[Ffmpeg-devel-irc] ffmpeg-devel.log.20120816

[burek]

[01:50] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r6ecf1eb0dd 10ffmpeg/libavcodec/snow.h: 
[01:50] <CIA-40> ffmpeg: snowdec: add 2 av_asserts to check run validity
[01:50] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:50] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r4a596d19ff 10ffmpeg/libavcodec/snow.h: 
[01:50] <CIA-40> ffmpeg: snowdec: check log in get_symbol2()
[01:50] <CIA-40> ffmpeg: Fixes Ticket1635
[01:50] <CIA-40> ffmpeg: Found-by: Piotr Bandurski <ami_stuff at o2.pl>
[01:50] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:36] <burek> there is a little error in:  ffmpeg -formats | grep output
[02:36] <burek> they all show flags 'D' and 'E'
[02:36] <burek> but in the format name, the word "output" is present
[02:36] <burek> which is kinda misleading
[02:37] <burek> because (due to 'D' flag) it can also be an input too
[03:05] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r747774ece9 10ffmpeg/libavcodec/utils.c: 
[03:05] <CIA-40> ffmpeg: av_get_audio_frame_duration: fix ra144
[03:05] <CIA-40> ffmpeg: Fixes Ticket1612
[03:05] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:19] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rf0896a6bd9 10ffmpeg/libavcodec/sp5xdec.c: 
[03:19] <CIA-40> ffmpeg: sp5xdec: fix off by 1 error causing a crash
[03:19] <CIA-40> ffmpeg: Fixes Ticket1633
[03:19] <CIA-40> ffmpeg: Found-by: Piotr Bandurski <ami_stuff at o2.pl>
[03:19] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:36] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * ra39c5c4c6b 10ffmpeg/libavformat/dv.c: 
[04:36] <CIA-40> ffmpeg: dv: fix null ptr dereference
[04:36] <CIA-40> ffmpeg: Fixes Ticket1640
[04:36] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:56] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r11f8f2c0f3 10ffmpeg/libavformat/ (avformat.h utils.c): 
[04:56] <CIA-40> ffmpeg: avformat: export queue_attached_pictures
[04:56] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:56] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rc151e0ca15 10ffmpeg/ffplay.c: 
[04:56] <CIA-40> ffmpeg: ffplay: reque attached pictures on channel changes.
[04:56] <CIA-40> ffmpeg: Fixes Ticket1473
[04:56] <CIA-40> ffmpeg: Reviewed-by: Marton Balint <cus at passwd.hu>
[04:56] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:56] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r484aec4da3 10ffmpeg/LICENSE: (log message trimmed)
[04:56] <CIA-40> ffmpeg: LICENSE: correct incorrect claims on *GPL incompatibility
[04:56] <CIA-40> ffmpeg: There are currently no statements from any person qualified in the art
[04:56] <CIA-40> ffmpeg: of law about the LGPL compatibility of the libfaac license and others.
[04:56] <CIA-40> ffmpeg: Furthermore their is no consensus about their compatibility amongth
[04:56] <CIA-40> ffmpeg: people not qualified in the art.
[05:05] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rb1b84b068d 10ffmpeg/libavformat/version.h: 
[05:05] <CIA-40> ffmpeg: lavf: minor bump for avformat_queue_attached_pictures()
[05:05] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:03] <CIA-40> ffmpeg: 03Philip Langdale 07master * r6af680fa07 10ffmpeg/libavcodec/ (Makefile allcodecs.c avcodec.h srtdec.c): (log message trimmed)
[07:03] <CIA-40> ffmpeg: srtdec: Add timing-less "subrip" decoder.
[07:03] <CIA-40> ffmpeg: After various discussions, we concluded that, amongst other things,
[07:03] <CIA-40> ffmpeg: it made sense to have a separate subrip decoder that did not use
[07:03] <CIA-40> ffmpeg: in-band timing information, and rather relied on the ffmpeg level
[07:03] <CIA-40> ffmpeg: timing.
