[Ffmpeg-devel-irc] ffmpeg-devel.log.20130627

[burek]

[00:08] <saste> $ffplay ffplay
[00:10] <Daemon404> wut
[00:21] <saste> ^^ ffplay plays with itself
[00:24] <llogan> ffapper
[00:26] <cone-855> ffmpeg.git 03Stefano Sabatini 07master:205092bf478f: ffprobe: simplify branching logic in probe_file()
[00:26] <cone-855> ffmpeg.git 03Nicolas George 07master:a334b00cf62c: ffprobe: fix exit code with stream specifiers
[00:26] <cone-855> ffmpeg.git 03Stefano Sabatini 07master:1fc626f8d072: ffprobe: reindent after previous commit
[00:27] <cone-855> ffmpeg.git 03Stefano Sabatini 07master:5c616fe48bf6: ffprobe: always exit 1 in case of errors
[00:44] <troker> Hey all, looking for an older copy of ffmpeg (version .60 to be exact) for some research.  Anyone know where I might be able to find it?
[00:45] <Compn> yes
[00:45] <Compn> should be in release dir
[00:45] <Daemon404> http://ffmpeg.org/download.html
[00:45] <Compn> troker : http://ffmpeg.org/releases/
[00:46] <Daemon404> Compn, please do not encourage people to download older stable releases
[00:46] <troker> thanks!
[00:46] <Daemon404> 0.6.6 is compatible with 0.6.0
[00:46] <Daemon404> its the whole point of stable branches.
[00:49] <Compn> bwahaha
[00:49] <troker> Daemon404: Well Im looking for a build with this particular vuln http://www.securityfocus.com/archive/1/archive/1/514009/100/0/threaded -- would that be in 6.1?
[00:49] <troker> (again research, I know vulnerable code is vulnerable) 
[00:50] <Daemon404> man thats super old
[00:50] <Daemon404> before multithreading was even merged
[00:50] <Compn> 2010
[00:50] <Daemon404> the commit hash doesnt even exist
[00:51] <troker> Is the code gone? or is it findable
[00:51] <Daemon404> also
[00:51] <Daemon404> libavcodec <= 0.6
[00:51] <Compn> Daemon404 : see mr smarty pants, if he chose your .6.6 version, its probably the one with the vuln fix
[00:51] <Daemon404> is confusing
[00:51] <Daemon404> because libavcodec version != ffmpeg version
[00:51] <Compn> it means ffmpeg
[00:51] <Compn> in this case
[00:51] <Daemon404> yeah...
[00:51] <Compn> but yes confusing to me :D
[00:52] <troker> Yea I know its old, Im using it to test the methods here http://user.informatik.uni-goettingen.de/~krieck/docs/2011-woot.pdf
[00:52] <Daemon404> http://gitorious.org/~net1945/ffmpeg/net1945s-ffmpeg-mt/commit/16c592155f117ccd7b86006c45aacc692a81c23b
[00:52] <Daemon404> just check out the hash before this one
[00:52] <Compn> troker : try the 0.6.0 , not the 0.6.6 as it could possibly include a fix for that vuln
[00:52] <Compn> :P
[00:52] <Daemon404> or just use the exact hash given in teh cve
[00:52] <Daemon404> and save the hunting trouble
[00:54] <troker> Compn: I see no 0.6.0, do you mean 0.6.release? Or would that be patched
[00:54] <Daemon404> [18:52] <@Daemon404> http://gitorious.org/~net1945/ffmpeg/net1945s-ffmpeg-mt/commit/16c592155f117ccd7b86006c45aacc692a81c23b
[00:54] <Daemon404> [18:52] <@Daemon404> just check out the hash before this one
[00:54] <Daemon404> READ
[00:54] <Daemon404> this is the exact hash given in the cve
[00:54] <Daemon404> which fixes the fuc
[00:54] <Daemon404> er, bug
[00:54] <Daemon404> ergo, checkout the revision before it
[00:54] <Compn> yeah what Daemon404 says is better
[00:55] <Compn> http://ffmpeg.org/releases/ffmpeg-0.6.tar.gz is the .6 one . i dont know if its patched. but the tarball is 6-2010 and the vuln was fixed in 9-2010 so it seems to be right ...
