[Ffmpeg-devel-irc] ffmpeg-devel.log.20120128

burek burek021 at gmail.com
Sun Jan 29 02:05:03 CET 2012


[01:51] <CIA-108> ffmpeg: 03Paul B Mahol 07master * rf9ca1ac796 10ffmpeg/ (9 files in 3 dirs): 
[01:51] <CIA-108> ffmpeg: sws/pixfmt/pixdesc: add support for yuva444p
[01:51] <CIA-108> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[01:51] <CIA-108> ffmpeg: Conflicts:
[01:51] <CIA-108> ffmpeg:  libavutil/avutil.h
[01:51] <CIA-108> ffmpeg:  libavutil/pixfmt.h
[01:51] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * reb3f81e4ef 10ffmpeg/libavcodec/targa.c: 
[01:51] <CIA-108> ffmpeg: targadec: Simplify RLE out of packet check.
[01:51] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:51] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * r49f2056289 10ffmpeg/libavcodec/pngdec.c: 
[01:51] <CIA-108> ffmpeg: pngdec: Convert 2/4 bit formats to 8bit.
[01:51] <CIA-108> ffmpeg: This way 2 and 4 bit gray, rgb & rgba are supported.
[01:51] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:51] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * rf21b6159cf 10ffmpeg/libavcodec/mpegvideo_enc.c: 
[01:51] <CIA-108> ffmpeg: mpegvideoenc: Fail if a buffer size is specified without a max rate.
[01:51] <CIA-108> ffmpeg: This combination makes not much sense.
[03:22] <pasteeater> michaelni: can you check to see if my key was added for ffmpeg-web?
[03:23] <pasteeater> i sent it to -devel on 20-nov-11.
[03:24] <pasteeater> becaue i get "fatal: The remote end hung up unexpectedly" when i "git push --dry-run"
[03:41] <michaelni> pasteeater, you seem to be in there
[03:41] <michaelni> does reading / fetch work ?
[03:43] <pasteeater> git fetch works
[03:44] <durandal_1707> michaelni: swscale loses alpha when doing yuva444p -> rgba, some more generic check if yuv format have alpha is missing
[03:45] <durandal_1707> there is already isALPHA
[03:49] <michaelni> pasteeater, also make sure you use url = gil at ffmpeg.org:ffmpeg-web
[03:50] <michaelni> not the git://
[03:51] <durandal_1707> michaelni: will post patch if fate passes
[03:51] <michaelni> durandal_1707, thx
[03:52] <durandal_1707> well, if fate passes it would indicate that alpha is not tested at all
[03:58] <durandal_1707> michaelni: yuva444p conversion into rgba sets alpha to 128 instead of 255
[03:59] <pasteeater> michaelni: ah, progress. at first i thought gil was a typo for git.
[03:59] <durandal_1707> ffmpeg -i ~/git/fate-suite/dxtory/dxtory_mic.avi -vf format=yuva444p -pix_fmt rgba /tmp/test.xwd
[04:00] <durandal_1707> rgb24->rgba works fine
[04:01] <durandal_1707> yuva420p->rgba works fine too
[04:11] <durandal_1707> michaelni: fixed that to, patch sent to ml
[04:16] <durandal_1707> michaelni: for this last one you would need to regenerate lavfi ref files, or i can do that?
[04:16] <durandal_1707> so this is what caused libav fate failure, ffmpeg puts 0x80 and libav some random shit
[04:17] <Compn> lol
[04:17] <Compn> boy theres been an explosion of pixfmts lately
[04:17] <Compn> how many different ways can you pack colors!?
[04:18] <drv> back in my day, all we had was 0 and 1, and we were happy with them!
[04:18] <gnafu> Yeah!
[04:18] <Compn> crazyness with 9bit, 10bit, 16bit too
[04:18] <durandal_1707> ignore all that crap, now is time for 64 bit
[04:19] <Compn> now if only i could play vivo with ffmpeg... :P
[04:19] <Compn> its my white whale
[04:19] <ohsix> all that porn from 1996 you could be playing!
[04:20] <durandal_1707> Compn: you dont use mplayer vivo demuxer?
[04:20] <Compn> i do
[04:21] <Compn> but still... its such an old format, it deserves preservation!
[04:21] <pasteeater> where's my cinepak encoder?
[04:23] <durandal_1707> pasteeater: you wrote it?
[04:24] <pasteeater> no. shitty joke about old formats.
[04:24] <Compn> wasnt there a patch?....
