[Ffmpeg-devel-irc] ffmpeg-devel.log.20131109

[burek]

[00:20] <cone-424> ffmpeg.git 03Anton Khirnov 07master:d42db44cfe58: avienc: drop the vfr flag.
[00:20] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:9c52ab6f4f32: ffmpeg: treat avi as VFR in framerate conversion code
[00:20] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:c79c7ffc9da7: Merge remote-tracking branch 'qatar/master'
[01:27] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:891f33cc7052: avformat/h261dec: optimize probe
[01:27] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:557571ff7db5: avformat/h261dec: dont accept invalid gobs in probe as valid
[02:02] <ubitux> wm4: any news on the dvd subtitles issue you had?
[02:02] <ubitux> duration or sth
[02:02] <wm4> ubitux: just that the API fails to provide a way to indicate that the duration is unset
[02:03] <ubitux> ok
[02:03] <wm4> it uses 0 for that, so you can't distinguish to subtitles with duration 0 (which can exists too, but maybe only with text subs not image subs)
[02:03] <ubitux> should we add some flag of some sort? or maybe negative duration?
[02:03] <ubitux> (negative duration might break some stuff in users apps)
[02:03] <wm4> a negative value would be best I guess, bit it would be an API change
[02:04] <wm4> also there's a strange additional hunk of code in mplayer's dvd sub parser, but I don't know about this
[02:08] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:64f48169159b: avformat/mp3dec: 4x faster probing
[04:29] <BBB> durandal_1707: if you're interested, I can help you get that going
[04:30] <BBB> durandal_1707: it's slightly more involved, but still, not overly difficult
[04:30] <BBB> durandal_1707: I tried to structure it in such a way that that becomes really easy to add later on
[05:51] <Compn> http://lists.alioth.debian.org/pipermail/pkg-multimedia-maintainers/2012-May/thread.html#26678
[05:51] <Compn> lol
[11:03] <ubitux> can anyone css-add an horizontal scrollbar on fate when unfolding failing tests?
[11:04] <ubitux> (see for instance what happens when unfolding dragonegg-3.4 instance)
[11:14] <michaelni> ubitux, i can apply a patch, i dont know what to add where ...
[11:15] <ubitux> ok let me have a look
[11:19] <ubitux> meh fuck css, overflow:auto doesn't want to work
[11:23] <ubitux> oh well, i'll see later
[12:14] <cone-654> ffmpeg.git 03James Zern 07master:eb7f7b797f0a: riff: add VP9 fourcc
[12:39] <cone-654> ffmpeg.git 03Clément BSsch 07master:dbb41f93c16c: build: avoid stdin stall with GNU AS probing.
[15:35] <cone-654> ffmpeg.git 03Michael Niedermayer 07master:76a47d66c414: avformat/dv: about 3 times faster probe
[16:22] <cone-654> ffmpeg.git 03Michael Niedermayer 07master:9825b99eca0f: avformat/mpegts: do not hardcode type used in memset()
[16:22] <cone-654> ffmpeg.git 03Michael Niedermayer 07master:e624df5dc33f: mpegts: only calculate stat index when its needed during probe
[16:28] <wm4> what's the deal with optimizing probing?
[16:32] <Compn> wm4 : takes a long itme
[17:44] <kierank> need to get rid of these spammers
[18:11] <ods15> michaelni, odd, I never unsubscribed from nut-devel, I did not get those mails
[18:12] <ods15> oops, I know whats wrong, my email address is dead, I am no longer ods15 at ods15.dyndns.org
[18:13] <michaelni> kierank, trac supports several spam filter services, like akismet, typepad, defensio, http:bl. But all these need API keys AFAIK. If someone provides such keys i can enter them in trac, that may (or may not) improve spam filtering
[18:13] <ods15> I can't really say that I am still a maintainer for libnut, but I don't mind approoving these trivial patches, if you don't mind replying in my name in nut-devel
[18:13] <michaelni> ods15, sure, ok, will do
[18:14] <ods15> just - please add parenthesis over entire macro in ftello(a) - should be: #define ftello(a) ((off_t)(_ftelli64(a)))
[18:14] <ods15> patches OK otherwise
[18:28] <ubitux> ._.
[18:33] <michaelni> ods15, patches applied
[18:51] <cone-654> ffmpeg.git 03Michael Niedermayer 07master:ef8f29347285: avformat/dv: consider a header match at byte 0 to count as detection of dv in probe
[18:51] <saste> ubitux: do you know if it is possible to specify the metadata text encoding in a container?
