[Ffmpeg-devel-irc] ffmpeg-devel.log.20120530

Thu May 31 02:05:03 CEST 2012

[00:29] <ubitux> ./ffplay -f lavfi 'testsrc,scale=256:256,mp=geq=random(1)/hypot(X-cos(N*0.07)*W/2-W/2\,Y-sin(N*0.09)*H/2-H/2)^2*1000000*sin(N*0.02):128:128'
[00:29] <ubitux> beh :(
[00:29] <ubitux> crash :(
[00:29] <Daemon404> just what are you trying to od
[00:29] <ubitux> it's a fun geq in ffmpeg history
[00:30] <ubitux> but it only works with ffplay video -vf ...
[00:30] <Daemon404> whats a geq
[00:31] <ubitux> eval an equation for each pixel
[00:31] <ubitux> + read, so you can apply various custom filtering :)
[00:31] <ubitux> see mplayer man page
[00:31] <Daemon404> heh
[00:31] Action: Daemon404 doesnt see anyone but the devs having use for it
[00:32] <ubitux> i actually saw a lot of examples on the net
[00:32] <ubitux> there was a page with ppl doing some grid on a video
[00:32] <ubitux> and half the boxes were inverted yuv
[00:32] <ubitux> fun stuff.
[00:34] <ubitux> ah it works with rgbtestsrc
[01:07] <CIA-119> ffmpeg: 03Clément BSsch 07master * r8619362ff1 10ffmpeg/ffprobe.c: ffprobe: fix C escaping.
[01:08] <pasteeater> michaelni: i'd like to add images to the wiki for some filter pages. how should i go about doing this without having to host them myself or do you care about third-party mage hosting?
[01:08] <pasteeater> s/mage/image
[01:08] Action: pasteeater casts magic missle
[01:25] <e98> there seems to be a regression in 0.11
[01:25] <e98> 0.10.3 is quick at converting a video to output an mp3
[01:25] <e98> ffmpeg -i video.ext -ab 160k audio.mp3
[01:26] <e98> git-bisect thinks it could be introduced starting with 2dfea1205870c298461f08fc323c652be41888f8
[01:27] <e98> something interesting when bisecting is that
[01:28] <e98> often the revision bisect stopped at happened to only build avconv and other av* executables
[01:28] <e98> compared to the usual ff* executable in other revisions
[01:28] <e98> maybe this is some odd effect of merging stuff back and forth from ffmpeg to libav
[01:31] <e98> 0.11 is really slow
[01:31] <e98> it tends to display more info during conversion and takes seemingly endless time
[01:32] <e98> displayin 9fps
[01:33] <e98> have to go. will check channel log.
[01:37] <pasteeater> e98: consider submitting a bug report so we can keep track (i'm assuming you'll read the channel log)
[01:51] <CIA-119> ffmpeg: 03Samuel Pitoiset 07master * r3bdb438e65 10ffmpeg/libavformat/http.c: 
[01:51] <CIA-119> ffmpeg: http: Add support for using persistent connections
[01:51] <CIA-119> ffmpeg: Add a new AVOption 'multiple_requests', which indicates if we want
[01:51] <CIA-119> ffmpeg: to use persistent connections (ie. Connection: keep-alive).
[01:51] <CIA-119> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:52] <CIA-119> ffmpeg: 03Samuel Pitoiset 07master * re999b641df 10ffmpeg/libavformat/ (http.c http.h): 
[01:52] <CIA-119> ffmpeg: http: Add support for reusing the http socket for subsequent requests
[01:52] <CIA-119> ffmpeg: Introduce ff_http_do_new_request(), a new function which sends a new
[01:52] <CIA-119> ffmpeg: HTTP request, reusing the existing connection to the server.
[01:52] <CIA-119> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:52] <CIA-119> ffmpeg: 03Mans Rullgard 07master * r7e5880e0cb 10ffmpeg/tests/ (Makefile rotozoom.c utils.c videogen.c): 
[01:52] <CIA-119> ffmpeg: fate: teach videogen/rotozoom to output a single raw video stream
[01:52] <CIA-119> ffmpeg: This makes videogen/rotozoom output a raw video stream on stdout
[01:52] <CIA-119> ffmpeg: if no output directory is specified.