[04:26] <durandal_1707> it should not be hard once i build some exp, until then use mencoder
[04:26] <Compn> drv has a patch iirc
[04:26] <Compn> for vivo
[04:26] <Compn> i keep pestering him
[04:26] <drv> yeah, it'll be done sometime this decade
[04:26] <Compn> 2 years down, 8 to go :)
[04:26] <drv> unless the mayans were right
[04:27] <Compn> stop putting it off due to various calendar bugs
[04:27] <durandal_1707> vivo < tivo
[04:27] <drv> it is actually mostly working, i should send a patch to the ML and guilt trip somebody else into finishing it ;)
[04:27] <Compn> i'm guessing no one has looked at getting h263 decoder to decode viv2 tho
[04:28] <Compn> or vivoaudio being g723 something
[04:28] <durandal_1707> it is siren..somthing
[04:28] <drv> yeah, only the older files decode
[04:29] <drv> i dug up a bunch of samples from archive.org and such, but they are almost all the newer codecs
[04:29] <durandal_1707> how could one write decoder if he would first need to write demuxer?
[04:30] <Compn> durandal_1707 : use mplayer demuxer 
[04:30] <Compn> build mplayer with ffmpeg changes
[04:30] <durandal_1707> Compn: that is just calling for more problems....
[04:31] <Compn> mplayer must output correct video stream in order for binary codec to handle it
[04:31] <Compn> binary codec is very specific about what data it wants
[04:32] <durandal_1707> mplayer can play those new vivo files?
[04:32] <Compn> yep
[04:32] <Compn> viv1 / vivo = old, viv2 = new
[04:33] <Compn> 'new' as in, newer than vivo , still a 15 year old codec.
[04:33] <durandal_1707> so drv demuxer when using with mplayer can still play that files?
[04:33] <drv> i haven't tried actually
[04:33] <drv> but the older format is just h.263 and g.723.1, works with ffmpeg's decoders
[04:34] <Compn> you implement seeking in your demuxer ? :P
[04:35] <burek> one question about GPL, licenses, violations, blabla..
[04:36] <burek> would this kind of page be considered illegal: http://ffmpeg.test-lab.ch/uploads/ffmpeg-custom-build.html
[04:36] <drv> nah, the timestamps are busted too, one of the things I still need to fix
[04:36] <ohsix> that's what people pay lawyers for
[04:36] <burek> :)
[04:37] <durandal_1707> burek: libaacplus builds are nonfree
[04:37] <burek> yes i know, its just a dummy temp page
[04:37] <burek> to show the idea
[04:38] <Compn> burek : if you give a link to source somewhere its fine
[04:38] <durandal_1707> web page appears to be hosted from .ch?
[04:38] <Compn> non-free  license aside...
[04:38] <burek> well ok, i would create it in such way that it shows warnings all over when you select a non-free package and stuff
[04:38] <burek> but is it illegal?
[04:38] <durandal_1707> burek: stop this yagadada, and distribute ffmpeg via torrent
[04:39] <burek> :)
[04:39] <Compn> its legal to distribute as long as you also include the source
[04:39] <Compn> in a link
[04:39] <durandal_1707> burek: depends in which country is server/person
[04:39] <Compn> or something
[04:39] <burek> im not into piracy and stuff, i just want to make it easy to the people to build/use ffmpeg, thats all
[04:40] <Compn> most ffmpeg devs only care if the people distributing ffmpeg have a link to sourcecode
[04:40] <Compn> at least i hope so
[04:40] <Compn> and changes / patches published are really nice too
[04:40] <Compn> but i cant speak for whole project.
[04:40] <burek> yes, the link would be there too
[04:40] <pasteeater> enable-bzlib is autodetect
[04:40] <burek> pasteeater, it would be a static lib
[04:40] <burek> -lib +binary
[04:41] <durandal_1707> burek: how is gonna pay for electricity and data transfer?
[04:41] <burek> easy :) I can auto-upload to sourceforge or some other file sharing hosting if the upload starts going crazy :)
[04:41] <durandal_1707> imho better add precompiled stuff
[04:43] <durandal_1707> and i hope you will add x264, lame and xvid support at least
[04:44] <burek> well, I already managed to add all gpl stuff
[04:44] <burek> just alsa is making me headache so far
[04:58] <burek> how to tell cmake to build static libs, like --enable-static=yes ?:)
[05:05] <durandal_1707> google did not help?
[05:05] <burek> no :/
[05:12] <NapHtaKeRoSene> hi
[05:12] <NapHtaKeRoSene> my random is benchmarked at 3GB/s filling my graphic screen, the enhanced version is 470MB/s, /dev/urandom is 4.4MB/s
[05:15] <durandal_1707> how you watch images with mplayer?
[05:23] <Compn> mplayer mf://*jpg
[05:24] <Compn> -mf fps=1
[05:24] <Compn> -fps .1
[05:33] <NapHtaKeRoSene> interesting slideshow option
[05:48] <durandal_1707> i get strange last line with x11 output with mplayer, more noticeable with -zoom, can anyone here reproduce it?