[07:03] <CIA-40> ffmpeg: As this is 90% the same as the existing srt decoder, it's implemented
[07:03] <CIA-40> ffmpeg: 03Philip Langdale 07master * r6057de19b5 10ffmpeg/ (8 files in 5 dirs): 
[07:03] <CIA-40> ffmpeg: srtenc: Add timing-less "subrip" encoder.
[07:03] <CIA-40> ffmpeg: Unsurprisingly, if a timing-less subrip decoder is desireable, an
[07:03] <CIA-40> ffmpeg: encoder is as well. With this in place, we can move on to remove
[07:03] <CIA-40> ffmpeg: the use of the old encoder/decoder with embedded timing and move
[07:03] <CIA-40> ffmpeg: all timing handling the (de)muxer where they belong.
[07:03] <CIA-40> ffmpeg: Signed-off-by: Philip Langdale <philipl at overt.org>
[08:39] <thresh> hi
[08:39] <thresh> config.mak:2766: *** extraneous `endif'.  Stop.
[08:39] <thresh> on current ffmpeg
[08:39] <thresh> when issuing "make"
[08:48] <ubitux> thresh: no issue here, and fate green
[08:48] <ubitux> thresh: you should have a ifndef FFMPEG_CONFIG_MAK on top of that file
[08:48] <thresh> I do
[08:49] <thresh> I have two ifndef and three endifs
[08:49] <thresh> my configure line is: ./configure --disable-doc --enable-debug --enable-gpl --disable-avdevice --disable-devices --disable-filters --enable-ffplay --enable-libmp3lame --enable-libvpx --target-os=linux --enable-pic --enable-pthreads --enable-static --disable-shared --enable-asm --enable-ffmpeg
[08:50] <ubitux> with the same cmd line it seems ok to me
[08:50] <ubitux> only two endif
[08:50] <thresh> wtf then
[08:51] <ubitux> (one ifndef/endif for MAIN_MAKEFILE on top, and one global ifndef/endif with FFMPEG_CONFIG_MAK)
[08:51] <thresh> yup
[08:51] <thresh> looks like a heisenbug. works fine now :/
[08:58] <thresh> also, libvpxenc is broken: http://pastebin.com/raw.php?i=akt40p78 and crashes ffmpeg / vlc
[09:01] <ubitux> confirmed
[09:02] <ubitux> caused by the -vb 0
[09:02] <thresh> sure
[09:02] <thresh> that's the default bitrate we pass from VLC to lavc encoders
[09:07] <ubitux> is this a regression?
[09:07] <thresh> I don't have older VLC (read: older ffmpeg versions) to test
[09:07] <ubitux> these div by bitrate were added mid 2010
[10:20] <ubitux> meh ffplay doesn't support animated gif :(
[10:21] <CIA-40> ffmpeg: 03Carl Eugen Hoyos 07master * re701b0cf75 10ffmpeg/libavformat/lxfdec.c: 
[10:21] <CIA-40> ffmpeg: Support reading 16-channel lxf files.
[10:21] <CIA-40> ffmpeg: Fixes ticket #1608.
[10:21] <CIA-40> ffmpeg: Reported, analyzed and tested by Gabriel Gerard.
[10:58] <ubitux> ffmpeg -f lavfi -i testsrc -vf 'select=not(mod(n\,1000))' -frames:v 5 out%d.png
[10:59] <ubitux> something looks wrong here
[10:59] <ubitux> i got the first frame, then 4x the same image
[11:00] <ubitux> OTOH, this works fine: ffmpeg -f lavfi -i testsrc -vf 'select=not(mod(n\,1000)),tile=5x1' -frames:v 1 out.png
[11:08] <ubitux> anyway, reported.
[11:16] <j-b> 'morning
[15:03] <CIA-40> ffmpeg: 03Stefano Sabatini 07master * r419e1b7463 10ffmpeg/libavfilter/vf_hue.c: 
[15:03] <CIA-40> ffmpeg: lavfi/hue: apply misc fixes to default values setting
[15:03] <CIA-40> ffmpeg: In particular: signal error in case of bogus h:s values.
