[Ffmpeg-devel-irc] ffmpeg-devel.log.20141206

[burek]

[00:06] <cone-460> ffmpeg.git 03Lukasz Marek 07master:53b2809f2c1b: lavc/options: don't copy priv context when it is not AVOption object
[00:09] <ubitux> http://imgur.com/gallery/iDVTstR
[00:10] <jamrial> heh
[00:39] <cone-460> ffmpeg.git 03wm4 07master:1eec9bfc383f: APIchanges: mark the release 11 branch point
[00:39] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:a0d0799ea2a7: Merge commit '1eec9bfc383f6dca29d83a2bfb45433dd66561c9'
[00:45] <jamrial> btw, fate passes on linux x86_64 with the latest gcc 5 snapshot
[00:46] <iive> are you trying to tell, that there might be gcc x.y.0 that does not miscompile ffmpeg?
[00:47] <jamrial> still four months to go! some bugfix in the meantime may ruin things for us :p
[00:50] <cone-460> ffmpeg.git 03Luca Barbato 07master:1c77ead1fb14: rm: Use the correct codec_data_size signedness
[00:50] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:ef2f9073dda3: Merge commit '1c77ead1fb14d64e6798fc5e32af04abcb3cebb4'
[00:50] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:50f9de59a08f: avformat/rmdec: rm_read_extradata: add error message for oversized extradata
[00:50] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:03abf55f2529: avformat/rmdec: Check for overflow in ff_rm_read_mdpr_codecdata()
[00:57] <ubitux> michaelni: you can remove the "DVBSub error: " in the message if you add the context
[01:01] <cone-460> ffmpeg.git 03Vittorio Giovara 07master:b72727a5248f: lavc: mention that the parser callback never returns an error
[01:01] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:66f379da15ce: Merge commit 'b72727a5248f1ef02db99b378dce1eb48a46357a'
[01:02] <cone-460> ffmpeg.git 03Lukasz Marek 07master:8616c5b75acf: lavc/options: fix leaks in avcodec_copy_context
[01:29] <cone-460> ffmpeg.git 03Matthieu Bouron 07master:8c0a865ad96b: mov: skip version and flags attributes in mov_read_chan()
[01:29] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:95482092b1e5: Merge commit '8c0a865ad96b9e8542051f75b0edc424cb73994e'
[01:36] <cone-460> ffmpeg.git 03Derek Buitenhuis 07master:33c827f632f9: swscale: Properly scale YUV
[01:36] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:e827f65618f6: Merge commit '33c827f632f95ffe3399b695a5a0d47b366b6e20'
[02:00] <cone-460> ffmpeg.git 03Kieran Kunhya 07master:9a738c27dceb: v210enc: Add SIMD optimised 8-bit and 10-bit encoders
[02:00] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:1d048f762d2f: Merge commit '9a738c27dceb4b975784b23213a46f5cb560d1c2'
[02:08] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:ce800d4695b7: avcodec/dvbsubdec: Remove redundant "DVBSub error" prefix
[02:09] <michaelni> ubitux, removed
[02:38] <cone-460> ffmpeg.git 03Christian Suloway 07master:1c639fa6e02e: avformat/hlsenc: added segment filename option
[03:39] <cone-460> ffmpeg.git 03Timothy Gu 07master:cd6f0f28bc06: doc: Do not use the headings as links to TOC anchors
[03:39] <cone-460> ffmpeg.git 03Timothy Gu 07master:4a27f05d1450: doc: html: Use native viewport
[06:25] <cone-460> ffmpeg.git 03Reynaldo H. Verdejo Pinochet 07master:da95fe4ef81c: ffserver_config: reflow _parse_ffconfig()
[06:25] <cone-460> ffmpeg.git 03Reynaldo H. Verdejo Pinochet 07master:ac896f0cd1b1: ffserver_config: reflow _get_arg()
[06:26] <cone-460> ffmpeg.git 03Reynaldo H. Verdejo Pinochet 07master:7cdc99707bf7: ffserver_config: fix line width on fwd declarations
[08:58] <ubitux> http://bellard.org/bpg/ ... 
[09:08] <rcombs> huh, interesting
[10:07] <anshul_mahe> Inside init function of decoder is it safe to use value in context structure without doing memset
[10:19] <anshul_mahe> I got it, there av_mallocz of private data, so need to do it again
[10:20] <nevcairiel> the context should already be zero'ed when the init function runs
[11:07] <funman> [fun at amigo ~/ffmpeg]% git fetch -v                                     
[11:07] <funman> error: Could not read f13f5a7d4b6c57a11c63ba5e4f972f0bff5f306a
[11:07] <funman> then the same error for a bunch of other hashes and finally..
[11:07] <funman> fatal: pack has 166 unresolved deltas
[11:07] <funman> fatal: index-pack failed
[11:07] <funman> git gc fails too
[11:07] <funman> ::/
[11:10] <ubitux> git fsck?
