[Ffmpeg-devel-irc] ffmpeg-devel.log.20120308

[burek]

[00:00] <Orphis> It's like the compiler is optimizing them
[00:11] <CIA-17> ffmpeg: 03Nicolas George 07master * ra8cdb7b95a 10ffmpeg/libswresample/rematrix.c: 
[00:11] <CIA-17> ffmpeg: swr: special case for empty channels in rematrix.
[00:11] <CIA-17> ffmpeg: Speed += 10% overall for pan=stereo:FL=FL+FR from wav to null.
[00:13] <Orphis> ubitux: It's incredible, there's __attribute__((used)) and it's still optimized
[00:13] <ubitux> might be a bug in the compiler
[00:13] <ubitux> is it supposed to be honored?
[00:13] <CIA-17> ffmpeg: 03Nicolas George 07master * r78306c51ef 10ffmpeg/ffmpeg.c: ffmpeg: show fps more accurately when it is small.
[00:14] <Orphis> It's supposed to be compatible with GCC :P
[00:15] <Orphis> Both variables that are optimized are used the same way as the others that aren't
[00:19] <Orphis> ubitux: http://llvm.org/bugs/show_bug.cgi?id=3576
[00:21] <ubitux> i guess that could be added to https://ffmpeg.org/trac/ffmpeg/wiki/KnownCompilationIssues
[00:21] <Orphis> I'm trying with clang
[00:22] <Orphis> Worked fine with clang
[00:26] <Orphis> Recompiling again with brew using clang and seeing if the binary works
[00:57] <Orphis> ubitux: The binary worked fine with ffmpeg and x264 from HEAD
[00:58] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rf64cfba93b 10ffmpeg/libavutil/cpu.h: 
[00:58] <CIA-17> ffmpeg: libavutil: document av_force_cpu_flags(-1).
[00:58] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:58] <CIA-17> ffmpeg: 03Tomas Härdin 07master * rbf2290a80c 10ffmpeg/libavformat/mxfdec.c: 
[00:58] <CIA-17> ffmpeg: mxfdec: Truncate packets that extend past the next edit unit
[00:58] <CIA-17> ffmpeg: This fixes rare cases where OPAtom may be treated as OP1a, causing all essence
[00:58] <CIA-17> ffmpeg: to be read into RAM.
[00:58] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:58] <CIA-17> ffmpeg: 03Tomas Härdin 07master * r8c7721e709 10ffmpeg/libavformat/mxfdec.c: 
[00:58] <CIA-17> ffmpeg: mxfdec: Fix Avid AirSpeed files being misinterpreted as OP1a
[00:58] <CIA-17> ffmpeg: The "ECs != 1 -> OP1a" assumption was wrong. Luckily, the file that triggered
[00:58] <CIA-17> ffmpeg: that behavior had two ECs, not zero. Hence distinguishing between them is
[00:58] <CIA-17> ffmpeg: simple in this case.
[00:58] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:58] <CIA-17> ffmpeg: 03Tomas Härdin 07master * rf6daa9f4f6 10ffmpeg/libavformat/mxfdec.c: 
[00:58] <CIA-17> ffmpeg: mxfdec: Add intra_only flag to MXFTrack
[00:58] <CIA-17> ffmpeg: This allows future assumptions to be made without affecting non-intra files.
[00:58] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:58] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r4cda8aa1c5 10ffmpeg/libavformat/utils.c: 
[00:58] <CIA-17> ffmpeg: avformat_new_stream: fix initial last_dts value
[00:58] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:34] <burek> I saw Elvis yesterday :)
[02:35] <burek> omg wrong window :D
[03:08] <Compn> lol
[03:09] <Compn> burek : fat elvis or .... :P
[03:09] <Compn> bruce campbell 
[03:10] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * ra55d5bdc6e 10ffmpeg/libavcodec/mmvideo.c: 
[03:10] <CIA-17> ffmpeg: algmm: convert to bytestream2 API.
