[Ffmpeg-devel-irc] ffmpeg-devel.log.20120316

[burek]

[00:09] <ubitux> Compn: nice way of starting a troll, seems Patrick went for it with all his heart ;)
[00:55] <Compn> ubitux : we all must troll pulseaudio, every day, until it is gone from this planet. scorched earth!
[01:01] <durandal_1707> Compn: agreed
[01:02] <durandal_1707> alsa too
[01:02] Action: Daemon404 just thinks audio on linux needs to be completely crapped and redone.
[01:03] <durandal_1707> FreeBSD have OSS
[01:03] <ohsix> do it
[01:03] Action: Daemon404 cant comment on whatever freebsd uses since his only bsd box is a nas
[01:08] <ohsix> i don't understand this and it makes me angrehhhh
[01:08] <ohsix> audio is way too close to a real time system than most people are prepared to deal with
[01:09] <Daemon404> ohsix, the solution of course is mroe abstraction.
[01:09] <ohsix> do you mean to imply pulseaudio or alsa-lib is unneeded
[01:10] <ohsix> alsa just gives access to devices
[01:10] <Daemon404> im saying that audio on linux is a mess of libraries which abstract eacother
[01:10] <Daemon404> or wrap eachother
[01:10] <ohsix> fun thing about audio too, is that none of that matters, unless you're already doing something wrong
[01:11] <ohsix> a wrapper implies they do little to nothing
[01:12] <Daemon404> i am certaily no audio subsystem expert
[01:12] <Daemon404> but i do know a case of "too many ways to do X" when i see it
[01:12] <Daemon404> obligatory graph: http://blogs.adobe.com/penguinswf/files/penguinswf/linuxaudio.png
[01:13] <ohsix> too many ways to play samples? that's usually what you want to do
[01:14] <ohsix> but even that leads to a lot of things that you can't ignore, regardless of what you use to play samples; but in practical terms, people do; and get away with it to a huge degree
[01:17] <ohsix> i have no solution though, i've tried to compile references and stuff but there just aren't many; even the gurus in the domain get away with stuff they should know
[01:19] <ohsix> even if pulseaudio is thrown away in the end, it put a lot of pressure into making the devices work, and cleaning up a lot of ambiguities with alsa; which is awesome
[01:24] <fronk> is git://git.videolan.org/ffmpeg.git still considered a valid repo or should I update my remotes to track git://source.ffmpeg.org/ffmpeg.git
[01:26] <bryno> for the moment, source.ffmpeg.org points to git.videolan.org
[01:27] <Compn> videolan.org is still the official repo 
[01:27] <Compn> last i heard ;)
[01:27] <Compn> Daemon404 : the funny thing is, that graph (made by multimedia mike) was outdated AS SOON AS HE MADE IT
[01:28] <Daemon404> lol yeah
[01:28] <Daemon404> i know
[01:28] <Compn> pretty funny :)
[01:28] <Compn> no one updates penguin.swf after mike left adobe either :\
[01:28] <bryno> in theory, source.ffmpeg.org could point to something else some day :P
[01:28] <ohsix> it's all well and good to graph them all and their relations, but practically speaking you aren't supporting all of them or anything
[01:29] <Compn> ohsix : more bugs to workaround
[01:29] <Compn> depending on which chain is used
[01:29] <Compn> :P
[01:29] <Daemon404> thats back when mike was all like "let's support the most systems possible"
[01:29] <Daemon404> iirc
[01:29] <ohsix> it's open source, if you're working around it for more than a week you're doing it wrong
[01:29] <Compn> you do know that ffmpeg has been working around a gcc bug for ... years, right ?
[01:30] <ohsix> yea, not going to defend gcc tho
[01:30] <Compn> -fomit-frame-pointer :)
[01:30] <Daemon404> ohsix, it just means it's basically impossible to support a wide variety of linux systems with something like flash.