[01:52] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[01:52] <CIA-119> ffmpeg: 03Mans Rullgard 07master * r47b5996bb0 10ffmpeg/tests/ (Makefile fate-run.sh): 
[01:52] <CIA-119> ffmpeg: fate: allow tests to specify unit size for psnr comparison
[01:52] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[01:52] <CIA-119> ffmpeg: 03Vitor Sessak 07master * rbac0729d9e 10ffmpeg/libavcodec/x86/ (fft.c fft.h fft_3dn2.c fft_mmx.asm): 
[01:52] <CIA-119> ffmpeg: fate: fix acodec/vsynth tests for make 3.81
[01:52] <CIA-119> ffmpeg: GNU make 3.81 applies pattern rules in declaration order rather than
[01:52] <CIA-119> ffmpeg: by stem length as in 3.82. This moves the more generic patterns above
[01:52] <CIA-119> ffmpeg: the more specific ones such that they work with either make version.
[01:52] <CIA-119> ffmpeg: Some of the vsynth patterns are also simplified a little.
[01:52] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[01:52] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rb0387edd5e 10ffmpeg/: (log message trimmed)
[01:52] <CIA-119> ffmpeg: Merge commit 'f919cc7df6ab844bc12f89fe7bef4fb915a47725'
[01:52] <CIA-119> ffmpeg: * commit 'f919cc7df6ab844bc12f89fe7bef4fb915a47725':
[01:52] <CIA-119> ffmpeg:  fate: fix acodec/vsynth tests for make 3.81
[01:52] <CIA-119> ffmpeg:  pcm_mpeg: fix number of consumed bytes to include the header.
[01:53] <CIA-119> (2 lines omitted)
[02:07] <Zeranoe> Does anyone know if there is a way to tell avcodec_decode_audio4 that the sample rate should be 48000 or is the only way to use av_resample and resample the audio to 48000?
[02:10] <Daemon404> i think you need to call eitehr libavresample or libswresample manually
[02:10] <Daemon404> the decoder doesnt incorporate resampling afaik
[02:11] <Daemon404> btw Zeranoe, i like your mingw script -- i modified it to build me native windows binaries <_<
[02:11] <michaelni> pasteeater, theres a "attach file button" on the wiki pages with which images can supposedly be attached
[02:11] <michaelni> didnt try it though
[02:12] <michaelni> its right of the "edit this page" button
[02:13] <Zeranoe> Daemon404: Thank you, I'm working on the 3.0 release right now. Updates a lot of things and supports option/argument feeding. Thats interesting you modified it to support a native toolchain, would you mind sharing it with me?
[02:14] <Daemon404> Zeranoe, it's a giant hack atm :P
[02:14] <Daemon404> it doesnt bootstrap properly
[02:14] <Daemon404> it needs a native linux cross-compiler in the PATH already
[02:15] <Daemon404> i just changed CC and added a --host and --disable-win32-registry
[02:17] <Daemon404> and s/4.6.2/4.6.3/ but thats trivial
[02:27] <pasteeater> michaelni: ah, of course i didn't notice that before i asked.
[02:43] <pasteeater> michaelni: mind giving me WIKI_DELETE permissions? so i can delete my own attachments. user llogan.
[03:18] <michaelni> pasteeater, done
[03:18] <pasteeater> michaelni: thanks
[03:20] <pasteeater> several new users are getting bitten by libx264 using yuv422p instead of yuv420p when using DV input.
[04:43] <CIA-119> ffmpeg: 03Anton Khirnov 07master * r8daf21d567 10ffmpeg/ (3 files in 3 dirs): 
[04:43] <CIA-119> ffmpeg: avconv: merge configuration code for complex and simple filters
[04:43] <CIA-119> ffmpeg: Some tests change because -s now inserts the scaler to the end instead
[04:43] <CIA-119> ffmpeg: of beginning of the filtergraph.
[04:43] <CIA-119> ffmpeg: 03Anton Khirnov 07master * ra977d91e28 10ffmpeg/avconv.c: avconv: split configuring input filters out of configure_complex_filter
[04:43] <CIA-119> ffmpeg: 03Anton Khirnov 07master * rc9cc76290f 10ffmpeg/ (avconv.c configure doc/avconv.texi): 
[04:43] <CIA-119> ffmpeg: avconv: extend -r to work on any input stream.