[11:10] <funman> ah forgot about this one
[11:10] <funman> [fun at amigo ~/ffmpeg]% rm -fr .git/
[11:10] <funman> too late I guess ;)
[11:11] <funman> thanks, I'll remember it next time
[11:31] <anshul_mahe> what is nick of  nicolas george
[15:39] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:b4ece5494712: avfilter/avcodec: use NULL instead of 0 for pointer
[15:39] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:1adc1b97e1b8: avcodec/hevc_ps: Add context to av_log() call
[15:39] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:8280b7db4782: avfilter/buffersrc: add context to av_log() call
[15:47] <wm4> rcombs: what was the issue with libavformat seekhead again?
[15:48] <wm4> rcombs: I now noticed that the seekhead field can be MODIFIED while it's being read
[15:48] <wm4> which is complete bullshit
[15:48] <wm4> and causes the hang in #4162
[15:48] <rcombs> wm4: it doesn't parse seekheads recursively
[15:48] <wm4> yes it does
[15:48] <rcombs> or, at least, it didn't
[15:49] <wm4> matroska_parse_seekhead_entry can be called somewhere from matroska_execute_seekhead
[15:49] <wm4> it's not "recursive" of course
[15:49] <wm4> it just overwrites the seekhead... while it's still iterating
[15:50] <rcombs> seekhead handling in general could probably stand do be improved a fair bit
[15:50] <wm4> yes, it doesn't even do it correctly for sane files
[15:51] <rcombs> the high-level issue was that a file with a SeekHead to the Cues placed at the end of the file, and a SeekHead to that SeekHead placed at the start, didn't get its Cues parsed
[15:51] <wm4> and how did we fix it?
[15:51] <rcombs> and that was bad because matroskadec (sometimes?) failed to seek in the absence of Cues
[15:51] <wm4> since seekheads are not actually read recursively
[15:51] <rcombs> I think we didn't
[15:51] <rcombs> I had a patch for it, but I don't think it was applied
[15:52] <wm4> yeah, let's not fix the bug, just paint it over!
[15:52] <rcombs> (and I'm not 100% sure if it was correct)
[15:52] <wm4> does anyone know what happens if ebml_parse reads an element that was already read?
[15:52] <wm4> does it free the old elements, or memory leak?
[15:52] <rcombs> that was the key bit: nobody was quite sure about that at the time
[15:53] <wm4> so the mpv mkv demuxer is still much better than libavformat's, sigh
[15:53] <rcombs> the more important bug (seeking failed altogether in the absence of an index [in some cases?]) was fixed
[15:54] <rcombs> but proper SeekHead-pointing-at-a-SeekHead handling was never implemented
[15:55] <wm4> seems like it calls realloc in this case
[15:57] <rcombs> biggest concern was what might happen if 2 SeekHeads pointed at each other, or some similar loop occurred
[15:57] <wm4> yeah
[15:57] <wm4> best would be having a worklist of elements to read, with flags whether they were already read
[15:57] <wm4> but it's probably hard to do with matroskadec.c's "clever" design
[16:11] <wm4> I see, the code tries to workaround reading previous entries simply by comparing the byte file position
[16:36] <wm4> rcombs: actually I think it does read referenced seekhead entries... pretty confusing, dunno
[16:37] <wm4> rcombs: didn't you have a file that can be used to test this?
[16:49] <JEEB> land the thread goes rolling onl
[17:07] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:7ad742b2247a: avfilter/vf_signalstats: fix avframe leak on error
[17:18] <funman> v210enc.c doesn't check for pic->format == AV_PIX_FMT_NONE
[17:19] <funman> should a check be added to avcodec_encode_video2 ?
[17:19] <funman> I had to fix upipe: https://github.com/funman/upipe/commit/cfad85b4254d6aff2b86c90c2aacb9a0bfea15d3 - not sure if it's doing anything not kosher with AVFrame initialization
[17:25] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:c149f67e9b5b: avformat/apetag: More completely check avio_get_str() return value
[19:12] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:6be36ef3c123: avcodec/libxavs: disable global header code
[20:56] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:a4f387bf56a6: avformat/mpegts: Warn if ffio_ensure_seekback() failed
[20:56] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:890d8f44fdfe: avutil/buffer: add #if USE_ATOMICS around atomics specific functions to ensure sync types arent mixed by mistake
[21:05] <cehoyos> Hi! Is anybody on Debian to test ticket #4164?
[23:33] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:e15b29bb18be: avformat/mpeg: do not count PES packets inside PES packets during probing
[23:33] <cone-390> ffmpeg.git 03Michael Niedermayer 07master:74080ded9630: avformat/mpeg: add commented out av_log() to probe
[00:00] --- Sun Dec  7 2014