[07:28] <vtorri> hey
[07:28] <vtorri> in os_support.h, the definition of lseek looks strange to me
[07:28] <vtorri> in msdn:
[07:28] <vtorri> http://msdn.microsoft.com/en-us/library/1yee101t%28VS.80%29.aspx
[07:29] <vtorri> -lseeki64 is already declared in io.h
[07:29] <vtorri> _lseeki64, not -lseeki64 :)
[07:42] <vtorri> lots of warnings with mingw :)
[07:48] <vtorri> also
[07:48] <michaelni> vtorri, can you send a patch :)
[07:48] <vtorri> hehe
[07:48] <vtorri> also
[07:48] <michaelni> its 8 in the morning here so i dont want to change things i cant test :)
[07:48] <vtorri> problem when creatin .def and .lib
[07:49] <vtorri> same here
[07:49] <vtorri> our countries are neighbour :)
[07:49] <vtorri> (i think)
[07:49] <vtorri> SLIB_EXTRA_CMD is defined as -lib ***
[07:50] <vtorri> i'm cross compiling (i'm on linux, using mingw-w64)
[07:50] <vtorri> lib.exe is not found
[07:50] <vtorri> so no .lib and .def
[07:50] <vtorri> that's annoying
[07:51] <michaelni> you should talk with Zeranoe or ramiro 
[07:51] <michaelni> iam not the windows guy :)
[07:51] <vtorri> hehe
[07:51] <vtorri> i know
[07:52] <vtorri> i'm just descirbing the problem, in case someone with better skills than me read them and solve the problem
[07:53] <vtorri> lib.exe seems to be a microsoft tool
[07:53] <nevcairiel> it is, but it works much better then the mingw variant, which is why its being used
[07:55] <vtorri> problem is, when cross compiling on linux, there are no microsoft tools :)
[07:56] <nevcairiel> so the configure needs to take that into account, as long as it keeps being used when compiling on windows for windows. :)
[07:56] <vtorri> configure should also check for lib.exe avaialbility
[07:57] <vtorri> (on windows)
[07:57] <nevcairiel> also, whats wrong with lseek? doesnt look wrong to me
[07:59] <vtorri> i have that warning:
[07:59] <vtorri> libavformat/os_support.h:34:0: warning: "lseek" redefined [enabled by default]
[07:59] <vtorri> /home/torri/local/opt/mingw-w64-x86_32/bin/../lib/gcc/i686-w64-mingw32/4.7.0/../../../../i686-w64-mingw32/include/io.h:336:0: note: this is the location of the previous definition
[08:00] <vtorri> io.h is certainly included somewhere
[08:00] <vtorri> so lseek is already defined
[08:00] <vtorri> then in os_support.h, you redefine lseek
[08:01] <nevcairiel> the problem is that mingws lseek is an alias for _lseek (which only exists in mingw), which only takes a long position, not a int64 position
[08:01] <nevcairiel> so it cannot be used
[08:01] <vtorri> so maybe a:
[08:01] <vtorri> #ifdef lseek
[08:01] <vtorri> # undef lseek
[08:01] <vtorri> #endif
[08:01] <vtorri> before defining your lseek in os_support.h
[08:02] <vtorri> to remove that warning
[08:03] <vtorri> other redefinitions:
[08:03] <nevcairiel> shouldnt hurt, i suppose
[08:04] <vtorri> EPROTONOSUPPORT, ETIMEDOUT, ECONNREFUSED, EINPROGRESS, gai_strerror
[08:04] <nevcairiel> in my version of mingw, lseek is a function, not a define
[08:04] <nevcairiel> so it doesnt complain
[08:04] <vtorri> in libavformat/network.h
[08:04] <vtorri> nevcairiel: ok. I use mingw-w64
[08:05] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r5eda82aac9 10ffmpeg/libavcodec/frwu.c: 
[08:05] <CIA-108> ffmpeg: frwu: check if allocation was successful
[08:05] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[08:05] <CIA-108> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[08:05] <vtorri> mingw is not improving quickly
[08:05] <CIA-108> ffmpeg: 03Diego Biurrun 07master * r299ab0fd17 10ffmpeg/libavutil/ (Makefile file.c): libavutil: Remove pointless file test program.
[08:05] <CIA-108> ffmpeg: 03Mashiat Sarker Shakkhar 07master * rd51d6ae9c4 10ffmpeg/libavcodec/vc1dec.c: (log message trimmed)
[08:05] <CIA-108> ffmpeg: vc1dec.c: Don't use uninitialized values of dmv_[x, y]
[08:05] <CIA-108> ffmpeg: MVDATA may or may not be transmitted. If it is not, both
[08:05] <CIA-108> ffmpeg: dmv_x and dmv_y is to be assumed zero.
[08:05] <CIA-108> ffmpeg: This may not trigger wrong picture in all systems, but
[08:05] <CIA-108> ffmpeg: it's a bug nevertheless. Fixes SA10116.vc1 on my 64-bit
[08:05] <CIA-108> ffmpeg: Windows 7.