[15:03] CTCP re:  from CIA-40 (CIA-40!~CIA at 204.152.223.100) to #ffmpeg-devel
[15:03] <CIA-40> ffmpeg: 03Jemy Tran 07master * r191b77eb01 10ffmpeg/ (doc/filters.texi libavfilter/version.h libavfilter/vf_hue.c): 
[15:03] <CIA-40> ffmpeg: lavfi/hue: add named options support
[15:03] <CIA-40> ffmpeg: Old syntax has been kept for compatibility reasons.
[15:03] <CIA-40> ffmpeg: Signed-off-by: Stefano Sabatini <stefasab at gmail.com>
[15:03] <CIA-40> ffmpeg: 03Stefano Sabatini 07master * r5b0151a474 10ffmpeg/libavfilter/vf_hue.c: lavfi/hue: reindent after previous commits
[15:03] <CIA-40> ffmpeg: 03Stefano Sabatini 07master * r208d6e6615 10ffmpeg/libavfilter/vf_hue.c: lavfi/hue: add verbose debug message showing initial parameters
[15:04] <CIA-40> ffmpeg: 03Stefano Sabatini 07master * r6efce3a894 10ffmpeg/ (5 files in 3 dirs): 
[15:04] <CIA-40> ffmpeg: lavfi/mp: remove hue wrapped filter
[15:04] <CIA-40> ffmpeg: hue was integrated as a native libavfilter filter.
[15:06] <ubitux> saste: i wonder if we couldn't get rid of hqdn3d and tile as well
[15:07] <saste> ubitux: we need to check that it is bit-exact, and that it is no slower
[15:07] <saste> given that hqdn3d has had much more development than the mp corresponding filter, I suppose we could just drop it
[15:07] <saste> tile...
[15:08] <saste> it was sightly different from our tile with regards to the padding
[15:08] <saste> so we may consider to extend ours before to drop it
[15:08] <ubitux> ok
[15:08] <ubitux> there is a lot of telecine related filter :p
[15:08] <saste> there are many other filters which could be dropped, but I didn't for a reason or another
[15:09] <saste> for example mp=tinterlace is faster than tinterlace (altough broken)
[15:09] <saste> and I wanted to see how I could optimize it
[15:09] <ubitux> pullup, decimate, (field?), filmdint, (kern/mc deint?), softpulldown, telecine, tinterlace, ...
[15:10] <saste> tinterlace was already ported (but see above)
[15:10] <saste> field is a 10 minutes hack
[15:10] <ubitux> mmh :)
[15:10] <saste> decimate: there is a already a patch for that
[15:10] <saste> but there was some discussion with nicolas (about its possible simplification due to the poll API work)
[15:11] <saste> about the others I think nyuhu could have a look at them
[15:11] <ubitux> would be great ;)
[15:11] <saste> also i see there are a lot of eq variants
[15:12] <ubitux> yeah, an equalizer
[15:12] <saste> i have already ported eq (but never posted it, since I had to test/benchmark it)
[15:12] <ubitux> speaking of this
[15:12] <ubitux> i'm going to start an histogram filter
[15:12] <saste> eq or eq2?
[15:12] <saste> ubitux: please check histeq
[15:12] <saste> in the archive
[15:12] <ubitux> :o
[15:12] <saste> also speak with nyuhu, an histogram filter is in his todo
[15:13] <ubitux> oh :(
[15:13] <ubitux> ok :(
[15:13] <saste> why ":(" and not ":)"?
[15:13] <ubitux> cause i wanted to do it :D
[15:14] <ubitux> http://ffmpeg.org/pipermail/ffmpeg-devel/2010-November/098340.html
[15:14] <ubitux> nice... :)
[15:14] <ubitux> you wrote a lot of things
[15:14] <ubitux> but never upstream :(
[15:14] <saste> well my histeq is broken
[15:16] <saste> ubitux: http://pastebin.com/8VzM5e46
[15:16] <ubitux> hahaha
[15:16] <saste> yeah hacking is easy, polishing and testing and fixing is boring
[15:17] <ubitux> :)
[15:17] <ubitux> some very interesting stuff in there
[15:17] <ubitux> :D
[15:22] <ubitux> ok well, i'll postpone the histogram filter then
[15:22] <ubitux> got another huge project i'll focus on in the next weeks
[15:39] <saste> ubitux: any comment on the filter options stuff?
