[Ffmpeg-devel-irc] ffmpeg-devel.log.20121105

[burek]

[00:02] <michaelni> mike_f see av_log_set_level() and av_log_set_callback()
[00:06] <durandal_1707> michaelni: why lvf-flm is failing for some machines?
[00:14] <michaelni> didnt realize it was failing
[00:14] <mike_f> Thanks. I'll see if I can get av_log_set_level and av_log_set_callback to give me an idea of what's wrong. :)
[00:16] <mike_f> I think I'll try AV_LOG_VERBOSE first. :)
[00:22] <michaelni> durandal_1707, found it, was off by 1 rounding in the muxer
[00:22] <mike_f> It's telling my that I'm referencing an invalid PPS. I'm pretty sure that I force outputting a SPS/PPS on the first frame (though I shouldn't really have to force this).
[00:23] Action: durandal_1707 curses on useless git blame
[00:23] <durandal_1707> michaelni: what bug?
[00:23] <durandal_1707> michaelni: aha got it.... it came back to me
[00:24] <durandal_1707> if found another stupid and extremly old bug i try to find owner of it ...
[00:24] <durandal_1707> *I
[00:25] <durandal_1707> this is imposible mission: rename,rename,cosmetics,cosmetics,reindent,reindent,reaname ..... f*******
[00:28] <cone-190> ffmpeg.git 03Paul B Mahol 07a0348d0966a8: pnm: remove nonsense code
[00:29] <cone-190> ffmpeg.git 03Michael Niedermayer 071ba0d9b5d110: oss: warn when non block mode cannot be enabled.
[00:29] <cone-190> ffmpeg.git 03Michael Niedermayer 07a147c1b2b125: oggenc: remove unneeded null check.
[00:29] <cone-190> ffmpeg.git 03Michael Niedermayer 07e82afa3e2d63: filmstripenc: fix rounding that caused fate failure
[00:30] <durandal_1707> michaelni: funny, so some compilations did different rounding?
[00:32] <michaelni> it differed on 32 vs 64bit x86 it seems
[00:32] <cbsrobot> dmon_: that question belongs to #ffmpeg
[00:38] <cone-190> ffmpeg.git 03Paul B Mahol 07ae5ecb8e60ad: pnm: remove duplicated check for maxval > 255
[00:38] <cone-190> ffmpeg.git 03Paul B Mahol 07cc5bcaa48e6a: pnm: cosmetics: reindent
[00:40] <Daemon404> yay msvc failures
[00:41] <durandal_1707> missing sample - ignore
[00:41] <Daemon404> if it was a missing sample it wouldnt have output
[00:41] <Daemon404> now would it.
[00:41] <Daemon404> it has output -- it just differs
[00:42] <durandal_1707> hmm
[00:42] <Daemon404> im interested to see how mingw does
[01:54] <Daemon404> will libavcodec's mpeg2video decoder set use teh idct algo i specify in my avctx's idct_algo member?
[02:22] <cone-190> ffmpeg.git 03Michael Niedermayer 07b015cd7eda9f: fate: try to fix mxpeg failures
[04:45] <cone-190> ffmpeg.git 03Carl Eugen Hoyos 07c673ef12885f: Write tiff tag YCbCrPositioning if chroma_sample_location is top-left.
[06:29] <burek> btw, I'm gonna put the tutorial about compiling ffmpeg on raspberry pi and if someone here has already done it, well, please let me know so that we don't do things 2 times :)
[06:29] <burek> put at the wiki*
[06:34] <burek> building vlc now :)
[06:41] <adnap> i don't have av_err2str in my libavutil/error.h as referenced here: http://ffmpeg.org/doxygen/trunk/error_8h-source.html i have version 0.10.3 of media-video/ffmpeg installed. what version includes this macro?
[07:29] <ubitux> adnap: according to doc/APIChanges:
[07:29] <ubitux> 2012-04-11 - 359abb1 - lavu 51.58.100 - error.h Add av_make_error_string() and av_err2str() utilities to libavutil/error.h.
