[Ffmpeg-devel-irc] ffmpeg-devel.log.20140404

[burek]

[00:11] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:8710ee11d75e: avcodec/h264: clear cur_pic structure instead of duplicating it in ff_h264_update_thread_context()
[00:23] <J_Darnley> :) -Werror=empty-body is quite useful.  It has found another semi-colon here libavformat/mxfdec.c:511
[00:29] <J_Darnley> I don't suppose the maintainer of mxfdec is here, Tomas Härdin?
[01:04] <J_Darnley> Plorkyeran_: Do you want to be thanked for finding that warning option in my commit message?
[01:05] <Plorkyeran_> seems awfully trivial
[01:05] <J_Darnley> You still found it or just knew it
[01:12] <michaelni> mateo`, thardin, TimNich can you take a look at  0404  0:46 James Darnley   (0.9K) [FFmpeg-devel] [PATCH 3/3] lavf/mxfdec: delete trailing semi-colon 
[01:12] <michaelni> the code as is is obviously not correct
[01:13] <michaelni> i mean the code as is in git currently
[01:24] <J_Darnley> michaelni: Sorry about the noise on the list
[01:26] <J_Darnley> I'm going to sleep.  Send an email if you need to contact this idiot.
[01:50] <cone-109> ffmpeg.git 03James Darnley 07master:7307a7fd936e: configure: check and use -Wempty-body
[02:15] <cone-109> ffmpeg.git 03Luca Barbato 07master:9464b69da6b3: log: Unbreak windows support
[02:15] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:a692ff93ef5b: Merge commit '9464b69da6b373d1eeb6b3d269ad029d9647b498'
[02:24] <cone-109> ffmpeg.git 03Luca Barbato 07master:ad8f736b7404: mxf: Remove typo
[02:24] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:069e462bebde: Merge commit 'ad8f736b7404503d8f5164f4879175a2258dd64b'
[02:53] <cone-109> ffmpeg.git 03Carl Eugen Hoyos 07master:5a85cf0fc224: isom: add 'm1v ' tag
[02:53] <cone-109> ffmpeg.git 03Reimar Döffinger 07master:b0a54516eb3c: isom: add 'flic' tag
[02:53] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:121b5b7e0476: isom: add some BOXX tags
[02:53] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:1946ad4d8c08: Merge commit '121b5b7e0476d22408d748b914799b7c00acc3a4'
[03:05] <Compn> wm4 : theres a guy in #mplayer who wants to work on mplayer, can you believe it ? :)
[03:25] <cone-109> ffmpeg.git 03James Almer 07master:fa570a5548a3: riff: Add ITRK tag
[03:25] <cone-109> ffmpeg.git 03Clément BSsch 07master:235bf1dab531: riff: add ISMP/timecode tag
[03:25] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:a52a1bc91274: riff: add SMV2 FourCC for H264
[03:25] <cone-109> ffmpeg.git 03Carl Eugen Hoyos 07master:51ace0fb921e: riff: add H264 from QNAP Systems
[03:25] <cone-109> ffmpeg.git 03Carl Eugen Hoyos 07master:9d9d319a3c29: riff: Add V264 for H264 in CCTV recordings
[03:25] <cone-109> ffmpeg.git 03Carl Eugen Hoyos 07master:dcb6e145ebb3: riff: add FourCC for GeoVision cameras
[03:25] <cone-109> ffmpeg.git 03Carl Eugen Hoyos 07master:915524e541f0: riff: Support H.264 fourcc UMSV
[03:25] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:9ab2d5a33c26: riff: add "lsvm" FourCC for H263.