[15:40] <saste> it is ticket #1599
[15:41] <ubitux> "lavfi: add flags to filter specific options" ?
[15:41] <saste> yes
[15:41] <ubitux> give me a few minutes, i'll have a look
[15:42] <saste> i'm going to rebase/cleanup and push it tomorrow, together with the other related class-split patches
[16:02] <ubitux> saste: well it looks fine to me; i have no opinion on the A|F vs FLAGS
[16:02] <saste> i'll go for FLAGS since it will require more manual labour in case of changes
[16:04] <saste> michaelni: ^^ same for you, since the change will require more merging effort
[16:20] <michaelni> saste, "i'll go for X since it will require MORE manual labour" <--- ?? ;)
[16:20] <saste> michaelni: yes i am a masochist
[16:21] <saste> michaelni: it was inverted logic
[16:21] <saste> FLAGS needs to be changed in a single line, vs. changing each option line
[16:22] <saste> so it should require *less* manual work
[16:24] <CIA-40> ffmpeg: 03Samuel Pitoiset 07master * r93f257db6b 10ffmpeg/ (3 files in 2 dirs): 
[16:24] <CIA-40> ffmpeg: rtmp: Automatically compute the hash for SWFVerification
[16:24] <CIA-40> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[16:24] <CIA-40> ffmpeg: 03Martin Storsjö 07master * r70766c2182 10ffmpeg/ (5 files in 5 dirs): 
[16:24] <CIA-40> ffmpeg: Add some more missing includes after removing the implicit common.h
[16:24] <CIA-40> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[16:24] <CIA-40> ffmpeg: 03Martin Storsjö 07master * r33e112847d 10ffmpeg/ (4 files in 4 dirs): 
[16:24] <CIA-40> ffmpeg: Add more missing includes after removing the implicit common.h
[16:24] <CIA-40> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[16:24] <CIA-40> ffmpeg: 03Martin Storsjö 07master * r1d9c2dc89a 10ffmpeg/ (238 files in 11 dirs): 
[16:24] <CIA-40> ffmpeg: Don't include common.h from avutil.h
[16:24] <CIA-40> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[16:24] <CIA-40> ffmpeg: 03Diego Biurrun 07master * r6961bdface 10ffmpeg/libavcodec/x86/ (Makefile pngdsp-init.c proresdsp-init.c vp8dsp-init.c): x86: avcodec: Consistently name all init files
[16:24] <CIA-40> ffmpeg: 03Diego Biurrun 07master * rc292f6a24d 10ffmpeg/ (configure libavcodec/Makefile): 
[16:25] <CIA-40> ffmpeg: build: Factor out error resilience dependencies to CONFIG_ERROR_RESILIENCE
[16:25] <CIA-40> ffmpeg: A new hidden config variable is added for the codecs that depend on
[16:25] <CIA-40> ffmpeg: the error resilience parts.
[16:25] <CIA-40> ffmpeg: 03Diego Biurrun 07master * rf704a079a1 10ffmpeg/ (configure libavcodec/Makefile): 
[16:25] <CIA-40> ffmpeg: build: Factor out rangecoder dependencies to CONFIG_RANGECODER
[16:25] <CIA-40> ffmpeg: A new hidden config variable is added for the codecs that depend on
[16:25] <CIA-40> ffmpeg: the rangecoder parts.
