[Ffmpeg-devel-irc] ffmpeg-devel.log.20120628

burek burek021 at gmail.com
Fri Jun 29 02:05:04 CEST 2012


[00:15] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rb1a17953ec 10ffmpeg/libavcodec/lpc.c: 
[00:15] <CIA-119> ffmpeg: lpc: use av_assert
[00:15] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:15] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rd1c74ca2be 10ffmpeg/libavcodec/lsp.c: 
[00:15] <CIA-119> ffmpeg: lsp: use av_assert
[00:15] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:39] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * ree7214c59a 10ffmpeg/libavutil/log.c: 
[00:39] <CIA-119> ffmpeg: log: change color for filters from blue to bright green
[00:39] <CIA-119> ffmpeg: The blue is difficult to read on several peoples terminals with black background.
[00:39] <CIA-119> ffmpeg: Idea-by: Paul B Mahol <onemda at gmail.com>
[00:39] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:39] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r7803a04041 10ffmpeg/libavformat/http.c: 
[00:39] <CIA-119> ffmpeg: http: try to detect live akamai streams and dont enable seeking for them
[00:39] <CIA-119> ffmpeg: Fixes ticket1320
[00:39] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:39] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r4ccf6e3971 10ffmpeg/libswresample/ (resample.c resample_template.c x86/resample_mmx.h): 
[00:39] <CIA-119> ffmpeg: swr: MMX2 & SSSE3 int16 resample core
[00:39] <CIA-119> ffmpeg: about 4 times faster
[00:39] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:47] <ubitux> nice push :)
[00:51] <ubitux> michaelni: resample_mmx.h seems to be the only one using __asm
[00:51] <ubitux> why not __asm__?
[01:03] <michaelni> ubitux, no reason, i hadnt noticed,  feel free to change it
[01:05] <ubitux> i'm curious to see if it will make some fate instance red
[01:05] <ubitux> so maybe later
[01:09] <CIA-119> ffmpeg: 03Jordi Ortiz 07master * r58f3e09ee4 10ffmpeg/libavformat/tcp.c: 
[01:09] <CIA-119> ffmpeg: tcp: Set AI_PASSIVE when the socket will be used for listening
[01:09] <CIA-119> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:09] <CIA-119> ffmpeg: 03Jordi Ortiz 07master * ref882e464a 10ffmpeg/libavformat/tcp.c: 
[01:09] <CIA-119> ffmpeg: tcp: Pass NULL as hostname to getaddrinfo if the string is empty
[01:09] <CIA-119> ffmpeg: This gives you the proper v4 or v6 version of the "any address",
[01:09] <CIA-119> ffmpeg: allowing receiving connections on any address on the machine.
[01:09] <CIA-119> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:09] <CIA-119> ffmpeg: 03Martin Storsjö 07master * r75d339e044 10ffmpeg/ (configure doc/protocols.texi libavformat/udp.c): 
[01:09] <CIA-119> ffmpeg: udp: Support IGMPv3 source specific multicast and source blocking
[01:09] <CIA-119> ffmpeg: Based on an original patch by Stephen D'Angelo <SDAngelo at evertz.com>.
[01:09] <CIA-119> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:09] <CIA-119> ffmpeg: 03Ronald S. Bultje 07master * rfa84506177 10ffmpeg/ (configure libavcodec/dxva2_internal.h): 
[01:09] <CIA-119> ffmpeg: dxva2: include dxva.h if found
[01:09] <CIA-119> ffmpeg: Apparently, some build environments require dxva.h even for dxva2,
[01:09] <CIA-119> ffmpeg: while others lack this header entirely. Including it conditionally
[01:09] <CIA-119> ffmpeg: allows building in both cases.
[01:09] <CIA-119> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:09] <CIA-119> ffmpeg: 03Martin Storsjö 07master * r46df708b45 10ffmpeg/ (configure libavutil/libm.h): 
[01:10] <CIA-119> ffmpeg: the assumptions we do in the new code.
[01:10] <CIA-119> ffmpeg: Tested on an IMC sample: 430c -> 370c.