[03:25] <cone-109> ffmpeg.git 03Compn 07master:fa6a02112132: riff: add G264 to mpeg4, only decodes keyframes
[03:25] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:32b0942c99d4: riff: add DM4V FourCC
[03:25] <cone-109> ffmpeg.git 03Compn 07master:684e43c5e3fd: riff: add DYM4 mpeg4 fourcc
[03:25] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:ade32b4a8400: riff: add "SM4V" FourCC
[03:25] <cone-109> ffmpeg.git 03Carl Eugen Hoyos 07master:f5f18aecef27: riff: Support decoding ASP variant from QNAP Systems
[03:25] <cone-109> ffmpeg.git 03Compn 07master:96dd7b233230: riff: add PLV1 fourcc to mpeg4
[03:25] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:adf0ff000055: riff: add "GXVE" FourCC for WMV2
[03:25] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:6aa7cbd58c3e: riff: add dvis/pdvc FourCCs
[03:26] <cone-109> ffmpeg.git 03Carl Eugen Hoyos 07master:a1c58b3acd07: riff: Support NTSC forward dvcpro videos
[03:26] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:eaec09a57518: riff: Add SLDV FourCC for dvvideo
[03:26] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:771cb3e59239: riff: support BW10 as mpeg1
[03:26] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:9ff6429b0742: riff: Support XMPG as mpeg1
[03:26] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:c9787f034c6a: riff: Add "S263" FourCC
[03:26] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:7100f3c07dbf: riff: add "YUV8" FourCC
[03:26] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:ae58466994be: riff: support 'aas4' FourCC
[03:26] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:fb6c21fa8caa: riff: add escape130 FourCC
[03:26] <cone-109> ffmpeg.git 03Paul B Mahol 07master:228af18a9832: riff: Support FLIC FourCC
[03:26] <cone-109> ffmpeg.git 03Compn 07master:10a14bd57457: riff: map 0x0038 to amrnb
[03:26] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:b7a12ab95bbd: riff: add G723_1 wav tag
[03:26] <cone-109> ffmpeg.git 03Compn 07master:7b600ea0da8b: riff: add 0x64 to g726
[03:26] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:00ccdcd88c93: riff: add 0x594a TwoCC for Xan DPCM
[03:26] <cone-109> ffmpeg.git 03Piotr Bandurski 07master:1e8c4fb8fc1a: riff: support 0xa100 TwoCC for G723_1
[03:26] <cone-109> ffmpeg.git 03Carl Eugen Hoyos 07master:ce58ead40dd1: riff: Add an additional AAC TwoCC
[03:26] <cone-109> ffmpeg.git 03Michael Niedermayer 07master:634636eb5aeb: Merge remote-tracking branch 'qatar/master'
[03:28] <Compn> koda fixed my commit name :)
[05:52] <cone-109> ffmpeg.git 03James Darnley 07master:16509d3a286c: log: allow color highlighting in Cygwin's mintty
[15:09] <cone-703> ffmpeg.git 03Peter Ross 07master:ac4b32df71bd: On2 VP7 decoder
[15:09] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:fb61ed1e9fca: Merge commit 'ac4b32df71bd932838043a4838b86d11e169707f'
[15:15] <cone-703> ffmpeg.git 03Vittorio Giovara 07master:287432430c5c: fate: add a VP7 test
[15:15] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:a20480e6592e: Merge commit '287432430c5c9cb8ffe1da1376949de494f24e45'
[15:23] <michaelni> BBB, ubitux btw has someone checked if the vp7/8 code hits any of gccs inlining limits?
[15:24] <ubitux> i didn't follow the vp7 adventures
[15:24] <michaelni> i was wondering as the libav variant droped 2 function pointers and has checks inline
[15:26] <ubitux> btw, the 30 first frames are mostly useless
[15:26] <ubitux> (look at the hashes)
[15:26] <ubitux> you probably want to -ss 10 or something
[15:29] <ubitux> so, about avx2, i don't see any code yet
[15:30] <ubitux> how well does it integrate itself with ssse3 & friends?
[15:31] <nevcairiel> i'm sure the x86inc.asm takes care of all the dirty business of VEX coded instructions and cleaning the uper half of the registers when required
[15:32] <ubitux> i'm wondering if i can i easily just change the parts where it does 8’16b in 2 reg + repack 
[15:32] <nevcairiel> but how would you mix it? ssse3 stuff doesnt like ymm :)
[15:32] <nevcairiel> Skyler_ can enlighten you, I'm sure
[15:32] <ubitux> well that's the main question; i mean i don't even know if mX is 128 or 256b in the ymm context
[15:33] <nevcairiel> if you use INIT_YMM above, it should be ymm
[15:33] <nevcairiel> or so i understand
[15:33] <nevcairiel> you can poke at the avx code, which does exist in some limited degree
[15:35] <ubitux> i guess it doesn't matter if the same code works with 256b instead of 128 (as long as the load & store have the same fixed sizes)
[15:36] <nevcairiel> as i understand, the instructions also work with xmm, if you just want to use some new instructions that didnt exist before, instead of larger registers
[15:36] <ubitux> right, then that should do it
[15:37] <Compn> michaelni : whats going on with the copyright of vp7 when 4 or more authors have fixes but the patch was squashed so theres no way to tell in git blame ?
[15:39] <ubitux> Compn: nothing new here
[15:39] <nevcairiel> should it ever come to a dispute, they can't really proof what is theirs anymore, and thats that?
[15:41] <Compn> nevcairiel : no, it has to be rewritten , and since no one knows , all that code has to be ...
[15:41] <Compn> i'm asking because sometimes people like to relicense code...