[16:25] <CIA-40> ffmpeg: 03Martin Storsjö 07master * r232e35de2e 10ffmpeg/libavcodec/sparc/simple_idct_vis.c: 
[16:25] <CIA-40> ffmpeg: Fix even more missing includes after the common.h removal
[16:25] <CIA-40> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[16:25] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rc581cb4e4f 10ffmpeg/: (log message trimmed)
[16:25] <CIA-40> ffmpeg: Merge remote-tracking branch 'qatar/master'
[16:25] <CIA-40> ffmpeg: * qatar/master:
[16:25] <CIA-40> ffmpeg:  Fix even more missing includes after the common.h removal
[16:25] <CIA-40> ffmpeg:  build: Factor out rangecoder dependencies to CONFIG_RANGECODER
[16:25] <CIA-40> ffmpeg:  build: Factor out error resilience dependencies to CONFIG_ERROR_RESILIENCE
[16:25] <CIA-40> ffmpeg:  x86: avcodec: Consistently name all init files
[17:04] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rd472453efd 10ffmpeg/libavcodec/libvpxenc.c: 
[17:04] <CIA-40> ffmpeg: libvpxenc: fix division by zero
[17:04] <CIA-40> ffmpeg: Reported-by: thresh
[17:04] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:04] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r75cd9a62f4 10ffmpeg/libavcodec/libvpxenc.c: 
[17:04] <CIA-40> ffmpeg: libvpxenc: check for odd RC parameter combinations that could crash
[17:04] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:34] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r7cb9f1a8d0 10ffmpeg/libavcodec/x86/idct_sse2_xvid.c: 
[17:34] <CIA-40> ffmpeg: idct_sse2_xvid: remove libavutil/internal.h include
[17:34] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:34] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r501b681d95 10ffmpeg/libavcodec/x86/lpc_mmx.c: 
[17:34] <CIA-40> ffmpeg: lpc_mmx: remove libavutil/internal.h include
[17:34] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:35] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * ra3ececf329 10ffmpeg/libavfilter/x86/yadif.c: 
[17:35] <CIA-40> ffmpeg: yadif: remove libavutil/internal.h include
[17:35] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:35] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r191ffc7fe7 10ffmpeg/libavcodec/x86/mlpdsp.c: 
[17:35] <CIA-40> ffmpeg: x86/mlpdsp: remove libavutil/internal.h include
[17:35] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:35] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r64b23d7dec 10ffmpeg/libavcodec/x86/motion_est_mmx.c: 
[17:35] <CIA-40> ffmpeg: x86/motion_est_mmx: remove libavutil/internal.h include
[17:35] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:35] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r9bfeaf6f10 10ffmpeg/libavcodec/x86/simple_idct_mmx.c: 
[17:35] <CIA-40> ffmpeg: simple_idct_mmx: remove libavutil/internal.h include
[17:35] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:35] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rbb46b9a36f 10ffmpeg/libavcodec/x86/vc1dsp_mmx.c: 
[17:36] <nevcairiel> someone jumped the include bandwagon
[17:37] <ubitux> saste: what do you propose in place of "windowing"?
[17:38] <saste> ubitux: what does it represent?
[17:38] <saste> write it in plane english, and compress it :)
[17:38] <ubitux> windowing buffer?
[17:38] <ubitux> "Window function LUT"
[17:40] <saste> windowing_buf, window_lut whatever seems more descriptive/appropriate/consistent with the rest
[17:41] <ubitux> ok
[17:46] <cbsrobot> is there any reason almost all av_mallocs in swscale are unchecked ?
[17:46] <ubitux> consistency ;)
[18:13] <nevcairiel> better fix those checked allocs then :)
[18:13] <CIA-40> ffmpeg: 03Nicolas George 07master * r584acbf524 10ffmpeg/libavfilter/src_movie.c: 
[18:13] <CIA-40> ffmpeg: lavfi/movie: set pkt->stream_index for flushing.
[18:13] <CIA-40> ffmpeg: Otherwise, the value of stream_index is kept from the last packet
[18:13] <CIA-40> ffmpeg: read, which is not necessarily the stream that needs flushing.
[18:43] <CIA-40> ffmpeg: 03Giorgio Vazzana 07master * re6ee58fae6 10ffmpeg/libswscale/x86/rgb2rgb_template.c: 
[18:43] <CIA-40> ffmpeg: libswscale: fix #endif comments in rgb2rgb_template.c
[18:43] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:17] <llogan> michaelni, Compn: i'm going to be gone 18-23 August, so one of you will have to moderate the mailing lists then.
