[Ffmpeg-devel-irc] ffmpeg-devel.log.20120522

[burek]

[00:04] <CIA-105> ffmpeg: 03Samuel Pitoiset 07master * r5d603f1b65 10ffmpeg/libavformat/http.c: 
[00:04] <CIA-105> ffmpeg: http: Factorize the code by adding http_read_header()
[00:04] <CIA-105> ffmpeg: This function is used for reading http reply headers.
[00:04] <CIA-105> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[00:04] <CIA-105> ffmpeg: 03Diego Biurrun 07master * r9309bdfe2c 10ffmpeg/ (configure doc/fate.texi): 
[00:04] <CIA-105> ffmpeg: fate: change name of FATE samples location environment variable
[00:04] <CIA-105> ffmpeg: FATE_SAMPLES is now used directly by the Makefiles, which induces the test
[00:04] <CIA-105> ffmpeg: system to run a test with the value of the environment variable as name.
[00:04] <CIA-105> ffmpeg: Renaming the environment variable to LIBAV_SAMPLES avoids this problem.
[00:04] <CIA-105> ffmpeg: 03Jordi Ortiz 07master * r13a867a63e 10ffmpeg/libavcodec/ (dwt.c dwt.h): 
[00:04] <CIA-105> ffmpeg: dwt: K&R prettyprinting cosmetics
[00:04] <CIA-105> ffmpeg: Signed-off-by: Diego Biurrun <diego at biurrun.de>
[00:04] <CIA-105> ffmpeg: 03Diego Biurrun 07master * rdb9e00f469 10ffmpeg/ (4 files in 2 dirs): 
[00:04] <CIA-105> ffmpeg: Remove libnut wrapper
[00:04] <CIA-105> ffmpeg: libnut is unmaintained and known to be buggy; native NUT code exists.
[00:05] <CIA-105> ffmpeg: Merge remote-tracking branch 'qatar/master'
[00:05] <CIA-105> ffmpeg: * qatar/master:
[00:05] <CIA-105> ffmpeg:  dwt: K&R prettyprinting cosmetics
[00:05] <CIA-105> ffmpeg:  Remove libnut wrapper
[00:05] <CIA-105> ffmpeg:  fate: change name of FATE samples location environment variable
[00:05] <CIA-105> ffmpeg:  avformat: Add a flag to mark muxers that allow (non strict) monotone timestamps.
[00:05] <CIA-105> ffmpeg: 03James Zern 07master * re9cef89702 10ffmpeg/ (11 files in 2 dirs): 
[00:05] <CIA-105> ffmpeg: avformat: Add a flag to mark muxers that allow (non strict) monotone timestamps.
[00:05] <CIA-105> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[00:44] <Zeranoe> michaelni: I haven't had a chance to yet, I will soon
[00:45] <CIA-105> ffmpeg: 03Stefano Sabatini 07master * r27127ebe6c 10ffmpeg/doc/ (Makefile ffmpeg.texi ffplay.texi ffprobe.texi syntax.texi): doc: add syntax.texi file
[00:49] <RobertNagy> is lavfi supposed to be able to handle pts rollover? if not I think there is a bug in ffplay for looping input
[00:58] <saste> RobertNagy: I suppose not right now
[00:58] <saste> RobertNagy: command to reproduce it?
[00:59] <RobertNagy> I don't have any command just reading the code
[00:59] <RobertNagy> if the input loop then you get an pts rollover
[00:59] <RobertNagy> which can cause problems
[01:00] <RobertNagy> since there is no guarantee for proper handling in lavfi
[01:00] <RobertNagy> the same with seeking
[01:00] <RobertNagy> the filter graph should probably be reinitialized the same time as av_flush_codec is called
[01:01] <RobertNagy> or lavfi should guarantee proper handling of rollover pts
[01:02] <RobertNagy> e.g. yadif currently doesn't handle rollover pts properly after the libav merge
[01:02] <RobertNagy> I'm not very familiar with other filters
[01:02] <saste> did it before the merge?
[01:03] <RobertNagy> yes
[01:03] <RobertNagy> the question is
[01:03] <RobertNagy> should it handle it?
