[Ffmpeg-devel-irc] ffmpeg-devel.log.20141108

[burek]

[00:32] <cone-338> ffmpeg.git 03Martin Storsjö 07master:1d8a0c1b43e5: movenc: Allow to request not to use edit lists
[00:32] <cone-338> ffmpeg.git 03Michael Niedermayer 07master:2893d1b36d17: Merge commit '1d8a0c1b43e58332a3a15c67d4adc161713cade8'
[00:41] <cone-338> ffmpeg.git 03Martin Storsjö 07master:e7d20f12c5ef: movenc: Remove a now redundant check
[00:41] <cone-338> ffmpeg.git 03Michael Niedermayer 07master:872c0bcd3c04: Merge commit 'e7d20f12c5eff5570cd897f3ce3a88456024036b'
[01:00] <cone-338> ffmpeg.git 03Carl Eugen Hoyos 07master:e2e36a739d3a: Fix make checkheaders for libavcodec/vorbis_parser.h.
[01:11] <cone-338> ffmpeg.git 03Martin Storsjö 07master:9cbf70fa0e44: movenc: Write correct presentation timestamps in tfra
[01:11] <cone-338> ffmpeg.git 03Michael Niedermayer 07master:c9a9a7a1fd7e: Merge commit '9cbf70fa0e44613590b019cef1fe99aa3f3c5d9d'
[01:43] <cone-338> ffmpeg.git 03Xiaohan Wang 07master:33301f001747: Fix read-after-free in matroska_read_seek().
[03:05] <cone-338> ffmpeg.git 03Lukasz Marek 07master:2b14593148e0: ffmpeg_opt: free incorrect priv_data for feed stream
[10:56] <arwa> Can anyone help me out with color blending masks for RGB32 format?
[11:03] <cone-779> ffmpeg.git 03Martin Storsjö 07master:95a449d3ce8e: movenc: Remove an outdated comment
[11:03] <cone-779> ffmpeg.git 03Michael Niedermayer 07master:0bdc5db520ba: Merge commit '95a449d3ce8e15522df47a80a8a4593ea5c2b1bb'
[11:16] <cone-779> ffmpeg.git 03Martin Storsjö 07master:8cb7b7b461b5: movenc: Avoid leaking locally allocated data when returning on errors
[11:16] <cone-779> ffmpeg.git 03Michael Niedermayer 07master:4342b346d273: Merge commit '8cb7b7b461b52898765b38e3eff68c0ce88347f3'
[12:05] <Holden> cehoyos, pretty much since that commit (Jan 2014)
[12:05] <cehoyos> That commit?
[12:06] <Holden> cehoyos, http://git.videolan.org/?p=ffmpeg.git;a=commit;h=b1ad9312331759679a9c956233716a67ae681d89
[12:07] <cehoyos> So it is impossible to fix both bugs?
[12:10] <Holden> with both you mean the ones reported in the ticket? if so, yes
[12:12] <cehoyos> The bug that b1ad931 tried to fix and the one in ticket 3517
[12:12] <Holden> I could not reproduce the one federico reported on the commit message, but in any case that change does not seem correct to me, so it's better to revert
[12:12] <cehoyos> Of course, I just wanted to ask.
[12:13] <Holden> the bug that b1ad931 tried to fix will probably still be there, but I need more info or a way to reproduce to understand why it happens
[12:14] <cehoyos> Perhaps this should be added to the commit message?
[12:16] <Holden> sure
[12:26] <cone-779> ffmpeg.git 03Martin Storsjö 07master:9a5ac36b69ed: movenc: Require samples before trying to write edts
[12:26] <cone-779> ffmpeg.git 03Michael Niedermayer 07master:9ebfe38f3873: Merge commit '9a5ac36b69ede4563e9ecd734141b12ea3280fbc'
[12:41] <cone-779> ffmpeg.git 03Carl Eugen Hoyos 07master:4436a8f44ded: Remove fminf() emulation.
[12:41] <cone-779> ffmpeg.git 03Michael Niedermayer 07master:b0ed88b4c0e9: Merge remote-tracking branch 'cehoyos/master'
[12:48] <cone-779> ffmpeg.git 03Carl Eugen Hoyos 07release/2.4:2be7d565bb2a: Remove fminf() emulation.
[12:52] <cone-779> ffmpeg.git 03Michael Niedermayer 07master:6f21fb793238: Revert "v4l2: setting device parameters early"
[12:55] <cehoyos> Should this commit get backported?