[01:10] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[01:10] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r87df986dcf 10ffmpeg/: (log message trimmed)
[01:10] <CIA-119> ffmpeg: Merge remote-tracking branch 'qatar/master'
[01:10] <CIA-119> ffmpeg: * qatar/master:
[01:10] <CIA-119> ffmpeg:  mss1: validate number of changeable palette entries
[01:10] <CIA-119> ffmpeg:  mss1: report palette changed when some additional colours were decoded
[01:10] <CIA-119> ffmpeg:  x86: fft: replace call to memcpy by a loop
[01:10] <CIA-119> ffmpeg:  udp: Support IGMPv3 source specific multicast and source blocking
[01:10] <CIA-119> ffmpeg: 03Kostya Shishkov 07master * r8f5d573a83 10ffmpeg/libavcodec/mss1.c: mss1: report palette changed when some additional colours were decoded
[01:35] <durandal_1707> michaelni: xpm decoder needs adding more colors to color table in lavu/parseutils
[01:43] <michaelni> adding more should not be a problem
[02:48] <Nickname123> The other day I opened a feature request: https://ffmpeg.org/trac/ffmpeg/ticket/1483
[02:49] <Nickname123> I toyed around with implementing it: https://github.com/thenewguy/FFmpeg/commit/c92a070d376cff11af6011911325893847f240bb
[02:49] <Nickname123> Are github pull requests an acceptable way to submit a patch?
[02:50] <durandal_1707> i don't think so
[02:51] <Nickname123> I have read http://git.videolan.org/?p=ffmpeg.git;a=blob_plain;f=doc/issue_tracker.txt;hb=HEAD
[02:51] <durandal_1707> Nickname123: http://ffmpeg.org/developer.html#Contributing
[02:52] <Nickname123> How do I go about getting my patch approved by another person
[02:52] <Nickname123> okay Ill read that one
[02:52] <Nickname123> Oh nice.  Github is accepted =)
[02:53] <durandal_1707> pull request may be ignored, patches to ml list less so
[02:53] <Nickname123> Okay.  I didn't see a guide to creating patches.  Should they be from the base FFmpeg repo?
[02:54] <Nickname123> *guidelines
[02:54] <durandal_1707> Nickname123: http://ffmpeg.org/developer.html#patch-submission-checklist
[02:55] <Nickname123> well dang. my bad
[03:02] <Nickname123> thanks durandal_1707 btw
[03:09] <Nickname123> Is it possible to sign off on a commit after it has been committed?
[03:27] <durandal_1707> michaelni: already available colors in parseutils conflict with xpm/x11 definitions, like green color it is more for SVG compliance
[04:36] Action: Nickname123[A] is now away - Reason : sleep
[07:36] <ubitux> Nickname123[A]: please follow the surrounding coding style, use av_strtok() instead of strtok(), use strtol() instead of atoi()
[07:36] <ubitux> Nickname123[A]: also, since you're working on the segmenter, you should see with Stefano (saste) when he is here
[07:36] <ubitux> he's done some work on it
[07:36] <ubitux> (which he hadn't submitted yet...)
[07:37] <ubitux> see https://gitorious.org/~saste/ffmpeg/sastes-ffmpeg/commits/misc-segment-fixes-20120206 for instance
[07:37] <ohsix> is av_strtok like strtok_r
[07:37] <ubitux> ohsix: yes
[07:37] <ohsix> ic
[08:22] <jesk> good morning
[08:24] <jesk> before me opens bug maybe it would make sense to get confirmation here in the first place
[08:24] <jesk> i'am too unexperienced if that could be expected behavior
[08:25] <jesk> i have pretty simple configuration, more or less stripped down the example configs to only have one stream and one feed
[08:25] <jesk> when adding the multicast options to the stream configuration the server won't response anymore to http request
[08:26] <jesk> s,to http request, to any http request,
[08:26] <jesk> i'am using 0.11.1, osx build
[08:27] <jesk> when removing the multicast options the server responses again
[08:29] <jesk> i tried the debug flag, but tells nothing about this
[10:08] <michaelni> jesk, for which protocol do you need multicast and for what do you need http
[10:08] <CIA-41> ffmpeg: 03Michael Bradshaw 07master * rfc5999d027 10ffmpeg/libavformat/ (avformat.h options.c): 
[10:08] <CIA-41> ffmpeg: lavf: add proper enum type for fmt ctx duration esitmation method
[10:08] <CIA-41> ffmpeg: Signed-off-by: Michael Bradshaw <mbradshaw at sorensonmedia.com>
[10:08] <michaelni> i assume you arent trying to get multicast http to work :)
[10:21] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rccbcf482ad 10ffmpeg/libavformat/udp.c: 
[10:21] <CIA-41> ffmpeg: udp: use av_freep() instead of av_free()
[10:21] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[11:20] <mwkenna> Hi guys. I have a question with regards to FFMpeg and LibAv. What's the difference? Also - does the old FFMPEG API mailing list still run?