[08:05] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r324e818093 10ffmpeg/libavcodec/8bps.c: 
[08:05] <CIA-108> ffmpeg: 8bps: K&R formatting cosmetics
[08:05] <CIA-108> ffmpeg: Signed-off-by: Diego Biurrun <diego at biurrun.de>
[08:05] <CIA-108> ffmpeg: 03Paul B Mahol 07master * rfc115c80b7 10ffmpeg/ (9 files in 3 dirs): 
[08:05] <CIA-108> ffmpeg: sws/pixfmt/pixdesc: add support for yuva444p
[08:05] <CIA-108> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[08:05] <CIA-108> ffmpeg: 03Ronald S. Bultje 07master * r3b15a6d742 10ffmpeg/ (27 files in 4 dirs): 
[08:05] <CIA-108> ffmpeg: config.asm: change %ifdef directives to %if directives.
[08:05] <CIA-108> ffmpeg: This allows combining multiple conditionals in a single statement.
[08:05] <CIA-108> ffmpeg: 03Alex Converse 07master * r08628b6afb 10ffmpeg/ (9 files in 3 dirs): 
[08:05] <CIA-108> ffmpeg: Revert "sws/pixfmt/pixdesc: add support for yuva444p"
[08:05] <CIA-108> ffmpeg: This reverts commit fc115c80b7bf6f1f5a937ee7f6e14630d99eb4dd.
[08:05] <CIA-108> ffmpeg: Tests are broken.
[08:05] <CIA-108> ffmpeg: 03Ronald S. Bultje 07master * rb5d08c27c3 10ffmpeg/libswscale/x86/ (input.asm swscale_mmx.c swscale_template.c): 
[08:05] <CIA-108> ffmpeg: swscale: convert rgb/bgr24ToY/UV_mmx functions from inline asm to yasm.
[08:05] <CIA-108> ffmpeg: Also implement sse2/ssse3/avx versions.
[08:05] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r7063b6eaee 10ffmpeg/ (doc/APIchanges libavcodec/version.h tests/ref/lavf/ffm): 
[08:05] <CIA-108> ffmpeg: lavc: increase major version to 54.
[08:05] <CIA-108> ffmpeg: The lavf-ffm test results change because ffmenc writes
[08:05] <CIA-108> ffmpeg: AVCodecContext.flags/flags2 and the defaults for those change.
[08:05] <CIA-108> ffmpeg: 03Alexandra Khirnova 07master * r570f6aaab1 10ffmpeg/cmdutils.c: 
[08:05] <CIA-108> ffmpeg: cmdutils: fix options starting with 'no'.
[08:05] <CIA-108> ffmpeg: E.g. non_linear_quant mpeg2video private option.
[08:05] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r99ace37e68 10ffmpeg/libavcodec/ (avcodec.h mjpegdec.c options.c version.h): lavc: remove disabled FF_API_MJPEG_GLOBAL_OPTS cruft.
[08:05] <CIA-108> ffmpeg: 03Anton Khirnov 07master * ra17479dfce 10ffmpeg/ (doc/APIchanges libavformat/version.h): lavf: increase major version from 53 to 54.
[08:05] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r2439f2ca82 10ffmpeg/libavformat/ (avio.c avio.h aviobuf.c options.c url.h version.h): lavf: remove disabled FF_API_OLD_AVIO cruft
[08:05] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r38233fc131 10ffmpeg/libavformat/ (avformat.h metadata.c metadata.h version.h): lavf: remove disabled FF_API_OLD_METADATA2 cruft
[08:05] <CIA-108> ffmpeg: 03Anton Khirnov 07master * recd5f41a06 10ffmpeg/libavformat/ (avformat.h utils.c version.h): lavf: remove disabled FF_API_PARSE_DATE cruft
[08:05] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r64f6f6f836 10ffmpeg/libavformat/ (avformat.h img2.c version.h): lavf: remove disabled FF_API_GUESS_IMG2_CODEC cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r997420abf2 10ffmpeg/libavformat/ (avformat.h utils.c version.h): lavf: remove disabled FF_API_DUMP_FORMAT cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * rcc10ab79ec 10ffmpeg/libavformat/ (avformat.h utils.c version.h): lavf: remove disabled FF_API_FIND_INFO_TAG cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r21824e5b88 10ffmpeg/libavformat/ (avformat.h utils.c version.h): lavf: remove disabled FF_API_PKT_DUMP cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r2fb7501938 10ffmpeg/libavformat/ (avformat.h utils.c version.h): 
[08:06] <CIA-108> ffmpeg: lavf: remove disabled FF_API_FORMAT_PARAMETERS cruft
[08:06] <CIA-108> ffmpeg: Also remove now unused AVFormatParameters struct and
[08:06] <CIA-108> ffmpeg: AVOutputFormat.set_parameters().
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r7232bfbd93 10ffmpeg/libavcodec/ (avcodec.h libx264.c libxavs.c options.c version.h): lavc: remove disabled FF_API_X264_GLOBAL_OPTS cruft.