[19:28] <Compn> llogan : probably the spam will sit there for 5 days :)
[19:28] <Compn> ehe
[19:34] <llogan> Compn: there have been a good number of valid non-member messages lately. 5 this morning.
[19:34] <llogan> (in ffmpeg-user)
[20:22] <philipl> michaelni: Did you see my last email to the list about changing the reported CODEC_ID for text subtitles coming out of matroska?
[20:40] <michaelni> philipl, is any ffmpeg user using these subtitle ids ?
[20:40] <philipl> mplayer.
[20:40] <philipl> at least.
[20:40] <Compn> mplayer uses lavf demuxer to show those ids i think
[20:41] <michaelni> hmm, i would suggest to ask reimar
[20:41] <michaelni> because he would be most affected by this
[20:41] <Compn> i dont think mplayer has a problem with it being changed
[20:41] <philipl> Well, obviously I will fix mplayer, but I was more asking what does this mean in terms of the API version.
[20:41] <philipl> Is this a major bump?
[20:42] <michaelni> if ID_TEXT was a bug then its a bugfix, otherwise one could argue it would need a major bump
[20:43] <michaelni> but if nothing uses it / all users agree it should just be changed then its also ok for sure
[20:45] <philipl> It's a little philosophical - every media player understood that the text could be using SubRip tags but it was marked ID_TEXT. The practical problem is that if they're using ID matching, that code will fail silently if you rev ffmpeg, which sounds like a major bump requirement to me.
[20:47] <CIA-40> ffmpeg: 03Nicolas George 07master * r12d2bbf324 10ffmpeg/libavfilter/src_movie.c: lavfi/movie: extend a doxygen comment.
[20:49] <michaelni> philipl, is there an alternative to changing it ? I dont think its a really good idea to block a large part of the subtitle work for a long time
[20:50] <philipl> The alternative is to retroactively declare ID_TEXT to really mean Subrip tags.
[20:50] <philipl> We could abuse the enum and reuse the ordinal for ID_SUBRIP and make a new ID_TEXT.
[20:51] <philipl> That would retain binary compatibility but apps would still need source-code updates.
[20:56] <Compn> philipl : whats the actual thing that needs fixing ?
[20:57] <Compn> that people are storing image subtitles as id_text ?
[20:57] <philipl> No.
[20:57] <michaelni> philipl, i suggest you do what you consider more correct/sane while looking forward. Either way may cause minor annoyances to existing code.
[20:57] <philipl> The problem is that ffmpeg tags srt subtitles coming out of matroska containers as CODEC_ID_TEXT.
[20:57] <philipl> But they are srt subtitles.
[20:58] <philipl> The wrong ID prevents us from being able to run the text into the SRT decoder, so you can't properly remux the subtitles into a new container.
[20:58] <Compn> isnt text an all encompassing word for text subtitles ?
[20:58] <Compn> ah
[20:58] <philipl> (or do format conversions that preserve tags)
[20:58] <Compn> so your patch just changes mkv demuxer ?
[20:59] <philipl> Yes. It changes the demuxer to report the right ID, but that has fallout for the media players, who will have code to treat the stream correctly.
[20:59] <philipl> I've looked at mplayer and I know it will fail silently.
[20:59] <Compn> codec_id_text was designed before we had a subrip decoder :P
[20:59] <michaelni> ubitux, ticket/1617 is a regression since 8f3eebd
[21:00] <philipl> yeah, and amusingly we don't have an ID_TEXT encoder or decoder.
[21:00] <michaelni> according to carl at least
[21:31] <ubitux> michaelni: mmh ok quite old, i certainly did braindead things at that time; i'll try to have a look
[21:32] <michaelni> thx
[21:48] <jesk> sorry for asking this here as it's not really a developer question...
[21:48] <jesk> should it be possible to simple stream-copy mkv to mp4 *including* subs?
[21:49] <jesk> all streams work but when including also the sub it interrupts with errors
[21:49] <michaelni> jesk, everything should be possible :)
[21:51] <jesk> michaelni, i'am not familar with subs so didn't know if there is even support for this kind of sub transcoding
[21:51] <ubitux> it should but iirc there is some issues; philipl might know better about that
[21:51] <ubitux> iirc it was one of the limitation but maybe it got fixed?
