/en/dvdstudiopro/usermanual/index.html#chapter=19%26section=13%26tasks=true
[15:35] <ubitux> anyone has some samples for this?
[15:50] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r68c845cdba 10ffmpeg/libavcodec/ (snow.c snow.h snowdec.c snowenc.c): 
[15:50] <CIA-119> ffmpeg: snow: yuv444 support
[15:50] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:02] <CIA-119> ffmpeg: 03Paul B Mahol 07master * r10d2a4028c 10ffmpeg/libswresample/swresample.c: 
[16:02] <CIA-119> ffmpeg: swresample: make .class_name more descriptive
[16:02] <CIA-119> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[16:50] <CIA-119> ffmpeg: 03Carl Eugen Hoyos 07master * r15bde06084 10ffmpeg/libavutil/mathematics.h: 
[16:50] <CIA-119> ffmpeg: Revert "mathematics.h: remove a couple of math defines"
[16:50] <CIA-119> ffmpeg: This reverts commit a2b186a188c126dd0402414d9e8198ec4df63ebe.
[16:50] <CIA-119> ffmpeg: The defines were present for several years, removing
[16:50] <CIA-119> ffmpeg: them without a version bump may break API.
[16:50] <CIA-119> ffmpeg: (Fixes MPlayer compilation.)
[17:14] <Daemon404> who's the build system expert here?
[17:14] <Daemon404> particularily the mingw side
[17:20] <michaelni> Daemon404, zeranoe or ramiro should know about mingw
[17:21] <Daemon404> ok
[17:21] <Daemon404> neither is ever really on irc... lame
[17:22] Action: Daemon404 guesses he uses mail
[17:25] <ubitux> seems this will keep me busy a while: http://tech.ebu.ch/ebu-tt
[17:25] <ubitux> yet another subtitle standard T_T
[17:26] <ubitux> we are lucky it seems they didn't implement karaoke this time
[17:29] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rfba18ef8cc 10ffmpeg/libavcodec/x86/dsputil_mmx.c: 
[17:29] <CIA-119> ffmpeg: x86/dsputil_mmx: support 4 sample edges
[17:29] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:29] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r1e7943c315 10ffmpeg/libavcodec/ (snowdec.c snowenc.c): 
[17:29] <CIA-119> ffmpeg: snow: yuv410 support
[17:29] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:56] <cbsrobot> ubitux: meaning it will be added at the end of your todo list ?
[17:56] <ubitux> possibly
[18:31] <Daemon404> nevcairiel, http://forum.doom9.org/showpost.php?p=1579296&postcount=11381
[18:31] <Daemon404> wanna bet that is due to frame multithreading + null frames?
[19:18] <CIA-119> ffmpeg: 03Wolfram Gloger 07master * r258dfff839 10ffmpeg/libavutil/log.c: 
[19:18] <CIA-119> ffmpeg: extend log line buffer sizes
[19:18] <CIA-119> ffmpeg: The string passed to av_vlog() from libx264's X264_log overflows the 512
[19:18] <CIA-119> ffmpeg: byte part_size, truncating x264 parameter output. Double this size
[19:18] <CIA-119> ffmpeg: and put it in a #define so it only needs to be changed in one place.
[19:18] <CIA-119> ffmpeg: Signed-off-by: Wolfram Gloger <video06 at malloc.de>
[19:18] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:21] <nevcairiel> Daemon404: i looked at the lagarith decoder, and i couldnt find any evidence of such null frames
[19:21] <nevcairiel> but maybe i'm blind
[19:24] <Daemon404> well theyre a thing
[19:24] <Daemon404> maybe we dont implement them
[19:24] <Daemon404> let me look.
[19:24] <nevcairiel> the decoder is fairly simple, i doubt i missed it
[19:25] <nevcairiel> in any case, windows thumbnail generation is fragile on its best days, plain broken on others, so i wouldn't rule out failure on that end :p
[19:25] <Daemon404> nevcairiel, i disbale
[19:25] <Daemon404> cause ms's own decoderts make OTHER APPS crash when they have an open dialogue to a folder with thumbnails
[19:25] <Daemon404> sometimes
[19:25] <Daemon404> its insanely annoying.
[19:29] <Daemon404>     // according to the avi specs, the calling application is responsible for handling null frames.