[11:21] <mwkenna> (I am using avcodec for my projects also)
[11:26] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r782763ed2b 10ffmpeg/libswresample/resample.c: 
[11:26] <CIA-41> ffmpeg: swr: fix compilation with ancient toolchain that doesnt support SSSE3
[11:26] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[13:08] <DEgITx> hellow guys, may i ask some questions about ffmpeg api?
[13:10] <michaelni> yes
[13:13] <michaelni> j-b, about the vscprintf issue, can you provide more details about the build environment (to nicolas) ?
[13:14] <DEgITx> first one about av_read_frame()... is there any posible way to know return maximum packet size?
[13:14] <michaelni> both mingw 32 & 64bit fate.ffmpeg clients are also green 
[13:16] <michaelni> DEgITx, the maximum is limited by the file format, by INT_MAX of the size field and in some cases by the maximunm packet size of the codec
[13:16] <michaelni> its also limited by max_alloc_size indirectly
[13:18] <michaelni> it also is limited by the filesize in many cases
[13:26] <DEgITx> ok ... MAX_INT bad choise for limitation in my way), so is any posible simple ways to get codec/file limit (independently from codec and file format) ?
[13:27] <michaelni> theres no lisit in ffmpeg about such limits no, btw why do you need to know it ?
[13:27] <DEgITx> i mean AVCodecContext or something like this
[13:30] <DEgITx> i'm trying to play file while it's downloading (such task), and care about av_read_frame() to have enough bytes for reading
[13:32] <michaelni> you would have to add support to the demuxer to return EAGAIN in if theres not enough data
[13:32] <michaelni> that is if you want it to work well
[13:32] <michaelni> with low latency
[13:33] <michaelni> max frame/packet size can be quite large compared to average with many file formats
[13:34] <DEgITx> about reading again, what do you mean (any example?)
[13:36] <michaelni> that a demuxer and  av_read_frame would return EAGAIN if there is not enough data instead of blocking or retunrning EOF
[13:36] <michaelni> and that they then could continue/be retried when theres more data
[13:36] <michaelni> but as said you would need to add support for this in libavformat to the demxuer that you need
[13:42] <DEgITx> and there is no problem with truncation of packets in such way?
[13:44] <michaelni> no, the demuxer would have to seek back and restore any changes to its state before retunring EAGAIN
[13:55] <DEgITx> ok, i understand that it's good variant to add support direct to the demuxer... but i'm trying don't touch sources of ffmpeg and making realization separately from it as well as caring about integrity of file
[14:44] <j-b> michaelni: normal debian, i586-migw32msvc from mingw.org
[14:47] <michaelni> j-b, thx, i forwarded the awnser to nicolas
[15:01] <funman> mingw.org disappeared from sid btw
[16:52] <nevcairiel> people are still using old legacy-mingw?
[16:57] <j-b> yes.
[16:59] <nevcairiel> they should stop :P
[17:03] <j-b> when ming-w64 will be stable, indeed
[17:06] <funman> if you have an issue with mingw-w64 just tell me
[17:09] <j-b> lack of maturity, so far.
[17:10] Action: funman slaps j-b
[17:12] <j-b> la violence est le dernier recours de l'incompétent
[17:17] <av500> didnt you agree to be nice to each other?
[17:17] <av500> in #videolan
[17:17] <av500> apply the same rules here
[17:18] <j-b> av500: we love each other.
[17:18] <av500> pics or it did not happen
[17:22] <j-b> av500: http://people.videolan.org/~funman/pics/coding/IMG_20120619_170645.jpg
[17:23] <av500> ok, seems mostly IRC-love :)
[17:30] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * re377208d43 10ffmpeg/ffmpeg.c: 
[17:30] <CIA-41> ffmpeg: ffmpeg: dont copy creation_time as the destination file is not created at that time
[17:30] <CIA-41> ffmpeg: Fixes Ticket1439
[17:30] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:42] <TimNich> Thanks Michael...
[17:46] <cbsrobot> j-b: the picture seems to be photoshoped - i do not see any porn
[19:18] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r61123fb8f7 10ffmpeg/libavutil/parseutils.c: 
[19:18] <CIA-41> ffmpeg: parseutils: fix sorting of colors
[19:18] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[19:19] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r3f59bbf5b6 10ffmpeg/libavfilter/af_amerge.c: 
[19:19] <CIA-41> ffmpeg: lavfi/amerge: silence warnings
[19:19] <CIA-41> ffmpeg: The warning silenced was: libavfilter/af_amerge.c:144:29: warning: conversion specifies type 'long long' but the argument has type 'int' [-Wformat]
[19:19] <CIA-41> ffmpeg: The warning was introduced after FF_API_SAMPLERATE64 removal.