[04:43] <CIA-119> ffmpeg: This is done by automatically inserting a setpts filter.
[04:43] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r16e25ef34b 10ffmpeg/: (log message trimmed)
[04:43] <CIA-119> ffmpeg: Merge remote-tracking branch 'qatar/master'
[04:43] <CIA-119> ffmpeg: * qatar/master:
[04:43] <CIA-119> ffmpeg:  avconv: extend -r to work on any input stream.
[04:43] <CIA-119> ffmpeg:  doc/avconv: expand documentation for the -s option.
[04:43] <CIA-119> ffmpeg:  avconv: don't print filters inserted by avconv in stream mappings.
[04:43] <CIA-119> ffmpeg: fate: make idroq test run again.
[04:43] <CIA-119> ffmpeg: Didnt investigate what broke it
[04:43] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:43] <CIA-119> ffmpeg: 03Anton Khirnov 07master * rcf6c38c62b 10ffmpeg/avconv.c: avconv: don't print filters inserted by avconv in stream mappings.
[04:43] <CIA-119> ffmpeg: 03Anton Khirnov 07master * r4f81a5072a 10ffmpeg/doc/avconv.texi: doc/avconv: expand documentation for the -s option.
[04:56] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rfd707add6b 10ffmpeg/tests/fate/audio.mak: 
[04:56] <CIA-119> ffmpeg: fate: try to fix fate-nellymoser-aref-encode make dependancies
[04:56] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:07] <CIA-119> ffmpeg: 03Lou Logan 07master * r73f3f6baff 10ffmpeg/libavcodec/utils.c: 
[06:07] <CIA-119> ffmpeg: lavc: clarify experimental codec message
[06:07] <CIA-119> ffmpeg: Should be easier for new users to get a working output.
[06:07] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:07] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r5855e46398 10ffmpeg/tests/ (fate/acodec.mak ref/fate/acodec-dca): 
[06:07] <CIA-119> ffmpeg: fate: resurrect acodec-dca
[06:07] <CIA-119> ffmpeg: This was lost in todays merge
[06:07] <CIA-119> ffmpeg: Ideas to do this less ugly are welcome!
[06:07] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:07] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * ref96ec97fb 10ffmpeg/tests/regression-funcs.sh: 
[06:07] <CIA-119> ffmpeg: fate: remove unused functions
[06:07] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:07] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r6aceed3b72 10ffmpeg/tests/codec-regression.sh: 
[06:07] <CIA-119> ffmpeg: fate: remove tests/codec-regression.sh
[06:52] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r7a2ae764f4 10ffmpeg/tests/fate/acodec.mak: 
[06:52] <CIA-119> ffmpeg: fate: disable fate-acodec-dca
[06:52] <CIA-119> ffmpeg: the new test doesnt work on all platforms, thus disable it until a better solution is found.
[06:52] <CIA-119> ffmpeg: the dca2 test remains and provides partial testing until this is resolved
[06:52] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[08:45] <durandal_1707> quick, what function gives number of comps for PIX_FMT ?
[09:13] <durandal_1707> hey, adding rgba64 O support is not hard
[09:35] <CIA-119> ffmpeg: 03Paul B Mahol 07master * rd66b0cd5c5 10ffmpeg/libavcodec/exr.c: 
[09:35] <CIA-119> ffmpeg: exr: alpha support
[09:35] <CIA-119> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[09:39] <ubitux> does anyone know if there is a way from the encoder and muxer context to ignore some metadata?
[09:40] <ubitux> for example, the matroska muxer seems to use the "language" metadata for specific heuristics
[09:40] <ubitux> but i can't tell where it's been drop
[09:40] <ubitux> (so it isn't also muxed with the other metadata)
[10:01] <durandal_1707> hmm does codecs in iff which use gray are palleted?
[10:18] <CIA-119> ffmpeg: 03Paul B Mahol 07master * r8f61526978 10ffmpeg/libavcodec/iff.c: 
[10:18] <CIA-119> ffmpeg: iff: check if there is extradata
[10:18] <CIA-119> ffmpeg: Fixes #1368.