[03:10] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[03:10] <CIA-17> ffmpeg: CC: libav-stable at libav.org
[03:10] <CIA-17> ffmpeg: 03Diego Biurrun 07master * r1e9d55e45e 10ffmpeg/libavcodec/x86/h264_qpel_mmx.c: x86: Remove duplicated AVG_3DNOW_OP / AVG_MMX2_OP macros from h264_qpel_mmx.c.
[03:10] <CIA-17> ffmpeg: 03Carl Eugen Hoyos 07master * ra294a7a1b3 10ffmpeg/libavformat/mov.c: 
[03:10] <CIA-17> ffmpeg: mov: Allow last chunk to have an arbitrary number of samples.
[03:10] <CIA-17> ffmpeg: Fixes ticket #673.
[03:10] <CIA-17> ffmpeg: (cherry picked from commit 8dcd2a41ecff8cc1e9b20cc267df54c59878ab3b)
[03:10] <CIA-17> ffmpeg: Signed-off-by: Alex Converse <alex.converse at gmail.com>
[03:10] <CIA-17> ffmpeg: 03Diego Biurrun 07master * r3c715383ea 10ffmpeg/libavcodec/ (vc1.c vc1.h vc1acdata.h vc1data.c vc1data.h vc1dec.c): 
[03:10] <CIA-17> ffmpeg: vc1: Move init code shared between decoder and parser to common code file.
[03:10] <CIA-17> ffmpeg: This fixes standalone compilation of the VC-1 parser.
[03:10] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * ra93b572ae4 10ffmpeg/libavformat/smacker.c: 
[03:10] <CIA-17> ffmpeg: smacker: error out if palette copy-with-offset overruns palette size.
[03:10] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[03:10] <CIA-17> ffmpeg: CC: libav-stable at libav.org
[03:10] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * re6ffd997cb 10ffmpeg/libavcodec/ (dca.c dcadata.h): 
[03:10] <CIA-17> ffmpeg: dca: prevent accessing static arrays with invalid indexes.
[03:10] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[03:11] <CIA-17> ffmpeg: CC: libav-stable at libav.org
[03:11] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * r1255eed533 10ffmpeg/libavcodec/eatgq.c: 
[03:11] <CIA-17> ffmpeg: tgq: convert to bytestream2 API.
[03:11] <CIA-17> ffmpeg: This protects against input buffer overreads.
[03:11] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[03:11] <CIA-17> ffmpeg: CC: libav-stable at libav.org
[03:11] <CIA-17> ffmpeg: 03Christophe GISQUET 07master * r7104c23bd1 10ffmpeg/libavcodec/rv34.c: 
[03:11] <CIA-17> ffmpeg: rv34: clean a bit availability checks.
[03:11] <CIA-17> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[03:11] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * rf1320dc3be 10ffmpeg/libavcodec/pcm-mpeg.c: 
[03:11] <CIA-17> ffmpeg: lpcm: fix sample size calculation for 20bit LCPM.
[03:11] <CIA-17> (52 lines omitted)
[03:11] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * re97efecec8 10ffmpeg/libavcodec/dvdata.c: 
[03:11] <CIA-17> ffmpeg: dv: check buffer size before reading profile.
[03:11] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[03:11] <CIA-17> ffmpeg: CC: libav-stable at libav.org
[03:11] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * r6e2821160b 10ffmpeg/libavcodec/aacsbr.c: 
[03:11] <CIA-17> ffmpeg: Revert "AAC SBR: group some writes."
[03:11] <CIA-17> ffmpeg: This reverts commit ba36f14e5d34bcca6ecf77ba1c7512d1b51a9740. It
[03:11] <CIA-17> ffmpeg: broke decoding on x86-32 on some systems.
[03:11] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * r37cc8600d0 10ffmpeg/libavcodec/ (cook.c cookdata.h): 
[03:11] <CIA-17> ffmpeg: cook: extend channel uncoupling tables so the full bit range is covered.
[03:11] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[03:11] <CIA-17> ffmpeg: CC: libav-stable at libav.org
[03:11] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * r3416d0805e 10ffmpeg/libavcodec/dvdata.c: dv: cosmetics.
[03:11] <CIA-17> ffmpeg: 03Diego Biurrun 07master * r8ca6e523a6 10ffmpeg/libavcodec/ (7 files): wma: Refactor common code to fix standalone compilation of WMA lossless decoder.