[07:30] <ubitux> version 0.10.x is getting old
[07:33] <cone-190> ffmpeg.git 03Clément BSsch 07a8b3130181f6: lavf/srtdec: recognize subtitles starting at event index 0.
[10:34] <ubitux> http://pastie.org/5188316
[10:35] <ubitux> ’ http://pastie.org/5188321
[10:36] <ubitux> (there is no problem, just a random code to show how to use the api)
[10:36] <ubitux> (at least how i think it is supposed to be used)
[12:26] <cone-480> ffmpeg.git 03Paul B Mahol 0717e67c532100: fate: add smjpegenc test
[12:53] <durandal_1707> anyone have pmp(Playstation Portable) files?
[12:55] <JEEB> I used to have those, it's a simplistic container made by people before Sony let people play H.264 from MP4
[12:56] <JEEB> lemme see if I still have some
[12:56] <JEEB> ever since Sony started supporting MP4 with H.264 the need for that format was removed
[12:58] <Compn> theres some files on repo
[12:58] <Compn> durandal_1707
[12:59] <Compn> http://samples.ffmpeg.org/playstation/psp/
[13:00] <durandal_1707> JEEB: there is still missing support for AAC in pmp
[13:00] <JEEB> I think I don't have any of that kind
[13:01] <JEEB> back when I was making those files it was all AVC+MP3 ÖD
[13:01] <JEEB> :D
[13:09] <JEEB> man, watching these old clips of mine made me all nostalgic :<
[13:09] <nevcairiel> last i heard, there was something truely weird about aac in pmp
[13:09] <JEEB> there probably was, the format was something really simple hacked up by the people at ps2dev.org
[13:10] <JEEB> (or the AAC support was then added by the Chinese who IIRC took the development of those PMP players after most people moved to MP4)
[13:19] <Prottey> how can I test a customly patched ffmpeg/libav tree while another version is being already installed from distr repos?
[13:19] <Prottey> w/o any need to uninstall the libs
[13:22] <JEEB> set a prefix and then add the prefix's /include and /lib into the related search paths when compiling your thing that you're using the libraries from
[13:23] <Prottey> ok thanks
[13:23] <JEEB> also with custom prefixes I generally recommend static linking because otherwise you'll have to have fun with setting the loading paths for shared libraries
[13:23] <JEEB> LD_LOAD_PATH or whatever it was when trying to run the linked thingy
[13:24] <JEEB> or I guess LD_LIBRARY_PATH
[13:40] Action: durandal_1707 gonna push avr demuxer
[14:01] <cone-480> ffmpeg.git 03Michael Niedermayer 07f4aaf987a588: frame_thread_encoder: fix handling of case where the encoder does not output a packet.
[14:39] <cone-480> ffmpeg.git 03Paul B Mahol 07e870a7dd4ab1: lavfi: stop using -1 instead use AV_PIX_FMT_NONE/AV_SAMPLE_FMT_NONE
[16:00] <cone-480> ffmpeg.git 03Paul B Mahol 072a2d39c72110: fate: add more pix_fmt tests for pam
[16:24] <durandal_1707> michaelni: zmbv decoder does some silly things like converting anything to rgb24, that is not optimal. It should switch pix_fmt instead.
[16:24] <durandal_1707> is midstream change of pix_fmt supported?
[16:25] <nevcairiel> lavfi will probably crash, but it crashes on basically any parameter change =p
[16:25] <michaelni> durandal_1707, yes
[16:26] <michaelni> nevcairiel, submit a bug report please
[16:26] <durandal_1707> michaelni: is there sample which do this?
[16:27] <michaelni> concatenate 2 mpeg files (ES PS TS hsould make no diff)
[16:27] <michaelni> with different resolutions or pixel formats
[16:27] <michaelni> i think resolution changes do work
[16:27] <michaelni> i think we even have a fate test for that
[16:38] <cone-480> ffmpeg.git 03Michael Niedermayer 079829ec1a9c0d: ffm: redesign header format to make it extensible
[16:38] <cone-480> ffmpeg.git 03Michael Niedermayer 07b2b67fd610b8: docs/ffserver: document ffm/ffm2
[16:54] <durandal_1707> hmm, why if i mux video only into mpeg there is no any pts?