[15:41] <ubitux> Compn: go vimdiff
[15:42] <nevcairiel> Compn: luckily its already lgpl, so relicense to what now? :d
[15:43] <Compn> 'lgpl should be good enough for anyone'
[15:43] <Compn> :P
[15:50] <cone-703> ffmpeg.git 03Janne Grunau 07master:f37815b1d5ee: arm: asm decode_block_coeffs_internal is vp8 specific
[15:50] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:35617366b581: Merge remote-tracking branch 'qatar/master'
[16:49] <cone-703> ffmpeg.git 03James Almer 07master:0f6298233556: oggenc: check return value of av_mallocz()
[16:57] <Skyler_> converting to avx2 is pretty easy; the main catch is that the lane aspect can require changes wheneve data crosses the middle boundary
[16:58] <Skyler_> and sometimes the optimal way to structure things does change a little bit.
[16:58] <ubitux> i might be able to restructure in a nice way since i might have one or two additionnal reg
[17:21] <cone-703> ffmpeg.git 03Peter Ross 07master:8348bd198ff8: avformat/wtvdec: ignore MPEG2VIDEO extradata when count is invalid
[18:30] <cone-703> ffmpeg.git 03Peter Ross 07master:0cc685e3add7: avformat/wtvdec: populate codec_tag from BITMAPINFOHEADER
[18:56] <cone-703> ffmpeg.git 03Roman Fietze 07master:5bb3f8825584: avcodec.h, avpacket.c: use const AVPacket pointers in copy functions
[21:17] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:506064ed10f6: ffmpeg: frame droping due to fps should not be a warning
[21:30] <ubitux> btw
[21:30] <ubitux> i see both mixed "INIT_XMM avx" and "INIT_YMM avx"
[21:30] <ubitux> i guess YMM happens when using the float ops?
[21:30] <nevcairiel> ymm happens when you want bigger registers
[21:41] <ubitux> seems i can't use movh with YMM/avx2
[21:43] <Skyler_> yes, you'll need to use mova with xm
[21:43] <Skyler_> we could add support for movh with avx, I'm not sure how hard it'd be macro wise
[21:44] <ubitux> can i use mova with xm for the XMM versions as well?
[21:44] <ubitux> or i need different code?
[21:45] <ubitux> (i'm asking because i really have a bunch of them)
[21:45] <ubitux> also since i'm doing some define movx mov[uh] that's gonna get pretty ugly
[21:51] <baptiste> ZMM
[21:52] <ubitux> Skyler_: what about movd?
[21:52] <ubitux> mmh i think i saw some kind of splating instr...
[21:53] <Skyler_> vpbroadcast*
[21:53] <Skyler_> movd should work fine.
[21:54] <ubitux> movd m0, food doesn't for a "foo" arg
[21:54] <ubitux> "error: invalid size for operand 1"
[21:55] <Skyler_> movd xm0, food
[21:55] <ubitux> right ok
[22:26] <cone-703> ffmpeg.git 03Lukasz Marek 07master:3aac5fcfa9d3: lavu/opt: validate range before dereference
[22:44] <cone-703> ffmpeg.git 03Christophe Gisquet 07master:dfd865e51b89: x86/synth_filter: remove the main loop when it's not needed
[22:51] <cone-703> ffmpeg.git 03Alessandro Ghedini 07master:25b6837f7cac: replaygain: fix gain sign decoding
[22:51] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:652061313979: Merge commit '25b6837f7cacd691b19cbc12b9dad1ce84a318a1'
[23:00] <cone-703> ffmpeg.git 03Alessandro Ghedini 07master:8542f9c4f171: replaygain: correctly parse peak values
[23:00] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:c1d053621c2c: Merge commit '8542f9c4f17125d483c40c0c5723842f1c982f81'
[23:05] <cone-703> ffmpeg.git 03Alessandro Ghedini 07master:1d55f8d5f6f2: flacdec: export replaygain tags from Vorbis comments
[23:05] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:8e8c8607fc23: Merge commit '1d55f8d5f6f2cd74fa27170e2be37a72441d9202'
[23:24] <cone-703> ffmpeg.git 03Alessandro Ghedini 07master:0048deb84cb6: oggparsevorbis: export replaygain tags from Vorbis comments
[23:24] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:849a8c4cdb70: Merge commit '0048deb84cb6d22ba7f4fd7c8b4ecc054fcc22d4'
[23:26] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:b97f2c7c49e3: replaygain: remove unused variable
[23:26] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:b00f6bb90a67: Merge commit 'b97f2c7c49e35724f3963677f8aeee28089dc64e'
[00:00] --- Sat Apr  5 2014