[03:11] <CIA-17> ffmpeg: 03Reimar Döffinger 07master * r6eda85e15b 10ffmpeg/libavcodec/x86/sbrdsp.asm: (log message trimmed)
[03:11] <CIA-17> ffmpeg: sbrdsp.asm: convert all instructions to float/SSE ones.
[03:11] <CIA-17> ffmpeg: Since the values are floats, using the float operations
[03:11] <CIA-17> ffmpeg: makes sense, improves performance on some CPUs and
[03:12] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rbf807a5e87 10ffmpeg/: (log message trimmed)
[03:12] <CIA-17> ffmpeg: Merge remote-tracking branch 'qatar/master'
[03:12] <CIA-17> ffmpeg: * qatar/master: (29 commits)
[03:12] <CIA-17> ffmpeg:  sbrdsp.asm: convert all instructions to float/SSE ones.
[03:12] <CIA-17> ffmpeg:  dv: cosmetics.
[03:12] <CIA-17> ffmpeg:  dv: check buffer size before reading profile.
[03:12] <CIA-17> ffmpeg:  Revert "AAC SBR: group some writes."
[03:12] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * r9487fb4dea 10ffmpeg/libswscale/output.c: 
[03:12] <CIA-17> ffmpeg: swscale: clip unscaled output intermediates.
[03:12] <CIA-17> ffmpeg: Fixes bug 240, as well as several integer overflows (visible as glitches)
[03:12] <CIA-17> ffmpeg: in other scaling output routines, e.g. YUV422.
[03:16] <Compn> anyone know how to access incoming and wants to remind me ?
[03:18] <Daemon404> iive, ironically the first ever file i checked after the the new scene standards was a messed-up mp4 file
[03:18] <Daemon404> hilariously so
[03:19] <Daemon404> (re: the conv we had a few days back)
[03:20] <iive> Daemon404: I should probably make the "Told you" on a macro. Happens to me a lot recently. 
[03:20] <Daemon404> lol
[03:20] <Daemon404> im sort of astounded though
[03:20] <Daemon404> it's not very hard to mux mp4 files that will play back on a pc...
[03:21] <iive> but to be honest, I already had encountered badly edited file before our conversation.
[03:21] <Daemon404> ah
[03:22] <Daemon404> i guess i forget tha ta lto of groups encode as it airs (if they dont have week-early satellite feeds)
[03:22] <Daemon404> and just cut/join
[03:22] <Daemon404> ... also typos.
[03:22] <iive> still it was bad editing... only the portion after the last advertisements was ... out of sync.
[04:23] <sj_underwater> any1 here?
[13:33] <CIA-17> ffmpeg: 03Stefano Sabatini 07master * r4ecfb91b85 10ffmpeg/doc/ffmpeg.texi: doc/ffmpeg.texi: apply misc fixes to the -map_channel documentation
[16:43] <ubitux> is it me or the depth of the atoms in mov is quite insane?
[16:43] <ubitux> http://blackhole.ubitux.fr/bbb-mov.txt (i'm sure we can go deeper, i've omitted quite a bunch of atoms)
[16:44] <av500> mov is insane
[16:44] <ubitux> well i don't find it that bad
[16:44] <ubitux> it's relatively simple
[16:45] <ubitux> but i can't really compare
[16:45] <ubitux> from what i saw, handling ebml and stuff is mkv doesn't look simpler for instance
[16:46] <av500> true
[16:46] <ubitux> any good container to mention? :)
[16:46] <av500> its a short list :)
[16:47] <ubitux> easier to remember then, give it a try :D
[16:47] <ubitux> nut? :)
[16:47] <av500> never looked into it
[16:48] <av500> I dont like how mkv and mpv are not really codec agnostic
[16:48] <av500> mpv->mov
[16:48] <av500> i hate that mov has metadata separate from content
[16:48] <av500> its nice for editing, but not for consuming it
[16:49] <av500> but then, memory is cheap now
[16:51] <ubitux> what about flv?