[13:21] <Holden> cehoyos, yes, if possible it should be backported
[14:27] <cehoyos> Holden: Done
[14:28] <cone-779> ffmpeg.git 03Michael Niedermayer 07release/2.2:418e9a6113a7: Revert "v4l2: setting device parameters early"
[14:28] <cone-779> ffmpeg.git 03Michael Niedermayer 07release/2.3:b152305bb34f: Revert "v4l2: setting device parameters early"
[14:28] <cone-779> ffmpeg.git 03Michael Niedermayer 07release/2.4:56e11cbe3237: Revert "v4l2: setting device parameters early"
[15:02] <Holden> cehoyos, thanks
[16:11] <ChenFang> I'm writing a demuxer for Android with ffmpeg for mpeg ts files. And I found a problem that I cannot read the first keyframe after start or seek. anyone known why?
[16:30] <ePirat> hello, would it be possible to make the content-type of output format avaliable to a protocol?
[16:51] <kierank> wtf http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavfilter/vf_interlace.c;h=e07963f79a4688410f07be836649553b0a2dbfb9;hb=HEAD#l207
[16:52] <wm4> lol
[16:53] <kierank> never trust a pts
[18:58] <ePirat> any dev here?
[18:59] <rcombs> nope, nobody here but us chickens
[18:59] <iive> ePirat: don't ask to ask, just ask.
[18:59] <ePirat> I already did
[19:00] <iive> well, you got the answer too.
[19:00] <ePirat> hmm I got nothing& maybe due to connection to my znc timed out& sorry then if I missed it
[19:01] <iive> <rcombs> nope, nobody here but us chickens
[19:01] <ePirat> I meant: [16:30:17]  <ePirat>	hello, would it be possible to make the content-type of output format avaliable to a protocol?
[19:02] <iive> well, I guess nobody knowns...
[19:02] <rcombs> as in, MIME type?
[19:02] <ePirat> yes
[19:03] <ePirat> would be good for http protcol and would increase usability of icecast protocol a lot
[19:03] <rcombs> I don't think ffmpeg uses MIME types for much of anything
[19:03] <ePirat> most supports requests we get are due to wrong or not set mimetype
[19:04] <rcombs> (or libavformat in particular)
[19:04] <rcombs> but apparently it does at least have them listed
[19:08] <michaelni> it should be possible to add a mimetype field to URLContext and initialize that
[19:09] <rcombs> HTTP already supports it via AVOptions
[19:11] <michaelni> each protocol which wants it also could have a private context with a mimetype instead thats set for the muxing case via AVOptions
[19:11] <rcombs> could have ffmpeg_opt.c automatically set that option if not set via CLI
[19:12] <michaelni> something in libavformat would be better than ffmpeg so it works for all applications
[19:18] <rcombs> so, it'd need to be provided by avio_open2
[21:44] <cone-990> ffmpeg.git 03Michael Niedermayer 07master:29b1af40f3a0: avformat/mpegts: add scan_all_pmts option
[21:44] <cone-990> ffmpeg.git 03Michael Niedermayer 07master:41ad87ad8efe: avformat/mpegts: improve first valid PMT heuristic
[21:46] <lukaszmluki> dorek666
[22:18] Action: Daemon404 wonders why 4436a8f44dedc83767b3d9da9beb85d1fae2ca30 wasnt on the ML
[22:22] <cehoyos> Do you want me to revert?
[22:23] <Daemon404> considering it was only recently added, i assume there was actually a reason
[22:24] <cehoyos> Yes, the reason was code that was removed the very same day the workaround for code added a day before was removed: 867c02acd
[22:26] <Daemon404> well ok
[22:26] Action: Daemon404 notes none of this stuff is mentioned in any of the commit messages for any of the related commits
[22:29] <cehoyos> I tried hard to add all relevant information to my commit...
[22:29] <cehoyos> (I edited it twice.)
[22:30] <cehoyos> michaelni: Shouldn't ffmpeg (the application) set "-scan_all_pmts 1"?
[22:32] <michaelni> cehoyos, that might make sense, yes, want to post a patch ?
[22:33] <cehoyos> I will try;-)
[23:18] <cehoyos> michaelni: I added "av_dict_set(&o->g->format_opts, "scan_all_pmts", "1", 0);" to open_input_file() and this works to set the option but how can I find out if the user had already set the option?
[23:24] <cehoyos> Solved using AV_DICT_DONT_OVERWRITE.
[23:56] <cone-990> ffmpeg.git 03Michael Niedermayer 07master:f9fa560597cf: avcodec/aacenc: check input for NaN
[23:58] <cehoyos> michaelni: I don't think that was ticket 3762;-)
[23:58] <michaelni> argh
[23:58] <michaelni> 4084
[23:59] <cehoyos> Shouldn't the encoder ignore the data instead of failing?
[23:59] <cehoyos> Please ignore the patch on the mailing list, I will send an update.
[00:00] --- Sun Nov  9 2014