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * rb32e30faa1 10ffmpeg/libavformat/ (avformat.h version.h): lavf: remove disabled FF_API_AVSTREAM_QUALITY cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r2d70912cc2 10ffmpeg/libavformat/ (avformat.h movenc.c options.c version.h): lavf: remove disabled FF_API_FLAG_RTP_HINT cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r3bbfd3026c 10ffmpeg/libavformat/ (avformat.h gif.c version.h): lavf: remove disabled FF_API_LOOP_OUTPUT cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * rdf0bb26aca 10ffmpeg/libavformat/ (avformat.h dvenc.c gxfenc.c movenc.c mxfenc.c version.h): lavf: remove disabled FF_API_TIMESTAMP cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * r6dc345555f 10ffmpeg/libavformat/ (avformat.h img2.c version.h): lavf: remove disabled FF_API_LOOP_INPUT cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * rf0cb13958d 10ffmpeg/libavformat/ (avformat.h mpegenc.c mpegtsenc.c options.c version.h): lavf: remove disabled FF_API_MUXRATE cruft
[08:06] <CIA-108> ffmpeg: 03Anton Khirnov 07master * radad5b88f8 10ffmpeg/libavformat/ (rtsp.c version.h): lavf: remove disabled FF_API_RTSP_URL_OPTIONS cruft
[08:06] <CIA-108> (63 lines omitted)
[08:08] <vtorri> there are also some conversion characters not supported by the printf family on Windows
[08:08] <vtorri> t, T
[08:09] <vtorri> hoo
[08:09] <vtorri> ./libavcodec/w32pthreads.h:226:5: warning: return with a value, in function returning void [enabled by default]
[08:09] <vtorri> :)
[08:13] <vtorri> nevcairiel: do you want the complete output of my build ?
[09:01] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * r72bbe768ba 10ffmpeg/libavdevice/alldevices.c: 
[09:01] <CIA-108> ffmpeg: devices: Disable v4l to fix compilation until someone
[09:01] <CIA-108> ffmpeg: has time to update it to the new API.
[09:01] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[09:16] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * r81ab42a334 10ffmpeg/libavcodec/x86/ (diracdsp_yasm.asm dwt_yasm.asm): 
[09:16] <CIA-108> ffmpeg: dirac_yasm: fix linking failure due to %ifndef
[09:16] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[11:05] <CIA-108> ffmpeg: 03Clément BSsch 07master * r23b639c44f 10ffmpeg/ (6 files in 5 dirs): Fix "ouput" typo and add a check.
[12:41] <CIA-108> ffmpeg: 03Clément BSsch 07master * r3bc92e34a4 10ffmpeg/cmdutils.c: cmdutils: re-add return 0 accidentally remove from opt_protocols().
[15:28] <nevcairiel> I'm seeing some rather puzzling  behaviour - i have the demuxer and the decoder running on completely different format contexts, and they only communicate via AVPackets. Now i am playing a paletted format which has the palette in side-data, however i never forwarded the side data from the demuxer to the decoder
[15:28] <nevcairiel> Still, it decodes fine
[15:28] <nevcairiel> how the hell do the two communicate? :d
[15:38] <saste> nevcariel: telepathy
[15:48] <nevcairiel> seriously, i dont understand why it works
[15:48] <nevcairiel> not that i mind, but i want to know why. :D
[15:51] <saste> nevcariel: yes be happy as long as it works, even if you can't understand why ;)
[15:53] <nevcairiel> oh i see, it puts the side data into the raw data stream
[15:54] <nevcairiel> with merge/split_side_data
[15:54] <nevcairiel> another mystery solved
[18:16] <ubitux> if i need common code between lavf & lavc, where is it prefered to put it? lavc?
[18:45] <Compn> lavu
[18:45] <Compn> :P
[18:46] Action: Compn has no clue
[18:46] <Compn> what does the code do ?
[18:46] <ubitux> it's basically a text format directives array, it doesn't really belongs to lavu
[18:46] <ubitux> it's codec/format specific
[18:56] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r762311ff94 10ffmpeg/ (6 files in 2 dirs): 
[18:56] <CIA-108> ffmpeg: swscale: fix conversion into yuva444p from non-alpha formats
[18:56] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[18:56] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:56] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r7bab43f29c 10ffmpeg/libswscale/yuv2rgb.c: 
[18:56] <CIA-108> ffmpeg: yuv2rgb: better check if source format have alpha
[18:56] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[18:56] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:19] <ubitux> it seems lavc has no dep on lavf, but lavf has deps on lavc
[19:20] <ubitux> mh.
[19:20] <ubitux> it seems lavc has no dep on lavf, but lavf has *no* depson lavc
[19:20] <ubitux> what am i saying&
[19:20] <ubitux> i'm totally braindead tonight
[19:21] <ubitux> lavc has no dep on lavf, lavf has deps on lavc
[19:21] <ubitux> so common code belongs to lavc.