[16:53] <wbs> it's very simple and straightforward, but doesn't support multiple streams
[16:53] <wbs> (packets only have a video/audio flag)
[16:53] <wbs> or perhaps there was some field that nobody uses anyway, that might be interpreted like that
[16:54] <ubitux> no subtitle in flv?
[16:54] <wbs> no official support at least
[16:54] <ubitux> beh
[16:54] <ubitux> ok
[16:54] <wbs> iirc lu_zero might have thought about hacking support of his own
[16:55] <ubitux> wbs: is j-b quitting because of you? :]
[16:55] <wbs> apparently ;P
[16:55] <ubitux> it looks like there is not much sane container
[16:56] <ubitux> avi isn't that bad, right? there were just too much really bad muxer?
[16:56] <wbs> and even if you invent a "perfect" one it doesn't help as long as nobody else uses it
[16:56] <ubitux> that wasn't my goal :)
[16:57] <ubitux> i was just curious about the current state
[16:59] <wbs> ah
[17:00] <wbs> avi is really really bad afaik, but I haven't ever worked with it so I don't know the details
[17:00] <ubitux> most complains i heard were limitations and hack made by muxers around these limitations which made things worse
[17:04] <ubitux> <+av500> I dont like how mkv and mpv are not really codec agnostic // do you know a container which is?
[17:04] <av500> well, avi is/wax
[17:04] <av500> is/was
[17:04] <av500> i like asf for that
[17:04] <nevcairiel> you can mux anything into mkv that has a fourcc
[17:05] <av500> yes
[17:05] <av500> and then suddenly its not pts and more but dts
[17:05] <av500> any more
[17:05] <av500> so mkv is all pts unless when it is not
[17:05] <av500> great design!
[17:06] <nevcairiel> isnt that more of a convention then actually declared in the spec
[17:06] <av500> well, then the spec sux too
[17:07] <av500> if you are going to have dts or pts, then you better prepare for that
[17:07] <nevcairiel> the fourcc mode was meant for compat with microsoft things (specifically avi), and they never really had the concept of pts, avi all contains dts
[17:08] <nevcairiel> but yeah, its a stupid idea to redefine the meaning
[17:08] <nevcairiel> mov/mp4s design is just way too complicated to be useful
[17:11] <JEEB> mp4's status on chapters is unfunny too
[17:11] <nevcairiel> last i heard, there is no official chapter support =P
[17:11] <JEEB> 1) they tell you that you cannot use the MOV chapter format
[17:12] <av500> mov chapters are a mega pain
[17:12] <av500> the chapter titles are buried all over the file
[17:12] <JEEB> 2) they give you BIFS and I guess you're supposed to use that
[17:12] <av500> since its a chapters "track"
[17:12] <JEEB> yeah
[17:12] <JEEB> thus yes, mp4 has no official way of having chapters
[17:13] <av500> so I implemented only the chapter time, I'm not going to seek over http over a multi BG file for chapter names
[17:13] <av500> GB
[17:13] <JEEB> I guess you're supposed to create your own system on BIFS or something, but that means that there is no single standard way of doing it :D
[17:13] <av500> bifs?
[17:13] <CIA-17> ffmpeg: 03Nicolas George 07master * rb50767c31d 10ffmpeg/libavfilter/vf_pad.c: 
[17:13] <CIA-17> ffmpeg: vf_pad: keep a reference to the output buffer.
[17:13] <CIA-17> ffmpeg: Once fixed, the end_frame function does exactly what
[17:13] <CIA-17> ffmpeg: avfilter_default_end_frame does; therefore, end_frame
[17:13] <CIA-17> ffmpeg: can be removed to let avfilter_default_end_frame work.
[17:13] <CIA-17> ffmpeg: Fixes ticket #1038.