[16:58] <durandal_1707> changing pix_fmt works fine with ffplay
[17:31] <TimNich>  Hmmm, so why is the fate test for DNX set to use interlaced coding from a progressive source&.
[17:34] <durandal_1707> thougth it does not work for >1 threads ...
[17:43] <cone-480> ffmpeg.git 03Piotr Bandurski 07461797f6aac5: riff: add "lsvm" FourCC for H263.
[17:44] <cone-480> ffmpeg.git 03Piotr Bandurski 078d88920578a2: dpcm: use AV_LOG_ERROR for error message.
[17:48] <merbanan> TimNich: for a regression test it doesn't matter what the source is
[17:51] <TimNich> merbanan: Well it does if you want to start handling interlaced flagged material intelligently!
[17:52] <merbanan> TimNich: elaborate, I don't follow your reasoning
[17:55] <TimNich> at the moment the round trip raw video->dnx->raw video assumes that you handle things progressively, *but* as the dnx is flagged as interlaced and if you handle it as such then the psnr drops through the floor and may mask other issues if you make that the default.
[17:58] <durandal_1707> michaelni: img2enc have same strange looking code for handling jpeg2000 codecs but that should not be there at all
[17:58] <TimNich> r, to put it another way, if the fate test used progressive on the round trip one could experiment with interlace handling without constantly having to worry about the fate tests being compromised. Then later on make a proper fate test for interlace.
[17:58] <merbanan> well I just assumed the dnx regtest uses crc
[17:59] <TimNich> its not the checksum change that's the big problem, but the psnr, same for the DV tests.
[18:24] <durandal_1707> why i get 00000.... when muxing into sox
[18:32] <cone-480> ffmpeg.git 03Paul B Mahol 07962314fe2769: AVR demuxer
[19:11] <llogan> ubitux: you mention -qscale 0 in trac 1835. i thought the scale for mpeg* was 1-31 (i'm a lazyass. I didn't look at the code)
[19:20] <ubitux> isn't it the generic lossless value?
[19:20] <ubitux> feel free to correct me if something is wrong
[19:38] <llogan> ubitux: i don't know.
[19:58] <michaelni> TimNich, about the field patch, if you dont want / dont have the time to implement the passing prior to the header muxing. Then the patch should be ok as it is just not optimal  ...
[19:59] <michaelni> i also suspect the avi flags arent used by much as noone complained to them being as they are
[20:46] <ubitux> ods15: no way to remove the ugly buffer[50000] from the vorbisenc?
[20:52] <cone-480> ffmpeg.git 03Michael Niedermayer 0730d27685b177: segment: fix null ptr deref
[21:02] <durandal_1707> omg what an awfull hack is in img2enc
[21:35] <cone-480> ffmpeg.git 03Michael Niedermayer 07282a6308bfd2: udp: check ff_socket_nonblock() return code
[21:47] <cone-480> ffmpeg.git 03Paul B Mahol 07efbe162f7dc2: cljr: use meaningful error code
[21:47] <cone-480> ffmpeg.git 03Paul B Mahol 0770727e8774dd: lavf/avr: make .long_name consistent with others
[21:47] <cone-480> ffmpeg.git 03Paul B Mahol 0710ac70d249c7: avrndec: use more meaningful error codes
[21:48] <cone-480> ffmpeg.git 03Martin Storsjö 07f0615557e035: rtmpproto: Use av_strlcat instead of strncat
[22:40] <cone-480> ffmpeg.git 03Martin Storsjö 07d578f94746dd: rtmp: Use av_strlcat instead of strncat
[22:40] <cone-480> ffmpeg.git 03Diego Biurrun 076c104826bd6e: x86: vc1dsp: port to cpuflags
[22:40] <cone-480> ffmpeg.git 03Diego Biurrun 07dbb37e771174: x86: PABSW: port to cpuflags