[19:21] <ubitux> pfiu. that was hard.
[19:22] <nevcairiel> i would recommend less drinking
[19:22] <ubitux> i don't drink alcohol at all
[19:22] <ubitux> maybe i should actually.
[19:28] <durandal_1707> av_set_pts_info was not removed
[19:51] <nevcairiel> alot of those removals really seem to make it harder to write an external demuxer
[19:51] <nevcairiel> i'll have to use private symbols again!
[19:52] <durandal_1707> why external demuxer? why not internal one?
[19:53] <nevcairiel> because its way easier to debug when i keep it in the code thats compiled with MSVC instead of the GCC code. :)
[20:09] <durandal_1707> ffmpeg/libav code smells full of memleaks
[20:14] <Compn> lol
[20:15] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * rafe9105f66 10ffmpeg/doc/examples/muxing.c: 
[20:15] <CIA-108> ffmpeg: example: update muxing example to purple API
[20:15] <CIA-108> ffmpeg: Reviewed-by: Paul B Mahol <onemda at gmail.com>
[20:15] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:15] <CIA-108> ffmpeg: 03Michael Niedermayer 07master * rcb73e35660 10ffmpeg/doc/examples/decoding_encoding.c: 
[20:15] <CIA-108> ffmpeg: example: Update decoding/encoding example to the new API.
[20:15] <CIA-108> ffmpeg: Reviewed-by: Paul B Mahol <onemda at gmail.com>
[20:15] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:33] <ubitux> seriously, the ppl who did jacosub were on LSD
[20:33] <ubitux> \D Insert current date into the title. The date will be of the format DD MMM YYYY, as in 2 Apr 1996.
[20:34] <ubitux> wtf is this related to subtitles& ? :D
[20:34] <nevcairiel> wth is jacosub
[20:34] <ubitux> old school ASS
[20:35] <ubitux> for amiga
[20:35] <ubitux> http://unicorn.us.com/jacosub/
[20:35] <durandal_1707> av_log_missing_feature()
[20:36] <ubitux> btw, does anyone know if there is a "reset to default style" tag in ASS?
[20:36] <durandal_1707> maybe for those artistic type of films...
[20:37] <ubitux> oh, \r
[20:37] <ubitux> great.
[20:37] <durandal_1707> i read commit log of mplayer and they ignore \r
[20:38] <durandal_1707> perhaps mplayer have own ass parser ....
[20:38] <ubitux> huh?
[20:40] <durandal_1707> https://github.com/richardpl/mplayer2/commit/1ef4fe59e2c25964c64bd54b198e561bc60e755a
[20:40] <iive> durandal_1707: please name the forks correctly.
[20:40] <durandal_1707> iive: that is mplayer(1)
[20:40] <ubitux> durandal_1707: i meant "{\\r}"
[20:41] <ubitux> ass tag
[20:41] <iive> durandal_1707: why is it named mplayer2 ?
[20:41] <ramiro> hi
[20:42] <durandal_1707> iive: mplayer2 is my own fork of mplayer2, it have branch master,mplayer2,mplayer1 and several others
[20:42] <ramiro> michaelni: which commit exactly broke dshow? I'm somewhat confused with those git merges with lots of commits
[20:45] <durandal_1707> i dont have impression it is broken, michaelni just did not removed old part of code which is already removed in libav
[20:46] <Compn> durandal_1707 : lol @ mplayer2 fork being named mplayer2
[20:47] <durandal_1707> Compn: i could name it mplayer3 and it would certainly cause bit popularity from anime fans
[20:48] <Compn> lol
[20:48] <durandal_1707> Compn: just give me some remote server and i will do that
[20:49] <Compn> i dont need it
[20:49] <Compn> just thought it was funny :)
[20:51] <durandal_1707> Compn: you dont need ordered matroska chapters and gapless playback?
[20:52] <Compn> durandal_1707 : the only use i've seen of ordered chapters has been to watch intros before each anime episode
[20:52] <Compn> and i certainely dont need to watch the same intro 40 times if i'm watching 40 episodes
[20:52] <Compn> once is enough.
[20:53] <nevcairiel> the feature sure has potential (like, mixing theatrical and directors cut of a movie into one file), but its only used for anime intros, so .. meh!
[20:53] <Compn> and gapless i just dont care about
[20:53] <Compn> not saying other people, just me
[21:16] <durandal_1707> michaelni: should i stop spamming ml?
[21:20] <michaelni> ramiro, 2fb7501938b7103624c9bef740ca498258cacdab
[21:20] <michaelni> this removed AVFormatParameters
[21:21] <michaelni> codec->time_base  = ap->time_base; <---- this used AVFormatParameters
[21:21] <michaelni> durandal_1707, please continue sending patches
[21:22] <michaelni> the more the better
[21:23] <ramiro> michaelni: can I just set codec->time_base to (AVRational){framerate.den, framerate.num}?