[01:30] <Compn> lol
[01:30] <Compn> well theres lots of bugs like that
[01:30] <Compn> gcc is just the one we like to troll about 
[01:30] <ohsix> why would you do that, and they seem to be doing ok regardless
[01:30] Action: Daemon404 uses clang :P
[01:30] <Compn> sdl had some bug i remember
[01:30] <fronk> Compn: I thought the -fomit-frame-pointer problems went away with gcc 4.4
[01:30] <Compn> but i cant remember specifics
[01:31] <Compn> fronk : its been a while since i tested... the bug comes and goes anyhow
[01:31] <Daemon404> fronk, cant be
[01:31] <Compn> anyways, done trolling, need to do other stuff
[01:31] <Daemon404> ive run into that exact bug while workign with 4.6
[01:31] <Daemon404> and it was a bitch to figure out wtf was going on
[01:32] <Daemon404> or maybe it is a slightly different one
[01:33] <durandal_1707> hate lazy programmers when they do not care when malloc fails
[01:33] <Daemon404> durandal_1707, the ones who never free are much worse.
[01:34] <Daemon404> (hi ms office 97 engineering team)
[01:34] <ohsix> what can you do when malloc fails? it obviously depends on the program, but it usually means you can't make any meaningful progress anyways
[01:34] <Daemon404> likely itll end up just segfaulting.
[01:34] <fronk> I'd rather have people not call free than double-freeing
[01:35] <ohsix> plus a lightweight abort will free memory!
[01:35] <durandal_1707> Daemon404: actually it may not, it may continue to work in unpredictive ways
[01:35] <Daemon404> of course
[01:35] <durandal_1707> like deleting your porn torrents....
[01:36] <ohsix> i'm a fan of asserts that define your expectations rigidly, and breaking early :<
[01:37] <ohsix> some types of software is just hard and that's not acceptable tho
[01:48] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * raba7a827aa 10ffmpeg/libswscale/x86/scale.asm: swscale: convert hscale to cpuflags().
[01:48] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * r98b9da2ac7 10ffmpeg/libavutil/x86/x86inc.asm: x86inc: add *mp named argument support to DEFINE_ARGS.
[01:48] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * r45fdcc8e2d 10ffmpeg/libswscale/x86/scale.asm: swscale: convert hscale() to use named arguments.
[01:48] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r2636e691ce 10ffmpeg/avconv.c: 
[01:48] <CIA-17> ffmpeg: avconv: remove a pointless check.
[01:48] <CIA-17> ffmpeg: output_video_filter is always guaranteed to be set and is in fact
[01:48] <CIA-17> ffmpeg: dereferenced right above the check.
[01:48] <CIA-17> ffmpeg: 03Anton Khirnov 07master * rce326c1130 10ffmpeg/avconv.c: avconv: check for get_filtered_frame() failure.
[01:48] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r251422982b 10ffmpeg/avconv.c: 
[01:48] <CIA-17> ffmpeg: avconv: remove pointless checks.
[01:48] <CIA-17> ffmpeg: OutputStream.picref is guaranteed to be non-NULL if
[01:48] <CIA-17> ffmpeg: get_filtered_video_frame() succeeds.
[01:48] <CIA-17> ffmpeg: 03Anton Khirnov 07master * rcd9a3c3512 10ffmpeg/libavformat/utils.c: lavf: don't select an attached picture as default stream for seeking.
[01:48] <CIA-17> ffmpeg: 03Paul B Mahol 07master * r702985b8b7 10ffmpeg/libavcodec/dxa.c: 
[01:48] <CIA-17> ffmpeg: dxa: remove useless code
[01:48] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[01:48] <CIA-17> ffmpeg: Signed-off-by: Anton Khirnov <anton at khirnov.net>
[03:14] <Compn> http://www.youtube.com/watch?v=gSrN2zIRwN8 , bunch of big names debate war on drugs, julian assange, ian blair, russle brand, richard branson etc
[03:30] <iive> will watch tomorrow.