[01:03] <RobertNagy> I can send a patch that re adds the rollover handling
[01:04] <michaelni> i think filters that can easily handle timestamp discontinuities, should handle them
[01:04] <michaelni> that way they can easily be preserved if someone wants to preserve them
[01:05] <RobertNagy> well, I think we either need to review all the filter to handle it, or have ffplay reinit the graph, as it is now I would suspect that it is broken
[01:07] <michaelni> reinit is problematic
[01:07] <michaelni> it destroys state
[01:07] <RobertNagy> indeed, and since the graph has a source filter
[01:07] <RobertNagy> that shouldn't be reinit
[01:07] <RobertNagy> makes it even more problematic
[01:09] <RobertNagy> oh, vf_fps is merged as well
[01:09] <RobertNagy> there is a problem with vf_fps and sink_buffer
[01:10] <RobertNagy> vf_fps sometimes returns 0 when it should'nt, which causes sink_buffer to return EINV instead of EOF
[01:12] <RobertNagy> hm, or maybe that was fixed, I will need review it closer
[01:16] <saste> RobertNagy: please report on list if you find problems, so we can have a look at them
[01:16] <RobertNagy> vf_fps doesn't handle rollover either
[01:16] <RobertNagy> saste: will do
[10:26] <ubitux> http://onlinelib.de/opensource.html
[10:26] <ubitux> vimdiff libavformat/{mpegtsenc.c,iphoneenc.c}
[10:30] <cbsrobot> x264 is a very good H.264 Codec
[14:06] <CIA-105> ffmpeg: 03Paul B Mahol 07master * r672f8226bb 10ffmpeg/libavcodec/shorten.c: 
[14:06] <CIA-105> ffmpeg: shorten: fix lossy decoding
[14:06] <CIA-105> ffmpeg: Closes #1299, #1300 and #1301.
[14:06] <CIA-105> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[16:06] <CIA-105> ffmpeg: 03Michael Niedermayer 07master * r696452e4c5 10ffmpeg/libavcodec/ (avcodec.h version.h): 
[16:06] <CIA-105> ffmpeg: avcodec: add parser flag to indicate that codec TS are the only available and should be used
[16:06] <CIA-105> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:06] <CIA-105> ffmpeg: 03Michael Niedermayer 07master * ra6b3471c44 10ffmpeg/libavformat/ (avformat.h utils.c): 
[16:06] <CIA-105> ffmpeg: avformat: add needs_parsing type to enable codec TS use.
[16:06] <CIA-105> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:06] <CIA-105> ffmpeg: 03Michael Niedermayer 07master * r035f6620ce 10ffmpeg/libavcodec/mpeg4video_parser.c: 
[16:06] <CIA-105> ffmpeg: mpeg4videoparser: support using timestamps from codec.
[16:06] <CIA-105> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:06] <CIA-105> ffmpeg: 03Michael Niedermayer 07master * r1dd71e1cd5 10ffmpeg/libavformat/rawdec.c: 
[16:06] <CIA-105> ffmpeg: rawdec: force timestamps from codec to be used
[16:06] <CIA-105> ffmpeg: This fixes playback timing with at least m4v files
[16:06] <CIA-105> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:07] <CIA-105> ffmpeg: 03Matthieu Bouron 07master * rf452d6c34d 10ffmpeg/libavformat/mxfdec.c: 
[16:07] <CIA-105> ffmpeg: mxfdec: fix frame height computation for mixed fields layout
[16:07] <CIA-105> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:07] <CIA-105> ffmpeg: 03Robert Nagy 07master * r965b807278 10ffmpeg/libavfilter/vf_yadif.c: 
[16:07] <CIA-105> ffmpeg: vf_yadif: Fixed potential access violation on EOF.
[16:07] <CIA-105> ffmpeg: The correct invariant is (cur != NULL -> next != NULL). Not the other way around.
[16:07] <CIA-105> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:16] <Tjoppen> ah, the mxf patch looks fine btw. just spotted it
[16:45] <michaelni> Tjoppen, the one i applied ? or did i miss another ?
[16:48] <Tjoppen> nah, that one
[17:01] <burek> when browsing git with a web browser
[17:01] <burek> how can I find this commit git-N-28506-g3660b5b
[17:02] <burek> whatever I type in a search box here, I get nothing: http://git.videolan.org/?p=ffmpeg.git
[17:02] <Tjoppen> try 3660b5b
[17:02] <burek> "No match."
[17:03] <Daemon404> michaelni, that sample i had that used to generate infinite silence, now just segfaults ffmpeg with your async changes
[17:03] <Tjoppen> you can git clone and use gitk 3660b5b
[17:04] <burek> well, it's not that convenient to clone entire repos just to find out some date/time info :)
[17:04] <nevcairiel> the interface is a bit buggy, it actually finds something, but it still claims it doesnt
[17:04] <Tjoppen> commit 3660b5bca9aa076214e8ee7f6c3cceee8641f31f
[17:04] <Tjoppen> Author: j at v2v.cc <j at v2v.cc>
[17:04] <Tjoppen> Date:   Thu Mar 17 17:57:52 2011 +0100
[17:04] <michaelni> Daemon404, which sample ?