[19:29] <Daemon404>     if ( compressed_size == 0 ){
[19:29] <Daemon404>         #ifndef LAGARITH_RELEASE
[19:29] <Daemon404>         MessageBox (HWND_DESKTOP, "Received request to decode a null frame", "Error", MB_OK | MB_ICONEXCLAMATION);
[19:30] <Daemon404>         #endif
[19:30] <Daemon404>         return ICERR_OK;
[19:30] <Daemon404>     }
[19:30] <Daemon404> well there u go
[19:30] <nevcairiel> oh yeah i think avformat doesnt send null frames out
[19:30] <nevcairiel> if its really size 0 without anything else
[19:30] <nevcairiel> in any case, nothing in the avcodec decoder
[19:32] <Daemon404> yeah
[19:34] <aquarat> is H.264 MVC (3D) video supported in ffmpeg or being worked on ?
[19:34] <aquarat> and hello
[19:34] <nevcairiel> no, and not to my knowledge
[19:35] <nevcairiel> of course you never know what some developer is cooking up in his private repo somewhere
[19:35] <aquarat> should I create a request for it ?
[19:35] <aquarat> and upload samples
[19:35] <aquarat> lol @ cooking
[19:36] <michaelni> samples are welcome if there are none yet
[19:36] <michaelni> a patch would be great too ;)
[19:37] <aquarat> I only really know Java :P
[19:41] <JEEB> http://c.learncodethehardway.org/
[19:42] <aquarat> mmm
[19:44] <aquarat> looks cool
[19:44] <aquarat> and it's free
[19:44] <aquarat> I'll read it ;)
[19:44] <aquarat> thanks
[19:46] <JEEB> no problem, also you might want to start looking at some of the simpler video decoders and possibly format parsers
[19:47] <ramiro> hi
[19:48] <Daemon404> hi
[19:48] <aquarat> I've played a bit with wav decoding
[19:48] <aquarat> and I wrote an application in Java that generated sine waves modulated with time for my watch
[19:49] <aquarat> I live in a country that doesn't have a long wave time transmitter
[19:50] <Daemon404> ramiro, you are fimiliar with the mingw aspect of teh build system, right?
[19:50] <ramiro> yep
[19:50] <Daemon404> how do we generate our .def files?
[19:50] <ramiro> well i was until early last year
[19:51] <ramiro> gcc generates them
[19:51] <Daemon404> hmm shit.
[19:51] <Daemon404> when i link libutvideo, which has c++ mangled symbols, teh def files only contains the mangled c++ symbols (all of 2!) and ignores the rest of the normal C ones
[19:51] <Daemon404> even though theyre all present, and exported from the dll
[19:51] <Daemon404> i dont really know how to go about fixing this
[19:52] <ramiro> the line that configures to make the def file is in configure: SHFLAGS='-shared -Wl,--output-def,$$(@:$(SLIBSUF)=.def) [...]
[19:53] <ramiro> I don't know how it works for c++. so this is a libutvideo issue, not ffmpeg, right?
[19:57] <Daemon404> ramiro, no
[19:57] <Daemon404> it's a ffmpeg issue
[19:57] <Daemon404> it hapopens with any c++ 
[19:57] <Daemon404> like libstagefright
[19:59] <ramiro> hm, let me see. if you link against any c++ library, the ffmpeg dlls export c++ symbols only, is that right?
[20:00] <Daemon404> ramiro, no
[20:00] <Daemon404> teh dlls themselevs export teh c symbols too
[20:00] <Daemon404> at least when i checked
[20:00] <Daemon404> the .def and teh .dll.a onyl have teh c++ symbols
[20:00] <ramiro> ah, ok
[20:01] <ramiro> i remember digging into binutils' source code and the function that created the .def file was really stupid
[20:04] <Daemon404> probably
[20:05] <ramiro> http://sourceware.org/cgi-bin/cvsweb.cgi/src/binutils/dlltool.c?rev=1.106&content-type=text/x-cvsweb-markup&cvsroot=src
[20:05] <ramiro> gen_def_file
[20:08] <Daemon404> right... binutils sitll uses cvs
[20:08] <ramiro> heh
[20:09] <ramiro> it does cplus_demangle(). I have no idea what that does, but it looks like it should demangle the symbol name
[20:09] <ramiro> can you pastebin the output you get?
[20:09] <Daemon404> the resulting .def?
[20:14] <ramiro> yes
[20:17] <Daemon404> let me rerun it
[20:41] <Daemon404> ramiro, https://dl.dropbox.com/u/77349181/avcodec-54.def
[20:42] <ramiro> that's it?
[20:42] <Daemon404> yup.
[20:42] <Daemon404> those are the 2 c++ symbols i called through libavcodec*.cpp
[20:43] <Daemon404> i think
[20:43] <ramiro> that's odd
[20:43] <ramiro> is this in the ffmpeg tree? or some code you're working on?