[21:52] <philipl> jesk: That's pretty much the punchline of the current problem.
[21:52] <philipl> Everything I'm working on right now is to get us to where you can stream copy mkv to mp4.
[21:53] <jesk> that sound like a lot of todo :)
[21:53] <philipl> Well, it's two basic problems at this point. 1) Getting the right codec id. 2) Getting the timing information into the right place.
[21:53] <philipl> Those are both on the matroska side.
[21:54] <ubitux> re-encoding the subtitle should work, right?
[21:54] <philipl> Yes. Given a subrip source, it will work.
[21:54] <philipl> but there are no working sources today.
[21:55] <philipl> Oh, well, I think srtdec works correctly today.
[21:55] <ubitux> ah right mp4 doesn't support ass mmh
[21:55] <jesk> so no way for: mkv -> extract sub -> "inject" into mp4
[21:55] <philipl> Oh yes. True. You can remux an mkv to mp4 if you have ass subtitles.
[21:55] <ubitux> well, iirc we have a proper srtenc
[21:55] <philipl> but srt won't work.
[21:55] <ubitux> so it should do the trick, no?
[21:55] <philipl> Hmm. Probably...
[21:55] <ubitux> ah right the movsub doesn't support that properly?
[21:56] <ubitux> meh.
[21:56] <philipl> I can't remember right now if srtdec uses duration correctly.
[21:56] <ubitux> haha
[21:56] <philipl> Anyway, it's all way harder than it should be but we're almost there.
[21:56] <ubitux> brainchaos
[21:57] <philipl> quite.
[21:57] <jesk> for me more like brainapocalypse
[21:58] <jesk> so i should forget about it at the moment, right? :)
[21:58] <philipl> Well, you can build my development tree, then it will work.
[21:58] <philipl> https://github.com/philipl/FFMPEG/tree/mov_text
[21:58] <ubitux> jesk: try to play with -c:s mov_text maybe
[21:58] <jesk> investing hours wouldnt be worth to be hontest
[21:59] <jesk> at least not at the moment
[21:59] <jesk> philipl, so you are quite far, cool stuff
[22:00] <jesk> ubitux, thanks I will give it a shot
[22:02] <philipl> jesk: actually that should work now. I've fixed enough. You should be able to extract the subtitle to a separate srt file and then mux that into your mp4.
[22:03] <CIA-40> ffmpeg: 03Clément BSsch 07master * r3250231a02 10ffmpeg/ (8 files in 5 dirs): lavfi: add edgedetect filter.
[22:04] <saste> \o/
[22:04] <ubitux> @_@
[22:05] <jesk> philipl: brilliant!
[22:05] <philipl> You have to use -c:s mov_text when doing the mp4 muxing.
[22:06] <beastd> yay, more virtual edges for all of us :D
[22:09] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * re71df841ee 10ffmpeg/libavcodec/fft-test.c: 
[22:09] <CIA-40> ffmpeg: fft-test: switch to new cpu flags API
[22:09] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:09] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r540bd5a95f 10ffmpeg/libavutil/cpu.c: 
[22:09] <CIA-40> ffmpeg: av_parse_cpu_caps: add "mmxext" as alias for mmx2
[22:09] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:09] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r4b529edff8 10ffmpeg/libavutil/cpu.h: 
[22:09] <CIA-40> ffmpeg: deprecate av_parse_cpu_flags
[22:09] <CIA-40> ffmpeg: This function is problematic in several ways, its also quite
[22:09] <CIA-40> ffmpeg: unpredictable which flags it ends up turning on
[22:09] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:35] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r3d7817048c 10ffmpeg/libavcodec/escape124.c: 
[22:35] <CIA-40> ffmpeg: escape124: fix integer overflow leading to excessive memory allocation
[22:35] <CIA-40> ffmpeg: Fixes Ticket1629
[22:35] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:00] --- Fri Aug 17 2012