[00:55] <Daemon404> goddammit Compn 
[00:55] <Compn> bwahaha
[00:55] <Daemon404> stop leading people down shit paths
[00:56] <Compn> hes the one trying to root someones box using a malformed file
[00:56] <Compn> and you're helping him
[00:56] <Daemon404> if he was trying to root someone's box, he'd already know the exact version
[00:56] <Daemon404> also people running vuln cod from 2010 have it coming.
[00:56] <Compn> i didnt say he was good at it
[00:56] <Compn> :P
[00:56] <troker> Yea all us security guys are evil. 
[00:57] <Compn> yes, you are
[00:57] <Compn> making ffmpeg, a video encoding project, fix vulns for hours on end when that time could be better spent adding more codecs and features :P
[00:57] <saste> diary of a bug hunter also contains a few nice exploits of ffmpeg vulnerabilities
[00:57] Action: Compn bitter
[00:58] <troker> That paper I linked up there ^ will *hopefully* cut down time spent doing just that
[00:58] Action: Compn also kidding
[00:59] <troker> I agree though! Bug hunting takes far too long! Lets optimize!
[00:59] <troker> With Science!
[00:59] <troker> (")
[00:59] <Compn> what, no cake ?
[00:59] <troker> That happens after you find the bug
[02:10] <cone-855> ffmpeg.git 03Marton Balint 07master:9fac752afa96: ffplay: simplify and fix flushing out old subtitles on seeking
[02:10] <cone-855> ffmpeg.git 03Michael Niedermayer 07master:e4198d2fc3d9: Merge remote-tracking branch 'cus/stable'
[06:16] <anshul> i was trying to solve ticket "https://ffmpeg.org/trac/ffmpeg/ticket/2716" , have more than one solution,plz help
[09:16] <saste> michaelni, any reason why we don't store the timebase in AVFrame?
[10:05] <michaelni> saste, probably no reason
[10:45] <anshul__> is there any spevific format. to send an patch, i feel mine patch is ignored  on ffmpeg-devl mailing list
[10:46] <av500> git format-patch
[10:47] <nevcairiel> you posted your patch 2 hours ago, so .. seriously? complaining already?
[10:47] <nevcairiel> if there is a week without feedback you're allowed to ask
[10:47] <av500> nevcairiel: projects get created, forked and abandoned in less time than that on github :)
[10:48] <anshul__> i posted first time, so dont know what to expect
[10:50] <anshul__> what other people posted, there were lot of thing common. so i thought there wud be an format to send mail. like what is on linux!!!
[10:50] <av500> git format-patch
[10:50] <nevcairiel> its more like git send-email what he means
[10:51] <durandal_1707> anshul__: read http://ffmpeg.org/developer.html
[10:57] <anshul__> got it
[12:02] <xlinkz0> i'm curious to learn how seeking works, how would i find out which function is pointed to by read_seek2 in avformat_seek_file?
[12:03] <av500> depends on the format
[12:03] <xlinkz0> this is from the report on the file i'm testing : [mov,mp4,m4a,3gp,3g2,mj2 @ 0xc08100] Format mov,mp4,m4a,3gp,3g2,mj2 probed with size=2048 and score=100
[12:03] <av500> formats fill out callback functions
[12:03] <av500> libavformat/movdec.c or so
[12:03] <av500> or mov.c
[12:04] <xlinkz0> thanks
[12:05] <wm4> so I posted a patch for something nobody is interested in - how do I get someone to apply it?
[12:11] <saste> wm4, ping it
[12:11] <av500> advertise it
[12:12] <saste> wm4, which patch? the icy thing?
[12:12] <wm4> saste: yes
[12:12] <saste> wm4, i'm going to review it
[12:13] <wm4> thanks
[12:21] <anshul__> i am unable to run ./tests/fate.sh doc/fate_config.sh.template 
[12:21] <anshul__> in my template file i have written 
[12:21] <anshul__> slot=anshul-i386-openSuse-gcc-4.7.2 20130108                                    # some unique identifier
[12:22] <anshul__> but it is saying that "doc/fate_config.sh.template: line 1: anshul-i386-openSuse-gcc-4.7.2: command not found"
[13:19] <wm4> ubitux: is there timed text subtitle support?
[13:19] <wm4> appears it isn't
[13:34] <saste> wm4, no way to read metadata from a protocol?
[13:34] <saste> using options to read metadata is a bit weird
[13:35] <saste> who is our libavformat protocols expert?