[20:44] <ramiro> i don't know what libutvideo, but if it's a lib that you can compile into ffmpeg then i guess zeranoe has fixed whatever issue happens with it
[20:44] <Daemon404> tthats ffmpeg tree
[20:44] <Daemon404> ramiro, iy only happens with --enable-shared --disable-static
[20:45] <Daemon404> https://dl.dropbox.com/u/77349181/dll_nm.txt
[21:11] <ramiro> Daemon404: maybe ask the mingw-w64 guys? they know gcc inside out
[21:11] <Daemon404> me and them dont see eye-to-eye a lot
[21:11] <Daemon404> lets just put it that way
[21:12] <ramiro> hahaha
[21:12] <ramiro> too bad. i think they're nice people
[21:13] <ramiro> then ask zeranoe, he might know
[21:13] <Daemon404> the yhave communication issues (lack of native english speakers)
[21:13] <Daemon404> ill ask zeranoe maybe
[21:25] <ubitux> michaelni: is it me or AV_QSORT comparison is inverted?
[21:26] <ubitux> it looks like it is sorting the exact opposite of qsort()
[21:36] <ubitux> [libav-devel] [PATCH 01/11] lavfi: remove 'opaque' parameter from AVFilter.init() 
[21:36] <ubitux> ARGH
[21:36] <ubitux> no shit niggah...
[21:36] <ubitux> that could be actually very useful...
[21:37] <ubitux> it might be better to propose a buffersrc & buffersink struct instead of asking for the user to build a string.
[21:53] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rc8f4c83e70 10ffmpeg/libavcodec/ituh263dec.c: 
[21:53] <CIA-119> ffmpeg: h.263dec: use av_assert
[21:53] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:54] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r8aeed14459 10ffmpeg/libavcodec/ituh263enc.c: 
[21:54] <CIA-119> ffmpeg: h.263enc: use av_assert
[21:54] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:06] <ubitux> michaelni: btw, subtitle_out in ffmpeg.c is leaking
[22:06] <ubitux> any idea how this should be done?
[22:07] <ubitux> fate will become yellow if i push the subtitles test suite
[22:10] <michaelni> ubitux, what is leaking exactly ?
[22:11] <michaelni> ahh i see :)
[22:12] <michaelni> just needs a av_freep() in exit_program() 
[22:12] <ubitux> we could put the buffer globally and free it at the end but well, that is rather silly
[22:12] <michaelni> what is silly on that ?
[22:12] <ubitux> i think it should be global, but that's more like an api issue 
[22:12] <ubitux> should *not*
[22:13] <ubitux> well whatever.
[22:13] <CIA-119> ffmpeg: 03Clément BSsch 07master * r492cc9bcc4 10ffmpeg/ffmpeg.c: ffmpeg: remove unused buffer_size in InputFile.
[22:26] <ubitux> michaelni: did you see my AV_SORT question?
[22:26] <michaelni> it surely could be fliped, i didnt check
[22:26] <michaelni> feel free to flip it if you checked ....
[22:26] <ubitux> well i thought you tested your code :D
[22:27] <michaelni> i did test it :)
[22:27] <ubitux> so maybe it was wanted :)
[22:27] <michaelni> and i benchmatked against qsort
[22:27] <michaelni> but maybe i got the order wrong 
[22:28] <durandal_1707> michaelni: do you know args to encode lossless with snow?
[22:30] <michaelni> durandal_1707,  i think: -qscale .001 -pred 1
[22:31] <michaelni> ubitux, my test had: return (*sa > *sb) - (*sa < *sb); vs. CMP(a,b) (((b)[0] > (a)[0]) - ((b)[0] < (a)[0]))
[22:31] <ubitux> i just tried to use AV_QSORT() in place of the qsort() in my subtitles branch
[22:31] <ubitux> and the subtitles were inverted
[22:32] <ubitux> maybe i did something wrong though
[22:32] <ubitux> 'going to check again
[22:36] <ubitux> mmh i'm not able to make it work properly
[22:36] <CIA-119> ffmpeg: 03Mans Rullgard 07master * rc19e0ebe53 10ffmpeg/libavformat/avformat.h: 
[22:36] <CIA-119> ffmpeg: lavf: include libavutil/time.h instead of redeclaring av_gettime()
[22:36] <CIA-119> ffmpeg: This avoids some warnings about redundant declarations.