[03:30] <Compn> julian assange 'under house arrest' :D
[04:51] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rb89f94077d 10ffmpeg/libavfilter/vf_ass.c: 
[04:51] <CIA-17> ffmpeg: ass: fix DAR calculation.
[04:51] <CIA-17> ffmpeg: Found-by: REN Lifeng <renlifeng at wowfly.com>
[04:51] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:21] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rd9fe6b926c 10ffmpeg/libavformat/matroskadec.c: 
[06:21] <CIA-17> ffmpeg: matroskadec: Fix null pointer exception on corrupted input.
[06:21] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[10:37] <kriegerod> is it right that 0.10.1 tag will be taken from master branch?
[11:50] <dilaroga> a008
[13:05] <KnightRiderRetro> hi devs, does ffmpeg supports Adobe HDS (HTTP Dynamic Streaming)?
[13:07] <KnightRiderRetro> i know it supports Apple HLS (HTTP Live Streaming)
[13:15] <Compn> KnightRiderRetro : no, it does not
[13:16] <Compn> i'm also looking for something that supports adobe hds
[13:16] <Compn> so i can download some streams ;)
[13:16] <KnightRiderRetro> any chance of it's being implemented in near future?
[13:17] <KnightRiderRetro> i was also looking it for the same reason :D
[13:17] <Compn> KnightRiderRetro : for playing streams or creating them ?
[13:17] <KnightRiderRetro> playing
[13:17] <Compn> its a pretty funky format, and adobe hasnt released the f4f file format spec
[13:18] <Compn> so it will be a while 
[13:19] <Compn> KnightRiderRetro : i think adobe tied it into the flash plugin, so it will be hard for 3rd party program to play it at all
[13:19] <Compn> so , might not ever get in ffmpeg
[13:20] <KnightRiderRetro> yeah it only works in flash player
[13:22] <KnightRiderRetro> OSMF also supports it
[13:22] <KnightRiderRetro> http://opensource.adobe.com/svn/opensource/osmf/trunk/framework/OSMF/org/osmf/net/httpstreaming/f4f/
[13:23] <Compn> ah, good point
[13:23] <Compn> i wonder if i can use osmf player to play my f4f files
[13:23] <Compn> hmm
[13:24] <Compn> doubt it
[13:24] <KnightRiderRetro> yeah
[13:24] <KnightRiderRetro> you can
[13:24] <Compn> oh neat
[13:24] <Compn> what do you use to download f4f ?
[13:26] <Compn> and where is the osmf player? are there binaries somewhere ? i cant find it..
[13:26] <KnightRiderRetro> search for strobe media player
[13:29] <KnightRiderRetro> http://sourceforge.net/projects/osmf.adobe/files/
[13:29] <Compn> its hard to download f4f because the http server expects a bunch of tokens and cookies. are you able to download it easily ?
[13:29] <Compn> yeah i found it thanks
[13:32] <KnightRiderRetro> yeah that poses problems sometimes
[13:33] <Compn> i use replay media catcher on windows , and edit the 'auto download' http plugin to download video/f4f mimetype. which doesnt catch them all.
[13:35] <KnightRiderRetro> well it feels pointless to try to grab fragments which we can't use aftewards easily
[13:36] <Compn> its possible to download the f4m file too, but the urls in that file are unaccessible
[13:36] <av500> Compn: does f4m have any good p0rn?