[10:18] <CIA-119> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[13:02] <burek> can someone tell me what is a proper way to cache/buffer audio input for some time, to avoid network issues?
[13:03] <j-b> !tld ke
[13:03] <j-b> arf, not in this chan
[13:12] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r607461e50d 10ffmpeg/tests/fate/ac3.mak: 
[13:12] <CIA-119> ffmpeg: fate: add bitexact flag to ac3-fixed
[13:12] <CIA-119> ffmpeg: cant hurt, and should be there for all tests
[13:12] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[13:12] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rf5e0ef536a 10ffmpeg/tests/ (fate/acodec.mak ref/fate/acodec-dca): 
[13:12] <CIA-119> ffmpeg: fate: add a fate-acodec-dca test that works on all platforms second try
[13:12] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[13:23] <ubitux> michaelni: did you port the ffmpeg specific entries in codec-regression.sh to the Makefile rules before removing it?
[13:23] <ubitux> iirc we did a few changes to this file
[13:29] <michaelni> ubitux, all except one should be ported now
[13:29] <ubitux> michaelni: +fate-acodec-dca: SRC = tests/data/asynth-44100-2.wav  i'm not sure but i think you need $(SRC_PATH) here
[13:29] <ubitux> (not for the dep, but for the SRC only)
[13:31] <michaelni> ubitux, a quick grep for SRC tests/fate/*.mak
[13:31] <michaelni> doesnt show SRC_PATH
[13:31] <ubitux> mmh indeed
[13:31] <ubitux> i'm confused with sth, just a sec
[13:32] <ubitux> ah TARGET_PATH maybe
[13:33] <ubitux> https://lists.libav.org/pipermail/libav-devel/2012-May/028479.html
[13:33] <ubitux> that's what i used in the mapchan tests
[13:34] <michaelni> ubitux, feel free to add it if its needed ...
[13:34] <ubitux> k
[13:35] <ubitux> is the +bitexact really needed?
[13:35] <ubitux> fate-run.sh:FLAGS="-flags +bitexact -sws_flags +accurate_rnd+bitexact"
[13:35] <ubitux> it seems to be explicited in a few places though
[13:35] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rd24ef674ec 10ffmpeg/tests/ (fate/acodec.mak ref/fate/acodec-g723_1): 
[13:35] <CIA-119> ffmpeg: fate: fix long standing bug in the g723_1 test
[13:35] <CIA-119> ffmpeg: The test compared files of mismatching sample rate
[13:35] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[13:37] <michaelni> i dont really know but all tests should end up with bitexact flags one way or another
[13:46] <durandal_1707> can anyone create exr with luma/chroma but without compression?
[13:51] <cbsrobot> durandal_1707: sure
[13:52] <cbsrobot> durandal_1707: do you need rgb, rgba or another format ? 
[13:52] <durandal_1707> anything with luma
[13:57] <durandal_1707> cbsrobot: Y (gray), Y+A (gray with alpha), Y+RY + BY, .....
[14:13] <CIA-119> ffmpeg: 03Paul B Mahol 07master * rf5af8d5e76 10ffmpeg/libavcodec/exr.c: 
[14:13] <CIA-119> ffmpeg: exr: make message about missing feature more useful
[14:13] <CIA-119> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[15:46] <CIA-119> ffmpeg: 03Gavin Kinsey 07master * rbf6fb563b9 10ffmpeg/cmdutils.c: 
[15:46] <CIA-119> ffmpeg: Allow building of command line utils without swresample lib
[15:46] <CIA-119> ffmpeg: This allows building ffprobe without swresample
[15:46] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:46] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r89cf5bf39a 10ffmpeg/cmdutils.c: 
[15:46] <CIA-119> ffmpeg: cmdutils: replace 1 line #ifs by if()
[15:46] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:11] <ubitux> michaelni: oh i forgot to comment your colors patchset again
[16:11] <ubitux> michaelni: i'll do that tonight if you can wait :)
[16:21] <michaelni> ubitux, no hurry, btw we lost the ac3 seek test too and the remaining 2 fate tests are aref/vref which have been removed i think
[16:22] <michaelni> and zmbv vsynth1/2 was the not yet ported one i meant
[16:23] <michaelni> if someone wants to port it thats welcome ... if not ill do it in a few days or so
[17:29] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r96fe99cb11 10ffmpeg/ffmpeg.c: 
[17:29] <CIA-119> ffmpeg: ffmpeg: disable autoinsertion for -async on the output side of the filtergraph.