[17:04] <Tjoppen> remove definition of swab
[17:04] <nevcairiel> click on the header above the "No match." line :p
[17:05] <Daemon404> michaelni, i can share it, but it's rather large (700 mb)
[17:05] <Daemon404> i havent been able to cut it and reproduce it
[17:05] <burek> thanks Tjoppen :)
[17:05] <burek> I just need to find the way to get it over a web interface now :)
[17:05] <nevcairiel> you can also do http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=<hashhere>
[17:06] <michaelni> hmm, well i dont mind downloading 700mb if thats what i need to do to debug it
[17:06] <nevcairiel> so use http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=3660b5b
[17:06] <nevcairiel> never fails if you know the hash
[17:06] <Daemon404> http://pastebin.com/raw.php?i=26v2hbDX <-- backtrace
[17:08] <burek> nevcairiel, thanks!!
[17:08] <burek> :D
[17:08] <burek> it uses partial matching and was giving me correct output up until I shortened hash to 366 :)
[17:08] <burek> cool :)
[17:09] <nevcairiel> 7 chars is the usually accepted short-form
[17:09] <burek> I'll make a shortcut on the bot for that :)
[17:11] <burek> "<nevcairiel> click on the header above the "No match." line :p" <- lol :) didn't see this before :))) it does work :)
[17:11] <nevcairiel> i was wrong though, results in some other commit
[17:12] <nevcairiel> search seems to work on my gitweb, something is wrong on videolans
[17:12] <burek> I see, ok, h=hash will do already :)
[17:13] <nevcairiel> guess it doesnt expect you to search for a hash
[20:34] <CIA-105> ffmpeg: 03Carl Eugen Hoyos 07master * r50858a64bf 10ffmpeg/ (8 files in 5 dirs): Add Avid Meridien Uncompressed (AVUI) encoder.
[20:40] <CIA-105> ffmpeg: 03Michael Niedermayer 07master * ra645049bd8 10ffmpeg/libswresample/swresample.c: 
[20:40] <CIA-105> ffmpeg: swr: check count argument of realloc_audio()
[20:40] <CIA-105> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:40] <CIA-105> ffmpeg: 03Michael Niedermayer 07master * rd9b92980c9 10ffmpeg/libavcodec/libvorbis.c: 
[20:40] <CIA-105> ffmpeg: libvorbis: only store version string if bitexact is not set.
[20:40] <CIA-105> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:41] <CIA-105> ffmpeg: 03Michael Niedermayer 07master * r841c1be57e 10ffmpeg/libswresample/swresample.c: 
[20:41] <CIA-105> ffmpeg: swr: check return value of swr_inject_silence/drop_output.
[20:41] <CIA-105> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:51] <ubitux> git blame -C -C -C -C
[23:51] <ubitux> looks like a C-C-C-COMBO BREAKER
[23:56] <Compn> i miss killer instinct
[23:56] <Compn> ki2 was such a letdown
[23:57] <CIA-105> ffmpeg: 03Justin Ruggles 07master * rcc30080b3f 10ffmpeg/libavfilter/ (Makefile vf_split.c): 
[23:57] <CIA-105> ffmpeg: lavfi: rename vf_split.c to split.c
[23:57] <CIA-105> ffmpeg: This is in preparation for adding an audio split filter.
[23:57] <CIA-105> ffmpeg: 03Justin Ruggles 07master * rafeb3590fc 10ffmpeg/ (6 files in 3 dirs): 
[23:57] <CIA-105> ffmpeg: lavfi: add an audio split filter
[23:57] <CIA-105> ffmpeg: Based on current version of the asplit filter in FFmpeg written by
[23:57] <CIA-105> ffmpeg: Stefano Sabatini and others.
[23:57] <CIA-105> ffmpeg: 03Kieran Kunhya 07master * r5ff01259a8 10ffmpeg/libavcodec/ (10 files in 2 dirs): 
[23:57] <CIA-105> ffmpeg: Convert vector_fmul range of functions to YASM and add AVX versions
[23:57] <CIA-105> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[23:57] <CIA-105> ffmpeg: 03Michael Niedermayer 07master * rea5dab58e0 10ffmpeg/: (log message trimmed)
[23:57] <CIA-105> ffmpeg: Merge remote-tracking branch 'qatar/master'
[23:57] <CIA-105> ffmpeg: * qatar/master:
[23:57] <CIA-105> ffmpeg:  dwt: check malloc calls
[23:57] <CIA-105> ffmpeg:  ppc: Drop unused header regs.h
[23:57] <CIA-105> ffmpeg:  af_resample: remove an extra space in the log output
[23:57] <CIA-105> ffmpeg:  Convert vector_fmul range of functions to YASM and add AVX versions
[23:57] <CIA-105> ffmpeg: 03Diego Biurrun 07master * r3ea5429489 10ffmpeg/libavcodec/ppc/regs.h: ppc: Drop unused header regs.h
[23:57] <CIA-105> ffmpeg: 03Jordi Ortiz 07master * rc89e428ed8 10ffmpeg/libavcodec/dwt.c: 
[23:57] <CIA-105> ffmpeg: dwt: check malloc calls
[00:00] --- Wed May 23 2012