[13:36] <Compn> is xml manifest file, like dumb apple format
[13:40] <KnightRiderRetro> <?xml version="1.0" encoding="UTF-8"?>
[13:40] <KnightRiderRetro> <manifest xmlns="http://ns.adobe.com/f4m/1.0">
[13:40] <KnightRiderRetro>   <id>600_1</id>
[13:40] <KnightRiderRetro>   <streamType>live</streamType>
[13:40] <KnightRiderRetro>   <streamBaseTime>1331894043.891</streamBaseTime>
[13:40] <KnightRiderRetro>   <bootstrapInfo profile="named" id="bootstrap_600" url="600_f1efdf988d034fb5-p.bootstrap"/>
[13:40] <KnightRiderRetro>   <media bitrate="600" url="600_f1efdf988d034fb5-p_" bootstrapInfoId="bootstrap_600">
[13:40] <KnightRiderRetro>     <metadata>AgAKb25NZXRhRGF0YQgAAAAMAAhkdXJhdGlvbgAAAAAAAAAAAAAJZnJhbWVyYXRlAEA5AAAAAAAAAAV3aWR0aABAfVAAAAAAAAAGaGVpZ2h0AEB2AAAAAAAAAAx2aWRlb2NvZGVjaWQAQBwAAAAAAAAAF3ZpZGVva2V5ZnJhbWVfZnJlcXVlbmN5AEAAAAAAAAAAAA9hdWRpb3NhbXBsZXJhdGUAQOdwAAAAAAAADWF1ZGlvY2hhbm5lbHMAQAAAAAAAAAAADGF1ZGlvY29kZWNpZABAJAAAAAAAAAAIZmlsZXNpemUAAAAAAAAAAAAADGNhblNlZWtUb0VuZAEAAAAJ</metadata>
[13:40] <KnightRiderRetro>   </media>
[13:40] <KnightRiderRetro> </manifest>
[13:41] <Compn> yeah one of those
[13:41] <KnightRiderRetro> it's actual manifest file for  hds stream
[13:44] <Compn> hmm i cant get strobe media player to play all files
[13:44] <Compn> it works on some mp4, but not others
[13:45] <Compn> oh woops 
[13:45] <KnightRiderRetro> well i also tested only on few streams
[13:45] <Compn> doing it wrong again
[13:45] <Compn> i was putting things in the wrong box
[13:46] <Compn> do you have to play frag1 ?
[13:46] <Compn> or can you play frag4 ?
[13:47] <KnightRiderRetro> you can switch to any fragment
[13:48] <Compn> how do you play one? i'm trying to figure the steps so developers can use it easily 
[13:48] <Compn> i'm at this page right now , file://localhost/H:/StrobeMediaPlayback_1.6.328-full/Debug%20build%20for%20Flash%20Player%2010.1/setup.html
[13:48] <Compn> and i put some flv, mp4 and f4f files on local http server
[13:50] <KnightRiderRetro> well i only downloaded it yesterday night and not familiar with all stuff
[13:50] <Compn> i mean, how do you play f4f with it ?
[13:51] <Compn> i just randomly picked this way, might be wrong.. :)
[13:51] <KnightRiderRetro> i edited the sample page with actual test link
[13:52] <KnightRiderRetro> http://zattoohdslive-f.akamaihd.net/z/600_1@20415/manifest.f4m?hdnea=st=1331901573~exp=1331901633~acl=/z/600_1@20415/*~hmac=5022596186d8a7f4ca48afb5a25b7458198028b3b567e1c4a1035d55064e7e09
[13:52] <KnightRiderRetro> test link
[13:53] <Compn> oh ,  i was trying to play .f4f files i had on my harddrive 
[13:57] Action: Compn goes afk
[13:57] <Compn> bbl
[14:20] <KnightRiderRetro> http://download.macromedia.com/f4v/video_file_format_spec_v10_1.pdf
[14:20] <KnightRiderRetro> anex c for http streaming
[14:34] <Compn> o ksv whats up :D
[14:35] <Compn> KSV : thats f4v not f4f ...
[14:35] <Compn> adobe has f4f under nda
[14:36] <KSV> isn't f4v is the file format for HDS?
[14:36] <KSV> it seems so as per Annex c
[14:37] <KSV> page 63
[14:38] <Compn> yeah, end of the first paragraph
[14:38] <Compn> an http streaming presentation is composed of an http streaming manifest file f4m and http streaming segments (fragmented f4v files or f4fs)
[14:38] <Compn> f4v or f4f
[14:42] <Compn> KSV : fun huh? 