[19:19] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[19:57] <durandal_1707> michaelni: could i get help with making ffmpeg outputs filters AVOptions?
[21:00] <michaelni> durandal_1707, sure, what exactly do you mean ?
[21:17] <CIA-41> ffmpeg: 03Nicolas George 07master * r05d6cc116e 10ffmpeg/libavfilter/buffersrc.c: buffersrc: warn when there are too many buffers.
[21:17] <CIA-41> ffmpeg: 03Nicolas George 07master * rfcf8706ed9 10ffmpeg/libavfilter/sink_buffer.c: sink_buffer: warn when there are too many buffers.
[21:35] <jesk> why does ffserver stops responding to any http request as soon as those few multicast options are configured?
[21:48] <CIA-41> ffmpeg: 03Nicolas George 07master * r2c793b8501 10ffmpeg/ffmpeg.c: 
[21:48] <CIA-41> ffmpeg: ffmpeg: warn when -t is used for inputs.
[21:48] <CIA-41> ffmpeg: Using -t on an input already have surprising results.
[21:48] <CIA-41> ffmpeg: Furthermore, using it on an input or an output makes
[21:48] <CIA-41> ffmpeg: a real difference if there are speed-altering filters.
[21:48] <CIA-41> ffmpeg: Implementing -t for inputs will probably result in some
[21:48] <CIA-41> ffmpeg: behavour changes.
[21:49] <CIA-41> ffmpeg: 03Nicolas George 07master * r8069db8633 10ffmpeg/ffmpeg.c: ffmpeg: warn that -t does not work with -filter_complex.
[22:07] <durandal_1707> why I get "Audio timestamp 85 < 284 invalid, cliping" ?
[22:23] <durandal_1707> resolved: was because get_frame_defaults was not called....
[22:26] <CIA-41> ffmpeg: 03Jeff Downs 07master * r8d9fd58113 10ffmpeg/libavcodec/h264.c: (log message trimmed)
[22:26] <CIA-41> ffmpeg: h264: Fix maximum reference count check for non-b frames
[22:26] <CIA-41> ffmpeg: Below fixes the maximum reference count check for second reference list in
[22:26] <CIA-41> ffmpeg: non-B frames. There is nothing to prohibit full (field sized) reference
[22:26] <CIA-41> ffmpeg: list in this case as far as I can tell, and this fixes several syntax-test
[22:26] <CIA-41> ffmpeg: files here (this is a regression caused when this check was made more
[22:26] <CIA-41> ffmpeg: stringent by
[22:26] <CIA-41> ffmpeg: 03Nick Brereton 07master * r16f6c16ac0 10ffmpeg/libavcodec/dca.c: 
[22:26] <CIA-41> ffmpeg: Parse & decode DTS XXCH frames
[22:26] <CIA-41> ffmpeg: Reviewed-by: Benjamin Larsson <benjamin at southpole.se>
[22:26] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:26] <CIA-41> ffmpeg: 03Nick Brereton 07master * rcd8bef969a 10ffmpeg/libavcodec/ (dca.c dcadata.h): 
[22:26] <CIA-41> ffmpeg: Generate channel layout, reordering for DTS-XXCH extension and, undo embedded downmixes
[22:26] <CIA-41> ffmpeg: Reviewed-by: Benjamin Larsson <benjamin at southpole.se>
[22:26] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:28] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rb0fdd3489a 10ffmpeg/ (4 files in 2 dirs): 
[22:28] <CIA-41> ffmpeg: lavfi: deprecate av_buffersrc_buffer() function
[22:28] <CIA-41> ffmpeg: Favor av_buffersrc_add_ref() instead, which is more powerful.
[22:28] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r7877b50d18 10ffmpeg/ffplay.c: ffplay: simplify code by using avfilter_unref_bufferp()
[22:55] <ubitux> llogan: do you have write access to push your patch or a developer need to push it for you?
[23:14] <j-b> So, this is different codecs, almost, but same CODEC_ID?
[23:20] <llogan> ubitux: i believe i only have write access for ffmpeg-web, but right now i'd prefer if others push it anyway
[23:21] <llogan> if isn't not too much a PITA
[00:00] --- Fri Jun 29 2012


More information about the Ffmpeg-devel-irc mailing list