[17:29] <CIA-119> ffmpeg: This fixes the issue that since the last merge -async caused 2 aresample
[17:29] <CIA-119> ffmpeg: filters to be inserted
[17:29] <CIA-119> ffmpeg: users that generate funny timestamps in a filtergraph should probably
[17:29] <CIA-119> ffmpeg: manually insert a aresample filter where they need it.
[17:29] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:29] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * rfd4c1c0b70 10ffmpeg/libavcodec/truemotion1.c: 
[17:29] <CIA-119> ffmpeg: truemotion1: Check index, fix out of array read
[17:29] <CIA-119> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[17:29] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:29] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r32e60b6bfe 10ffmpeg/libavcodec/h264data.h: 
[17:29] <CIA-119> ffmpeg: h264: increase scantable sizes to avoid overread
[17:29] <CIA-119> ffmpeg: We could also check the index but this would slow speed critical code
[17:29] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:29] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * r39f0a45a1a 10ffmpeg/libavcodec/h264_cavlc.c: 
[17:29] <CIA-119> ffmpeg: h264_cavlc: check prefix before using it.
[17:29] <CIA-119> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[17:29] <CIA-119> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:37] <CIA-119> ffmpeg: 03Luca Barbato 07master * r528995589f 10ffmpeg/avprobe.c: 
[19:37] <CIA-119> ffmpeg: avprobe: move formatter functions in the context
[19:37] <CIA-119> ffmpeg: Avoid possible clashes.
[19:37] <CIA-119> ffmpeg: 03Justin Ruggles 07master * r8dfc122719 10ffmpeg/libavresample/audio_mix.c: lavr: add C functions for mixing 2 to 1 channels with s16p format
[19:37] <CIA-119> ffmpeg: 03Justin Ruggles 07master * rb75726cb79 10ffmpeg/libavresample/x86/ (audio_mix.asm audio_mix_init.c): lavr: add x86-optimized function for mixing 2 to 1 s16p with q8 coeffs
[19:37] <CIA-119> ffmpeg: 03Luca Barbato 07master * r43b50e62c4 10ffmpeg/avprobe.c: avprobe: correctly set the default formatter
[19:37] <CIA-119> ffmpeg: 03Mans Rullgard 07master * r00c78a0a15 10ffmpeg/tests/md5.sh: 
[19:37] <CIA-119> ffmpeg: fate: fix md5sum replacement on some systems
[19:37] <CIA-119> ffmpeg: On systems where the 'md5' command is used, there is a conflict
[19:37] <CIA-119> ffmpeg: with the md5() shell function in fate-run.sh. Using the 'command'
[19:37] <CIA-119> ffmpeg: keyword bypasses the shell function for correct behaviour.
[19:37] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[19:37] <CIA-119> ffmpeg: 03Mans Rullgard 07master * r0857e46dd5 10ffmpeg/tests/fate/ac3.mak: 
[19:37] <CIA-119> ffmpeg: fate: add missing $(TARGET_PATH) to ac3-fixed-encode
[19:37] <CIA-119> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[19:37] <CIA-119> ffmpeg: 03Christian Schmidt 07master * rd041dec3cb 10ffmpeg/libavcodec/pcm-mpeg.c: 
[19:37] <CIA-119> ffmpeg: pcm-mpeg: improve log message wording
[19:37] <CIA-119> ffmpeg: We support every defined value for channel layout, bitrate and sample depth.
[19:37] <CIA-119> ffmpeg: All other values are not unsupported, but reserved.
[19:37] <CIA-119> ffmpeg: Update comments to say "are used" instead of "are known or exist".