[22:40] <cone-480> ffmpeg.git 03Diego Biurrun 07930e26a3ea9d: x86: h264qpel: Only define mmxext QPEL functions if H264QPEL is enabled
[22:40] <cone-480> ffmpeg.git 03Michael Niedermayer 07e859339e7af5: Merge commit '930e26a3ea9d223e04bac4cdde13697cec770031'
[23:00] <cone-480> ffmpeg.git 03Diego Biurrun 0792fec47d69d2: build: The A64 muxer depends on rawenc.o for ff_raw_write_packet()
[23:00] <cone-480> ffmpeg.git 03Diego Biurrun 072253df76c191: fate: aac: Place LATM tests and general AAC tests in different groups
[23:00] <cone-480> ffmpeg.git 03Justin Ruggles 076a744d261930: flacenc: use a separate buffer for byte-swapping for MD5 checksum on big-endian
[23:00] <cone-480> ffmpeg.git 03Justin Ruggles 07dfde8a34e541: lavu: add av_ctz() for trailing zero bit count
[23:00] <cone-480> ffmpeg.git 03Justin Ruggles 07e78331632208: flacenc: remove wasted trailing 0 bits
[23:00] <cone-480> ffmpeg.git 03Justin Ruggles 075ff998a233d7: flacenc: use uint64_t for bit counts
[23:00] <cone-480> ffmpeg.git 03Michael Niedermayer 077d26be63c25e: Merge commit '5ff998a233d759d0de83ea6f95c383d03d25d88e'
[23:21] <cone-480> ffmpeg.git 03Justin Ruggles 07799e2324901c: flacdsp: move lpc encoding from FLAC encoder to FLACDSPContext
[23:21] <cone-480> ffmpeg.git 03Justin Ruggles 0713e1ee6c84f0: flacenc: add 24-bit encoding
[23:21] <cone-480> ffmpeg.git 03Justin Ruggles 072e76f3438776: flacenc: use RICE2 entropy coding mode for 24-bit
[23:21] <cone-480> ffmpeg.git 03Justin Ruggles 0731c3b9c95e2b: FATE: rename FLAC tests from flac-* to flac-16-*
[23:21] <cone-480> ffmpeg.git 03Justin Ruggles 078a58894fc63c: FATE: add a 24-bit FLAC encoding test
[23:21] <cone-480> ffmpeg.git 03Michael Niedermayer 07649384290089: Merge remote-tracking branch 'qatar/master'
[23:25] <juanmabc> is someone thinking/caring about this: https://ffmpeg.org/trac/ffmpeg/ticket/747 ?
[23:26] <durandal_1707> i plant to add smjpeg adpcm encoder (and so amv variant too) 
[23:27] <juanmabc> cool, short , medium or long term?
[23:27] <juanmabc> :D
[23:30] <durandal_1707> it may happen in next hour or in next 10 years
[23:30] <juanmabc> well, at least sure not in next 10 min
[23:34] <durandal_1707> juanmabc: did you mention some kind of bounty?
[23:35] <juanmabc> not me, but the reporter it seems
[23:40] <kierank> 1yr ago
[23:45] <Daemon404> the adpcm codec is easy
[23:45] <Daemon404> by adpcm*c are horrible
[23:45] <durandal_1707> huh, if i use ccache lib version does not change
[23:45] <Daemon404> durandal_1707, tip: ccache is useless with ffmpeg
[23:46] <durandal_1707> Daemon404: not at all on crappy machine like this one
[23:47] <Daemon404> durandal_1707, i mean its useless cause config.h changes every time you run configure
[23:47] <Daemon404> and thus teh checksum changes
[23:47] <Daemon404> and it recompiles anyway
[23:48] <durandal_1707> well if i do not run configure lib minor will not change?
[23:49] <Daemon404> (the part in config.h that changes is like, the date)
[23:49] <Daemon404> the timestamp
[23:49] <Daemon404> so -any- time you re-run it
[23:49] <Daemon404> i wish there was a way to disable it
[23:50] <cone-480> ffmpeg.git 03Paul B Mahol 07c99564492c57: fate: add more pix_fmt tests for xwd
[23:50] <Daemon404> nevcairiel, did you ever peel at the fate failure on msvc
[23:50] <Daemon404> peek*
[00:00] --- Tue Nov  6 2012