[18:51] <saste> or do we alwasy assume UTF-8?
[18:52] <saste> why there are so many files around with the wrong encoding?
[18:53] <ubitux> demuxers should be able to access sub_charenc option
[18:53] <ubitux> (to be confirmed)
[18:53] <ubitux> we assume utf-8 but user can set it with -sub_charenc
[18:53] <saste> sub_charenc only regards subtitles, right?
[18:53] <ubitux> ah sorry, metadata
[18:53] <ubitux> yes i think we assume utf-8
[18:54] <saste> should we extend it to metadata as well?
[18:54] <ubitux> feel free to do it
[18:54] <ubitux> since we added the iconv thing, it should be doable
[18:54] <saste> also, should we validate string option values?
[18:54] <ubitux> i'm not much into philosophy sorry
[18:55] <saste> right now we can still inject non valid UTF-8 through -metadata and other options
[18:56] <saste> i'm asking because i'm validating the value of -validate_string_policy, but doing it just for an option seems stupid
[18:57] <saste> philosophy? no it's more about opinions
[18:57] <ubitux> it was a joke, i don't have any opinion here :)
[18:57] <ubitux> not enough insight on the subject
[18:58] <saste> i think you thought hard about such stuff, since you're our subtitles guru
[18:58] <ubitux> yes but for charset i'm honestly not the best bet
[18:59] <ubitux> you should probably troll about that with wm4 and nicolas
[18:59] <wm4> <ubitux> demuxers should be able to access sub_charenc option
[18:59] <wm4> <ubitux> (to be confirmed)
[18:59] <wm4> <ubitux> we assume utf-8 but user can set it with -sub_charenc
[18:59] <wm4> I'd fucking hate this
[18:59] <ubitux> wm4: some formats include specify charset
[18:59] <ubitux> at least microdvd does
[19:00] <ubitux> (we don't support this though)
[19:00] <wm4> yes, you want to be able to know what charset it is if it's guaranteed
[19:00] <ubitux> SAMI might as well support something like that
[19:00] <wm4> e.g. webvtt subs are _always_ utf-8, no matter what
[19:00] <wm4> while srt can be anything
[19:00] <ubitux> so what would you hate?
[19:00] <wm4> and it would be nice to know that (without having to consult a whitelist)
[19:00] <wm4> demuxers recoding charset
[19:00] <wm4> like nicolas plans it
[19:01] <ubitux> i'm sure you'll find webvtt not utf-8
[19:01] <wm4> the day it is committed, I'll NIH _all_ subtitle parsers
[19:02] <wm4> ubitux: at least that's a case where you should try not to misdetect subtitles as some other charset
[19:03] <ubitux> i honestly didn't really follow the subtitles charset troll
[19:05] <wm4> it's not a troll, it's a flame
[19:05] <wm4> a very angry one
[19:05] <ubitux> grrr
[19:07] <ubitux> saste: opinion on the fade color patch?
[19:07] <saste> ubitux, i'll review it in a few minutes
[19:08] <cone-654> ffmpeg.git 03Stefano Sabatini 07master:0f8bdfe3e807: ffmpeg: store value of -filter and -filter_script per-stream option in OutputStream struct
[19:08] <cone-654> ffmpeg.git 03Stefano Sabatini 07master:f591ba60ff4b: ffmpeg: generate more specific error message in case of inconsistent filtering options
[20:08] <saste> nobody will comment on my utf8 patch?
[20:53] <cone-654> ffmpeg.git 03Erik Olofsson 07master:f5bffd3a1a1a: libavf/avienc: store language metadata for XSUB streams
[21:00] <BBB> ubitux: can you review+commit that vp9 patch I put up?
[21:08] <ubitux> BBB: don't want to answer reimar question(s)?
[21:09] <ubitux> i don't mind committing but i can't really review blindly this
[21:25] <BBB> ubitux: it's a question for google, I can't answer it, right?
[21:26] <ubitux> maybe Reimar isn't aware you're not working at google anymore
[21:26] <ubitux> ...or that you have no contact with libvpx (anymore?)
[21:26] <BBB> hm, I see
[21:28] <BBB> well that should clear things up then for him
[21:35] <ubitux> BBB: applied
[21:35] <cone-654> ffmpeg.git 03Ronald S. Bultje 07master:dd3d0f46bdb0: vp9: fix mvref finding to adhere to bug in libvpx.
[00:00] --- Sun Nov 10 2013