[22:36] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[22:36] <CIA-119> ffmpeg: 03Mans Rullgard 07master * r539df61193 10ffmpeg/libavutil/log.c: 
[22:36] <CIA-119> ffmpeg: log: include unistd.h only when needed
[22:36] <CIA-119> ffmpeg: The only symbol this file uses from unistd.h is isatty(). By
[22:36] <CIA-119> ffmpeg: including the header only when this function is used, the file
[22:36] <CIA-119> ffmpeg: can be built on systems without unistd.h (which presumably also
[22:36] <CIA-119> ffmpeg: lack isatty).
[22:36] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[22:36] <CIA-119> ffmpeg: 03Mans Rullgard 07master * r542920b7f6 10ffmpeg/configure: 
[22:36] <CIA-119> ffmpeg: configure: check for unistd.h
[22:36] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[22:36] <CIA-119> ffmpeg: 03Mans Rullgard 07master * rd11baad055 10ffmpeg/libavutil/file.c: 
[22:36] <CIA-119> ffmpeg: lavu/file: include unistd.h only when available
[22:36] <CIA-119> ffmpeg: The unistd.h header is only needed for the close() declaration.
[22:36] <CIA-119> ffmpeg: If this header is not available, the close() declaration may be
[22:36] <CIA-119> ffmpeg: provided by another header, e.g. io.h.
[22:36] <michaelni> ubitux, wheres the problem ?
[22:37] <ubitux> michaelni: checkout this branch: https://github.com/ubitux/FFmpeg/compare/subtitles
[22:37] <CIA-119> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[22:37] <CIA-119> ffmpeg: 03Mans Rullgard 07master * rd3d3a32c9d 10ffmpeg/ (5 files in 3 dirs): (log message trimmed)
[22:37] <CIA-119> ffmpeg: lavu: add av_usleep() function
[22:37] <CIA-119> ffmpeg: This function implements a delay using the first available
[22:37] <michaelni> i mean after you flip AV_QSORTs order
[22:37] <CIA-119> ffmpeg: of the following functions:
[22:37] <CIA-119> ffmpeg: - nanosleep()
[22:37] <CIA-119> ffmpeg: - usleep()
[22:37] <CIA-119> ffmpeg: - Sleep() (Windows)
[22:37] <CIA-119> ffmpeg: 03Mans Rullgard 07master * r896bb0d742 10ffmpeg/ (8 files in 3 dirs): 
[22:37] <CIA-119> ffmpeg: Replace usleep() calls with av_usleep()
[22:37] <CIA-119> (16 lines omitted)
[22:37] <ubitux> michaelni: i didn't change the code actually, something looks broken
[22:38] <ubitux> michaelni: http://blackhole.ubitux.fr/0001-use-AV_QSORT.patch
[22:38] <ubitux> i'm trying this on top of the branch
[22:38] <ubitux> but i may be doing something wrong
[22:38] <ubitux> you can try make fate-subtitles to see the result
[22:39] <ubitux> or well, maybe just: ffmpeg -i ~/fate-samples/sub/JACOsub_capability_tester.jss out.ass
[22:46] <michaelni> ubitux, fixed
[22:47] <ubitux> does it pass make fate-subtitles?
[22:48] <michaelni> didnt try
[22:48] <michaelni> i just diffed the .ass file
[22:48] <michaelni> and it matches
[22:49] <ubitux> ok
[22:50] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r0567715a6b 10ffmpeg/libavutil/qsort.h: 
[22:50] <CIA-119> ffmpeg: qsort: flip compare direction to match libc qsort()
[22:50] <CIA-119> ffmpeg: Found-by: ubitux
[22:50] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:19] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * rc17808cebd 10ffmpeg/libavfilter/ (27 files): 
[23:19] <CIA-119> ffmpeg: lavfi: define macro AVFILTER_DEFINE_CLASS
[23:19] <CIA-119> ffmpeg: The macro can be used to define consistently the internal class of a
[23:19] <CIA-119> ffmpeg: filter, save some typing and factorize.
[23:19] <CIA-119> ffmpeg: 03Stefano Sabatini 07master * r012aa12a6b 10ffmpeg/libavfilter/vf_drawtext.c: lavfi/drawtext: remove unnecessary parentheses in init()
[23:53] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rc13e490dce 10ffmpeg/cmdutils.c: 
[23:53] <CIA-119> ffmpeg: codec_release_buffer: fix handling of non user buffers
[23:53] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:53] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r69971410f9 10ffmpeg/libavcodec/snow.c: 
[23:53] <CIA-119> ffmpeg: ff_snow_common_end: assert() that the last buffer freed differs from the current.
[23:53] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:00] --- Sat Jun 23 2012


More information about the Ffmpeg-devel-irc mailing list