[14:42] <Compn> ehe
[14:42] Action: Compn afk
[14:43] <Compn> KSV : i asked hyc about making a hdsdump project, but i dont think hes interested in it. do you think adobe will switch everyone over to hds ?
[14:43] <KSV> it's not adobe's choice
[14:44] <KSV> content providers are switching to hds
[14:44] <KSV> because adobe has assured them it's better than rtmp :P
[15:19] <kriegerod> is it right that 0.10.1 tag will be taken from master branch?
[15:22] <nevcairiel> should be from the 0.10 branch
[15:25] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r6f9803e5e0 10ffmpeg/libavcodec/sonic.c: 
[15:25] <CIA-17> ffmpeg: sonic: update to new API
[15:25] <CIA-17> ffmpeg: Fixes Ticket1075
[15:25] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:27] <KSV> Compn: as per my understanding f4f is f4v with some extra box types which are defined in the document i previously linked. for example bootstrap box used only in HDS defined on page 39. what else is required by devs to implement it properly? 
[15:28] Action: durandal_1707 michaelni fixing encoders/decoders which assumes linesize[x] is always positive integer?
[16:01] <michaelni> durandal_1707, not sure i understand your question but linesize in general isnt guranteed to be >0, also see CODEC_CAP_NEG_LINESIZES
[16:01] <durandal_1707> what that cap does?
[16:02] <durandal_1707> michaelni: qtrleenc, sunrasterenc (and probably some more) are broken with negative linesize..
[16:07] <durandal_1707> michaelni: nothing use that cap
[16:09] <iive> durandal_1707:  at first I understood your remark as inclining that the sonic bugfix somehow broke the linesize < 0 case.
[16:10] <iive> generally, it is good idea for the codec to be able to work with negative stride/linesize and fixes/patches to do that are welcome.
[16:11] <durandal_1707> iive: wtf, i never mentioned sonic and cant see why audio codec can have use of negative linesize ....
[16:11] <durandal_1707> iive: this is about -vf vflip....
[16:12] <iive> oh, it's that sonic... I guess I thought for another codec.
[16:12] <iive> libav removed sonic from their codebase, haven't they?
[16:20] <michaelni> durandal_1707, without setting that CODEC_CAP a codec should theoretically not have to deal with negative linesizes
[16:20] <michaelni> iive, yes IIRC libav removed sonic
[16:21] <michaelni> i kept it because i saw no reason to remove it
[16:21] <michaelni> its not as if it is a maintaince burden
[16:21] <michaelni> 1 hour every 6 month maybe ...
[16:22] <iive> some deletionists overuse this excuse :)
[16:23] <durandal_1707> michaelni: only ffplay use that cap
[16:23] <iive> I generally like the goal to make ffmpeg able to play every video ever created, even if this means handling obscure non-standard buggy behaviour.
[16:25] <durandal_1707> michaelni: bug no codecs use that flag at all....
[16:25] <durandal_1707> *but
[16:26] <michaelni> someone should test which codecs can handle negtaive linesizes and set the flags accordingly
[16:27] <durandal_1707> shouldn't cap check not be in ffplay/ffmpeg but in library itself?
[16:29] <michaelni> iam not sure how this could be done, the cap has to be passed into whatever allocates frames and thats libavfilter here
[16:31] <CIA-17> ffmpeg: 03Michael Niedermayer 07release/0.10 * r1749b0d74d 10ffmpeg/libavcodec/mmvideo.c: (log message trimmed)
[16:31] <CIA-17> ffmpeg: mmvideo: restore initial y value.
[16:31] <CIA-17> ffmpeg: This bug might have been exploitable (out of HEAP buffer writes)
[16:31] <CIA-17> ffmpeg: Bug introduced by libav
[16:31] <CIA-17> ffmpeg:  commit a55d5bdc6e28a2cfefc440d792de5cc4f02377e2
[16:31] <CIA-17> ffmpeg:  Date: Tue Mar 6 15:15:42 2012 -0800
[16:31] <CIA-17> ffmpeg:  algmm: convert to bytestream2 API.