[21:23] <ramiro> does it make sense?
[21:24] <durandal_1707> michaelni: vorbisenc, libfaac, alacenc (probably forgot about others) are harder to fix because they do not check for allocations failures at all 
[21:24] <michaelni> makes sense based on my limited knowledge of dshow.c
[21:25] <ramiro> libavdevice/v4l.c still has AVFormatParameters, shouldn't it be removed?
[21:26] <michaelni> yes
[21:26] <michaelni> configure didnt enable it here so i missed it
[21:26] <michaelni> and just quickly disabled in alldevices before going to bed to fix compile
[21:28] <durandal_1707> did 12/20
[21:49] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r7b21b0f155 10ffmpeg/libavcodec/adxenc.c: 
[21:49] <CIA-108> ffmpeg: adxenc: check return value of avcodec_alloc_frame()
[21:49] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[21:49] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:49] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r88fb935c1c 10ffmpeg/libavcodec/ac3enc.c: 
[21:49] <CIA-108> ffmpeg: ac3enc: check return value of avcodec_alloc_frame()
[21:49] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[21:49] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:49] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r82c252972b 10ffmpeg/libavcodec/adpcmenc.c: 
[21:49] <CIA-108> ffmpeg: adpcmenc: check return value of avcodec_alloc_frame()
[21:49] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[21:49] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:49] <CIA-108> ffmpeg: 03Paul B Mahol 07master * rf870fc2fb9 10ffmpeg/libavcodec/adpcmenc.c: 
[21:49] <CIA-108> ffmpeg: adpcmenc: refactor some code.
[22:01] <CIA-108> ffmpeg: 03Reimar Döffinger 07master * r0b378e8aa9 10ffmpeg/ (libavcodec/dfa.c tests/ref/fate/dfa4): 
[22:01] <CIA-108> ffmpeg: DFA: fix "skip frame" TSW1 encoding.
[22:01] <CIA-108> ffmpeg: Previously the decoder would raise an error.
[22:01] <CIA-108> ffmpeg: The end result is the same, the time stamps only change
[22:01] <CIA-108> ffmpeg: because regression tests create time stamps incorrectly.
[22:01] <CIA-108> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[22:01] <CIA-108> ffmpeg: 03Reimar Döffinger 07master * ra0eb8de5e0 10ffmpeg/ (libavcodec/eacmv.c libavformat/electronicarts.c): 
[22:01] <CIA-108> ffmpeg: Merge palette-only header with actual video frame.
[22:01] <CIA-108> ffmpeg: This fixes the video frame pts (off by one for each MVIh)
[22:01] <CIA-108> ffmpeg: and makes the "key frames" decode stand-alone (MVIh
[22:01] <CIA-108> ffmpeg: contains only palette, such a palette-only frame being
[22:01] <CIA-108> ffmpeg: marked as key frame is not really correct).
[22:01] <CIA-108> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[22:01] <CIA-108> ffmpeg: 03Reimar Döffinger 07master * r333f5b52e3 10ffmpeg/tests/fate/ (8 files): 
[22:01] <CIA-108> ffmpeg: Remove unnecessary -vsync 0.
[22:01] <CIA-108> ffmpeg: The tests work fine without it, and it will cause issues when the
[22:01] <CIA-108> ffmpeg: rawvideo decoder is changed to properly handle pts values.
[22:01] <CIA-108> ffmpeg: The H.264 conformance tests however are still broken, usually losing
[22:01] <CIA-108> ffmpeg: the first frames without it.
[22:01] <CIA-108> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[22:01] <CIA-108> ffmpeg: 03Reimar Döffinger 07master * rbb2d42c22f 10ffmpeg/libavformat/framecrcenc.c: (log message trimmed)
[22:02] <CIA-108> ffmpeg: Set AVFMT_TS_NONSTRICT for framecrc muxer.
[22:02] <CIA-108> ffmpeg: Since it is set for e.g. webm muxer we should make it possible
[22:02] <CIA-108> ffmpeg: to test such streams with framecrc, too.
[22:02] <CIA-108> ffmpeg: Though the primary reason is that this allows the H.264 tests
[22:02] <CIA-108> ffmpeg: to not run into this check when fixing raw video encode to
[22:02] <CIA-108> ffmpeg: pass pts values on.
[22:02] <CIA-108> ffmpeg: 03Reimar Döffinger 07master * r34751f8313 10ffmpeg/libavformat/mov.c: (log message trimmed)
[22:02] <CIA-108> ffmpeg: Demux all parts of concatenated H.264.
[22:02] <CIA-108> ffmpeg: We may or may not be able to play the latter parts
[22:02] <CIA-108> ffmpeg: but not demuxing at all seems like the worst possible behaviour.