[17:13] <JEEB> a very vague part of the spec
[17:14] <JEEB> that basically defines a system within the spec that you can use for RandomStuff
[17:14] <av500> i read chapters as handbrake writes them
[17:14] <av500> and I think handbrake emulates itunes
[17:14] <JEEB> then the unstandard ways that most muxers support are 1) mov chapters 2) nero chapters
[17:14] <nevcairiel> BIFS is MPEG-4 Part 11
[17:14] <JEEB> 1) is NG because the spec says you shouldn't use it, 2) doesn't even have an open specification :D
[17:14] <nevcairiel> very obscure
[17:14] <JEEB> yes
[17:15] <JEEB> very very obscure
[17:15] <av500> great
[17:15] <av500> anyway, our current UI does not even expose chapters and nobody complains :)
[17:15] <nevcairiel> Personally i dont usually use chapters much, i watch stuff from start to end, but i can see how they might be useful sometimes
[17:15] <nevcairiel> but not crucial
[17:16] <JEEB> anyways, mp4 has too many places where they've just done something obscure and vague that you can create your own stuff on
[17:16] <av500> the ui guys dont want to clutter the UI....
[17:16] <JEEB> nevcairiel, I like them for skipping stuff over or for repeating a certain spot :3
[17:17] <JEEB> for example, you like a series but it has a 1m30s opening in the beginning
[17:17] <JEEB> hit pagedown
[17:17] <av500> A-B  repeat ftw
[17:17] <JEEB> seek straight to the beginning of the episode
[17:17] <JEEB> or if you want to repeat an opening or ending or whatever, you can just hit pageup/whatever is the button for previous chapter
[17:18] <JEEB> tl;dr I do like those things and I'm relatively sad that mp4 is like it is with regards to them
[17:18] <JEEB> they're not something I'd die for, but useful :3
[17:20] <ubitux> there are mov with chapters in the wild?
[17:21] <JEEB> as far as I know, yes -- since that's where one of the two hacks of implementing chapters in mp4 comes from
[17:21] <nevcairiel> one thing i've learned about mov/mp4, no matter how obscure some feature is, you'll find someone using it
[17:21] <ubitux> :D
[17:23] <JEEB> and BIFS and stuff give random corporates etc. even more capabilities of hacking random shit into it :D
[17:30] <CIA-17> ffmpeg: 03Nicolas George 07master * rfe9818d004 10ffmpeg/ffplay.c: 
[17:30] <CIA-17> ffmpeg: ffplay: fix two const warnings.
[17:30] <CIA-17> ffmpeg: No change in assembly code produced on x86_64.
[17:31] <CIA-17> ffmpeg: 03Nicolas George 07master * r15d75dabee 10ffmpeg/libavcodec/utils.c: 
[17:31] <CIA-17> ffmpeg: lavc/utils: fix three const warnings.
[17:31] <CIA-17> ffmpeg: No change in assembly code produced on x86_64.
[20:04] <pasteeater> Daemon404: the rest of the samples (from ticket 534) decoded fine on x86.
[20:05] <Daemon404> so i just need to fix x64
[20:07] Action: Daemon404 is actually using it right now to batch transcode utvideo -> png/mov
[21:40] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r4093220029 10ffmpeg/libavformat/mov.c: 
[21:40] <CIA-17> ffmpeg: mov: Discard invalid CTTS.
[21:40] <CIA-17> ffmpeg: Fixes Ticket385
[21:40] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:40] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r12c65efb41 10ffmpeg/libavformat/oggparseogm.c: 
[21:40] <CIA-17> ffmpeg: oggparseogm: Fix order or arguments.
[21:40] <CIA-17> ffmpeg: Bug introduced by libav in 1bb3990b560e7ff7f46c06c14f959fe432e0b625
[21:40] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:00] <Compn> 3
[23:20] <CIA-17> ffmpeg: 03Lou Logan 07master * rf70b0ed7ca 10ffmpeg/configure: 
[23:20] <CIA-17> ffmpeg: configure: enable libutvideo is gpl warning
[23:20] <CIA-17> ffmpeg: This will warn the user when --enable-gpl is missing.
[23:20] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:20] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r4916a8fc44 10ffmpeg/libavcodec/proresenc_anatoliy.c: 
[23:20] <CIA-17> ffmpeg: prores_anatoliy: remove unused variable from prores_encode_frame()
[23:20] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:20] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * ra4524930d9 10ffmpeg/libavcodec/qpeg.c: 
[23:20] <CIA-17> ffmpeg: qpeg: remove unused var from decode_frame()
[23:20] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:00] --- Fri Mar  9 2012