[13:35] <wm4> saste: I don't know but I don't think there's metadata
[13:35] <wm4> and in fact, cookies, content type, mime type are all communicated through avoptions
[13:36] <saste> options are used for configuring components
[13:37] <saste> usually they contain user configurations, rather than content information
[13:37] <saste> not that i see a better solution at the moment
[13:39] <wm4> yeah, I don't think this is particularly elegant
[13:39] <wm4> in theory there should be some sort of out-of-band mechanism to transfer metadata, but then it gets too complicated
[13:39] <wm4> especially for something as crappy/obscure as shoutcast metadata
[13:44] <cone-855> ffmpeg.git 03Nigel Touati-Evans 07master:42bd0cd21ae6: Fix copying extradata to codec in mxfdec.c
[13:46] <saste> wm4 i'm not against it, but then i'm no libavformat maintainer
[13:46] <wm4> so who's responsible?
[14:16] <Compn> wm4 : whoever wants to submit a patch :D
[14:16] <wm4> since I came to the conclusion there's no nice way to implement it, I'm fine with my crappy patch
[14:17] <wm4> crappy as in using avoptions to communicate content
[14:17] <Compn> dumps it to a file ?
[14:17] <Compn> or you mean modify metadata from avopts ?
[14:18] <wm4> read
[14:19] <wm4> the server sends the metadata, and I make it available through an avoption entry
[14:19] <Compn> ah
[14:19] <Compn> kind of like mplayers' get_property 
[14:20] <wm4> in mplayer it'd be a STREAM_CTRL
[14:21] <Compn> like get property filename, get property ... metadata  :P
[14:22] <cone-855> ffmpeg.git 03Nigel Touati-Evans 07release/1.1:e2dcc4452dad: Fix copying extradata to codec in mxfdec.c
[14:22] <cone-855> ffmpeg.git 03Nigel Touati-Evans 07release/1.2:fcc6460dbfcc: Fix copying extradata to codec in mxfdec.c
[14:27] <Compn> whoa xvba decoder, nice
[14:35] <JEEB> can someone confirm whether or not this vorbis stream is borked? it used to work before the check for the codebook entry count was added, but now it just errors out when initing the decoder
[14:35] <JEEB> http://www.cccp-project.net/beta/test_files/possibly_broken_vorbis_stream.ogg
[14:35] <JEEB> and the audio seemed to sound OK when decoded
[14:35] <Compn> on new or old version ?
[14:36] <JEEB> pre-may 13th codebook entry count check
[14:36] <JEEB> after that check was added it just seems to fail that check -> doesn't init decoder
[14:36] <JEEB> if the sample's borked then I'll just push that to the person who provided the sample
[14:36] <Compn> it works in mplayer 2012 and ffplay version N-52523-g0fb64da
[14:37] <Compn>  built on Apr 28 2013 00:01:23 with gcc 4.7.3 (GCC)
[14:37] <JEEB> yeah, that's before the check
[14:37] <Compn> although ffplay freezes a few times
[14:37] <Compn> and loops 
[14:37] <JEEB> funky, works here on an april mplayer2 build
[14:37] <Compn> with -demuxer lavf or mplayer demuxer ?
[14:38] <JEEB> mplayer2/mpv use lavf demuxer by default
[14:56] <JEEB> anyways, if anyone has the time to check the stream for correctness, I'd love it <3
[14:58] <wm4> xmms1 plays it
[14:58] <wm4> it appears to use libvorbis (does that include a demuxer??)
[14:59] <JEEB> most probably
[17:28] <cone-855> ffmpeg.git 03Reuben Martin 07master:0de9d3dd1c73: gxf: Added GXF format code 25 which is used for DV codec in HD resolutions
[17:28] <cone-855> ffmpeg.git 03Reuben Martin 07master:7ebf3ab959a0: gxf: Factorize code in get_sindex()
[17:46] <cone-855> ffmpeg.git 03Reuben Martin 07master:c4db4b17569f: Added GXF format code to identify AVC Intra video streams. This was an extension to the origional GXF spec implemented by Grass Valley.