[22:02] <CIA-108> (13 lines omitted)
[22:08] <saste> michaelni: what is "purple API"?
[22:08] <ubitux> bikeshed API?
[22:08] <saste> ah got it :)
[22:09] <ubitux> it seems my jacosub encoder is starting to work
[22:09] <michaelni> :)
[22:10] <durandal_1707> michaelni: why ff_get_unscaled_swscale have check for YUVA420P?
[22:13] <durandal_1707> i think that check is not needed at all
[22:13] <michaelni> which ?
[22:15] <durandal_1707> YUVA420p to YUV420p and reverse
[22:16] <durandal_1707> \ /* simple copy */
[22:17] <michaelni> hmm, agree seems redundant
[22:18] <durandal_1707> hmm, need to check yuva444p to and from nv21
[22:24] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r3ac99cdc41 10ffmpeg/libavcodec/libgsm.c: 
[22:24] <CIA-108> ffmpeg: libgsm: check return value of gsm_create & avcodec_alloc_frame()
[22:24] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[22:24] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:24] <CIA-108> ffmpeg: 03Paul B Mahol 07master * ra136c2cf3e 10ffmpeg/libavcodec/libmp3lame.c: 
[22:24] <CIA-108> ffmpeg: libmp3lame: remove dupe assigment
[22:24] <CIA-108> ffmpeg: It is already done by avcodec_alloc_context().
[22:24] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[22:24] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:24] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r5291cedd6c 10ffmpeg/libavcodec/libgsm.c: 
[22:24] <CIA-108> ffmpeg: libgsm: remove dupe assigment.
[22:24] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[22:24] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:24] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r766e160e87 10ffmpeg/libavcodec/libopencore-amr.c: 
[22:24] <CIA-108> ffmpeg: libopencore-amr: check return value of avcodec_alloc_frame()
[22:24] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[22:25] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:25] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r2de6a7253e 10ffmpeg/libavcodec/roqaudioenc.c: 
[22:25] <CIA-108> ffmpeg: roqaudioenc: remove redundant assignment
[22:25] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[22:25] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:25] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r3ab90687fd 10ffmpeg/libavcodec/roqaudioenc.c: 
[22:25] <CIA-108> ffmpeg: roqaudioenc: check return value of avcodec_alloc_frame()
[22:25] <CIA-108> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[22:25] <CIA-108> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:25] <CIA-108> ffmpeg: 03Paul B Mahol 07master * r1fa24d08b2 10ffmpeg/libavcodec/pcm.c: 
[22:25] <CIA-108> (19 lines omitted)
[22:47] <ubitux> mmh about forgotten patches
[22:48] <ubitux> hasn't Aurelien Jacobs submitted a MicroDVD decoder?
[23:01] <michaelni> google finds a demuxer & muxer but i see no decoder, maybe iam searching for the wrong terms
[23:01] <ubitux> ok, then if it's only the demuxer & muxer it's upstream
[23:02] <ubitux> too bad we can't do much without the decoder
[23:03] <ubitux> is it common practice to have a common context struct layout between demuxer and decoder?
[23:03] <ubitux> in case of subtitles, i need to parse the timing stuff two times
[23:03] <ubitux> at demuxer level, and then at decoder level
[23:05] <ubitux> in the decoder, i can fetch it from the demuxer packet since it only has the start time, and not the duration
[23:08] <ubitux> (and i need a common context because there is context time shift and scale)
[23:15] <michaelni> demuxer and decoder cant access each others private context 
[23:15] <michaelni> if thats what you suggest
[23:15] <michaelni> they could run in different threads
[23:16] <michaelni> with fifo between them ...
[23:18] <ubitux> oh?
[23:18] <ubitux> how can i do that?
[23:18] <ubitux> (i mean access demuxer context in decoder scope)
[23:21] <CIA-108> ffmpeg: 03Reimar Döffinger 07master * rdbc7c6121e 10ffmpeg/libavcodec/libgsm.c: 
[23:21] <CIA-108> ffmpeg: libgsm: Fix compilation.
[23:21] <CIA-108> ffmpeg: Move libgsm_encode_close before its first use and call it
[23:21] <CIA-108> ffmpeg: with the correct number of arguments.
[23:21] <CIA-108> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[23:23] <ubitux> michaelni: through extradata, or another mean?
[23:46] <michaelni> ubitux, extradata, sidedata, normal AVPacket data
[23:48] <CIA-108> ffmpeg: 03Reimar Döffinger 07master * r97dece333d 10ffmpeg/libavcodec/rawenc.c: 
[23:48] <CIA-108> ffmpeg: rawenc: initialize coded_frame via avcodec_get_frame_defaults.
[23:48] <CIA-108> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[23:51] <ubitux> mmh
[23:51] <ubitux> ok
[23:52] <ubitux> thank you.
[00:00] --- Sun Jan 29 2012


More information about the Ffmpeg-devel-irc mailing list