[19:37] <CIA-119> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[19:37] <CIA-119> ffmpeg: 03Michael Niedermayer 07master * ra1fc1d2e1b 10ffmpeg/: (log message trimmed)
[19:37] <CIA-119> ffmpeg: Merge remote-tracking branch 'qatar/master'
[19:37] <CIA-119> ffmpeg: * qatar/master:
[19:37] <CIA-119> ffmpeg:  pcm-mpeg: improve log message wording
[19:37] <CIA-119> ffmpeg:  fate: add missing $(TARGET_PATH) to ac3-fixed-encode
[19:37] <CIA-119> ffmpeg:  fate: fix md5sum replacement on some systems
[19:37] <CIA-119> ffmpeg:  avprobe: correctly set the default formatter
[19:38] <CIA-119> ffmpeg: 03Justin Ruggles 07master * rc140fb2cbc 10ffmpeg/libavresample/x86/ (audio_mix.asm audio_mix_init.c util.asm): lavr: add x86-optimized functions for mixing 2 to 1 s16p with float coeffs
[20:20] <ubitux> Daemon404: you need access to these tags in an external app?
[20:20] <ubitux> CodecTags are not public; how do you expect to be able to make use of them?
[20:22] <Daemon404> ubitux, yes
[20:22] <Daemon404> there's a reason we added teh riff tags too
[20:22] <Daemon404> if you dont, libavcodec is UNUSABLE unelss you us elavf
[20:22] <Daemon404> just try using a muxer that isnt lavf
[20:24] <ubitux> can you be a bit more specific about the usage of these CodecTags for a lost soul like me?
[20:24] <Daemon404> ubitux, sec
[20:25] <Daemon404> http://ffmpeg.mplayerhq.hu/doxygen/trunk/group__riff__fourcc.html
[20:25] <Daemon404> read the description
[20:25] <Daemon404> its the same deal
[20:25] <Daemon404> unless you use lavf, which outputs an id, you have to use these tables
[20:25] <Daemon404> otherwise you cant actually use libavcodec.
[20:25] <Daemon404> fundemental design flaw etc
[20:28] <ubitux> that still doesn't tell me how you will manage to make use of the CodecTags without redifining the struct
[20:29] <Daemon404> i had a long discussion with elenril when teh riff tags were added to
[20:29] <Daemon404> too*
[20:29] <Daemon404> because "AVCodecTag is supposed to be opaque"
[20:29] <Daemon404> so you had to redefine that too
[20:29] <Daemon404> it's teh same deal here
[20:29] <ubitux> ok
[20:30] <Daemon404> its on both MLS for review
[20:30] <Daemon404> so uh
[20:30] <Daemon404> let the bikeshedding commence!
[20:31] <Daemon404> im not pushing to either project until both sides agree
[20:31] <ubitux> i was just questionning out of curiousity :)
[20:31] <ubitux> not judging your patch :p
[20:31] <Daemon404> ;P
[21:57] <Daemon404> michaelni, i dont think libav will be very keen on modifying AVCodecTag
[21:58] <Daemon404> which is why i avoided it
[21:58] <Daemon404> i can ask though
[22:05] <michaelni> Daemon404, whats your oppinon on using codec_tag as index into an array of strings ?
[22:05] <Daemon404> it sounds messy
[22:05] <Daemon404> but so does doign per-muxer functins
[22:05] <michaelni> Tjoppen, there are mxf patches on ffmpeg-devel, do you have time to review them ?
[22:06] <michaelni> Daemon404, yes
[22:07] <michaelni> AVCodecTags purpose was to map these identifers though so not using it feels messy in its own right
[22:12] <Daemon404> i know
[22:12] <Daemon404> doesnt everything but matroska use MKTAG?
[22:13] <michaelni> at the risk of causing nightmares ...
[22:13] <michaelni> theres ogg ...
[22:14] <michaelni> it identifies codecs by well a codec specific way to identify them
[22:15] <nevcairiel> there is also mpegts with its special way of dealing with things (if it even has the pmt and doesnt rely on probing)
[22:16] <Daemon404> oh lord
[22:16] <Daemon404> such insanity
[22:16] <Daemon404> remair made one good poitn on the l
[22:25] <Tjoppen> michaelni: kinda late, poke me in the morning
[23:06] <Tjoppen> ah, it's the muxer
[00:00] --- Thu May 31 2012