[17:51] <cone-855> ffmpeg.git 03Stefano Sabatini 07master:7eb6eb03d83a: lavc/mpegvideo_enc: simplify timestamp checks in load_input_picture()
[18:15] <cone-855> ffmpeg.git 03Reuben Martin 07master:86190afbda0c: gxf: Added codec ID to playback AVCHD
[18:57] <ubitux> wm4: you can do a av_asprintf() btw
[18:58] <wm4> ok
[18:58] <wm4> ubitux: about the subtitle utf-8 check, it seems nicolas didn't like the idea
[18:58] <ubitux> s->icy_meta_header = av_asprintf("%s%s: %s\n", s->icy_meta_header ? s->icy_meta_header : "", tag, p);
[18:58] <ubitux> or sth like that
[18:58] <ubitux> without the leak
[18:58] <ubitux> (not sure if that's simpler)
[18:59] <ubitux> wm4: yeah, 'told you :p
[18:59] <ubitux> dunno what to say except "i don't like it"
[18:59] <JEEB> ubitux, do you know vorbis :V
[18:59] <ubitux> i don't
[18:59] <JEEB> k
[18:59] <JEEB> will try to wait for someone who might know vorbis and check a sample's validity that used to decode before but now hits a sanity check
[19:02] <wm4> how do I add private options to AVCodecContext? I want an option that can be set from outside, but I don't want to add an AVCodecContext member
[19:12] <wm4> also why does avoptions have this complicated recursive option setting system, if you dump codec specific stuff into AVCodecContext anyway?
[19:18] <Daemon404> JEEB, sanity check as in an assert?
[19:19] <JEEB> a 'jooru' fix http://git.videolan.org/?p=ffmpeg.git;a=commit;h=e6b6ae46951e242d7caf11acf5b1f10f109e1c96
[19:23] <wm4> ubitux: sorry, that's probably a rather basic question, but... how do I set sub_charenc_mode?
[19:23] <wm4> -sub_charenc_mode pre_decoder doesn't work
[19:24] <ubitux> i don't if you can do much with that
[19:24] <ubitux> nicolas made me introduce it for a reason i forgot
[19:24] <ubitux> i thought it was for internal purpose only
[19:24] <ubitux> (but made "public" for interlib com)
[19:25] <wm4> ............
[19:25] <wm4> sometimes I think you don't want applications to use your libs
[19:25] <ubitux> but sorry, i forgot very quickly about stuff
[19:25] <ubitux> and thus i don't remember that
[19:25] <ubitux> so i might be wrong
[19:26] <Compn> why you need to use it pre-decoder ?
[19:26] <Compn> i mean, it doesnt work post decoder ?
[20:27] Action: Daemon404 jumps in to help wm4 
[20:29] Action: ubitux smiles with sympathy at his MUA
[20:30] <Daemon404> i dunno, ive been working with subtitles for 8 years now
[20:31] <Daemon404> i'd like to think my view is worth something.
[20:31] <Daemon404> key bit: as a user
[20:32] <ubitux> i didn't meant that way :)
[20:32] <ubitux> i'm just happy to finally see subtitles talks
[20:32] <ubitux> it's unfortunate i don't give a shit anymore
[20:32] <Daemon404> i blame webtvv
[20:32] <Daemon404> er
[20:32] <Daemon404> webvtt
[20:32] <Daemon404> feces. i meant feces.
[20:33] <wm4> Daemon404: thank you
[21:21] <kwizart> I don't understand the resolution of the libcdio support bug https://ffmpeg.org/trac/ffmpeg/ticket/2614
[21:21] <kwizart> it seems to me that the cdio/cdda.h isn't made available with libcdio-0.90
[21:22] <kwizart> I guess it's on purpose, but lot of projects still use them
[21:29] <wm4> kwizart: from what I remember the default paths were changed around
[21:32] <kwizart> wm4, probably also, or it has moved to another project ?: http://koji.fedoraproject.org/koji/rpminfo?rpmID=3732266
[21:32] <kwizart> because it's not anymore in libcdio
[21:33] <kwizart> http://koji.fedoraproject.org/koji/rpminfo?rpmID=3714393
[21:33] <kwizart> the related cdparanoia package
[23:26] <saste> foo = foo ? : 1;
[23:26] <saste> valid syntax?
[23:28] <ubitux> gnu extension afaik
[23:31] <saste> ubitux, http://en.wikipedia.org/wiki/%3F:#C
[23:31] <saste> yes
[23:31] <ubitux> you have the av_x_if_null() that somehow allows this
[00:00] --- Fri Jun 28 2013