[16:31] <CIA-17> ffmpeg: 03Michael Niedermayer 07release/0.10 * r6242dae507 10ffmpeg/libavcodec/sonic.c: 
[16:31] <CIA-17> ffmpeg: sonic: update to new API
[16:31] <CIA-17> ffmpeg: Fixes Ticket1075
[16:31] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:31] <CIA-17> ffmpeg: (cherry picked from commit 6f9803e5e02c557e1003cface9f3084a7e1e43e4)
[16:31] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:31] <CIA-17> ffmpeg: 03Michael Niedermayer 07release/0.10 * r0be85fd80f 10ffmpeg/libavcodec/diracdec.c: 
[16:31] <CIA-17> ffmpeg: diracdec: Check for negative quants which would cause out of array reads.
[16:31] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[16:31] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:32] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[16:32] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:32] <CIA-17> ffmpeg: (cherry picked from commit 0f13cc732b3752828890b8dff507615cfd454336)
[16:32] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:32] <CIA-17> ffmpeg: 03Michael Niedermayer 07release/0.10 * r9f253ebb41 10ffmpeg/libavcodec/diracdec.c: 
[16:32] <CIA-17> ffmpeg: diracdec: Fix integer overflow leading to out of global array read.
[16:32] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[16:32] <CIA-17> ffmpeg: Reviewed-by: Paul B Mahol <onemda at gmail.com>
[16:32] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:32] <CIA-17> ffmpeg: (cherry picked from commit 9729f140ae073f1df2041b6c5fd2068592eb9c48)
[16:32] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:57] <Compn> KSV : well if adobe tells people hds is better than rtmp, people will switch. therefor i blame adobe switching people over :P
[16:58] <kierank> i've had flames with adobe people about that
[17:01] <Compn> about which part? all of it ?
[17:01] <Compn> or adobe saying its their choice while also telling people that hds is better than rtmp? :P
[20:07] <CIA-17> ffmpeg: 03Michael Niedermayer 07release/0.10 * r313ddbfe48 10ffmpeg/libavcodec/proresdec2.c: 
[20:07] <CIA-17> ffmpeg: proresdec: Fix read via negative index in a global array.
[20:07] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[20:07] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:07] <CIA-17> ffmpeg: (cherry picked from commit 006508032057824a371bec4e629b66f8cbb26c47)
[20:07] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:07] <CIA-17> ffmpeg: 03Michael Niedermayer 07release/0.10 * r001f4c7dc6 10ffmpeg/libavcodec/jpeglsdec.c: 
[20:07] <CIA-17> ffmpeg: jpeglsdec: Prevent out of array write.
[20:07] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:07] <CIA-17> ffmpeg: (cherry picked from commit 00ab9cdae1a96dfea33cd505076a83823f390aa4)
[20:07] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:32] <Daemon404> i really need to fix my hilights
[20:32] <Daemon404> sha1s always alert me
[21:31] <CIA-17> ffmpeg: 03Wolfram Gloger 07master * rf8353d5fda 10ffmpeg/ (3 files in 2 dirs): (log message trimmed)
[21:31] <CIA-17> ffmpeg: mpegvideo: don't pretend the first frame is always a key frame
[21:31] <CIA-17> ffmpeg: Signed-off-by: Wolfram Gloger <wmglo at dent.med.uni-muenchen.de>
[21:31] <CIA-17> ffmpeg: Modify the parser initialization so that parsers can
[21:31] <CIA-17> ffmpeg: set pict_type themselves. Use this in the mpegvideo parser
[21:31] <CIA-17> ffmpeg: so that initial frames are not unconditionally I frames.
[21:31] <CIA-17> ffmpeg: I have had this in my tree for several years.
[21:32] <Compn> ooooo several year old patch :)
[00:00] --- Sat Mar 17 2012