[Ffmpeg-devel-irc] ffmpeg-devel.log.20131003

[burek]

[00:56] <cone-82> ffmpeg.git 03Thilo Borgmann 07release/2.0:842d7c9b3af0: configure: fix logic for threads in case of OpenCL is enabled.
[00:56] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/2.0:f089e67d515b: avcodec/imgconvert/get_color_type: fix type for PAL8
[01:26] <cone-82> ffmpeg.git 03Timothy Gu 07release/1.2:82cfb8c2db6c: doc/encoders: reformat libmp3lame doc
[01:27] <cone-82> ffmpeg.git 03Timothy Gu 07release/1.2:0f73cb454d14: doc/encoders: reformat and add some clarification in libtwolame doc
[01:27] <cone-82> ffmpeg.git 03Timothy Gu 07release/1.2:d2feaf2ba613: doc/encoders: Remove options that were not there when branch was cut from master
[01:27] <cone-82> ffmpeg.git 03Timothy Gu 07release/1.2:4431ee189643: doc/ffmpeg-formats: Add documentation for 3 parameters that have been missing
[01:27] <cone-82> ffmpeg.git 03Timothy Gu 07release/1.2:85eeab4d22cc: doc/encoders: improve libvo-aacenc doc
[01:27] <cone-82> ffmpeg.git 03Timothy Gu 07release/1.2:6db67ac2a4fa: doc/encoders: add doc for AAC encoder
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:9902eef0f531: avcodec/ffv1dec: check global header version
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:bb4126e250e0: ffv1dec: check that global parameters dont change in version 0/1
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:91e19ab9306c: ffv1dec: Check bits_per_raw_sample and colorspace for equality in ver 0/1 headers
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:a0d13f578b85: avcodec/dsputil: fix signedness in sizeof() comparissions
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:0f3cdddf3886: avfilter/vf_fps: make sure the fifo is not empty before using it
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:c17fd9f9ce2e: avcodec/mjpegdec: Add some sanity checks to ljpeg_decode_rgb_scan()
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:c3ee5b4c3641: avcodec/truemotion2: Fix av_freep arguments
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:f32051cd733d: avcodec/ffv1enc: update buffer check for 16bps
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:95e26d33d76d: avcodec/parser: reset indexes on realloc failure
[01:27] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:ca47aec665e2: Merge remote-tracking branch 'TimothyGu/release/1.2' into release/1.2
[01:36] <cone-82> ffmpeg.git 03Michael Niedermayer 07release/1.2:14e258d847f7: update for 1.2.4
[01:50] <cone-82> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n1.2.4': unknown revision or path not in the working tree.
[01:50] <cone-82> Use '--' to separate paths from revisions
[01:50] <cone-82> refs/tags/n1.2.4:HEAD: ffmpeg: add support to force specific AVCodecs
[02:42] <cone-82> ffmpeg.git 03Carl Eugen Hoyos 07master:b5f219c904b9: Print pix_fmt name on h264 context reinitialisation.
[03:05] <BBB> llogan: just toss it, should be ok
[03:43] <llogan> BBB: ok. sorry about that.
[03:44] <llogan> -devel size limit is now 550k
[04:58] <Compn> BBB : you can blame me for not checking the ml queue :)
[09:32] <cone-480> ffmpeg.git 03Stefano Sabatini 07master:f5b27b6d5f86: ffprobe: print stream channel_layout when available
[10:48] <ubitux> it's interesting, the difference i have with audio with the double seeking is similar to the video difference present in older ffmpeg
[10:56] <saste> ubitux, why are you double-seeking?
[10:56] <ubitux> saste: because it's faster?
[10:56] <ubitux> (and still accurate)
[10:58] <saste> ubitux: comments on av_get_utf8()?
[10:58] <ubitux> not yet but i trust reimar & nicolas on this one ;)
[10:59] <ubitux> saste: double seeking should be done internally :(
[11:54] <cone-480> ffmpeg.git 03Vittorio Giovara 07master:06e7301a2efa: vf_interlace: do not process an already interlaced frame
[11:54] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:c4f3b55f5079: Merge commit '06e7301a2efa8dededecbe6836ee5b59bf7e5e76'
[12:21] <cone-480> ffmpeg.git 03Anton Khirnov 07master:451606f7ce7e: lavc: mark deprecated AVCodec.max_lowres for removal on next bump
[12:21] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:5a63fb66936a: Merge remote-tracking branch 'qatar/master'
[12:36] <BBB> hm github is dead
[12:46] <Daemon404> DDoS
[12:46] <Daemon404> skiddies love it
[13:07] <j-b> Compn: money has been resent, but we need more info from you
[13:21] <BBB> michaelni: need me to send the patch by hand? I can also resend to ML
[13:50] <Compn> j-b : called my bank and got info. sorry i'm so much trouble :)
[13:52] <j-b> Compn: no, you should be sorry that your banking system is such a crap
[13:58] <Compn> lol
[13:59] <cone-480> ffmpeg.git 03Ronald S. Bultje 07master:848826f527bc: Native VP9 decoder.
[13:59] <cone-480> ffmpeg.git 03Ronald S. Bultje 07master:c07ac8d467f0: VP9 MC (ssse3) optimizations.
[13:59] <cone-480> ffmpeg.git 03Ronald S. Bultje 07master:f1548c008fc5: Full-pixel MC functions.
[13:59] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:4c292d5c34c9: Merge remote-tracking branch 'rbultje/master'
[14:00] <ubitux> yay.
[14:00] <Compn> psh, we want vp7 decoder ...
[14:01] <Compn> :)
[14:01] <ubitux> feed your sample to the vp9 decoder
[14:01] <ubitux> and see if it works
[14:01] <ubitux> if it doesn't, pretend the sample is broken
[14:01] <Compn> also old duck decoders, if theres any left we dont have
[14:02] <ubitux> ok i guess i have to finish the lut now
[14:15] <ubitux> ...and the optim
[16:07] <durandal_1707> who is this guy in all those vp9 samples on fate?
[16:20] <cone-480> ffmpeg.git 03Marton Balint 07master:5b5bb0d81c53: avfilter/vf_histogram: fix segfault in parade waveform mode for subsampled formats
[16:20] <cone-480> ffmpeg.git 03Marton Balint 07master:0362cf1a1928: avfilter/vf_histogram: add mirrored waveform mode
[16:40] <Compn> durandal_1707 : why, are you in looooove ?
[16:42] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:e57dba0d52df: avcodec: add av_codec_get_max_lowres()
[16:43] <durandal_1707> Compn: after watching it so many times.. certainly not
[16:45] <Compn> durandal_1707 : ah i dont have updated software to look at those samples :D
[16:46] <Compn> maybe grab a nightly build
[16:51] <durandal_1707> should telecine filter set frame flags?
[17:14] <durandal_1707> saste: why no comment mergeplanes?
[17:17] <saste> durandal_1707, i'll comment on it now
[17:20] <saste> what's the assumed range of GRAY8? (I keep discarding about that)
[17:21] <wm4> use the range flag?
[17:21] <wm4> at least I hope it works this way
[17:26] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:4d5d905eed1e: ffmpeg: avoid direct access to lowres use av_codec_g/set_lowres()
[17:27] <Daemon404> wm4, wishful thinkign on your part
[17:30] <wm4> ok it doesn't
[17:30] <wm4> not sure if my code's fault or libswscale
[17:37] <wm4> probably swscale's
[17:42] <Daemon404> is lavfi (in ffmpeg) threaded for audio too?
[17:43] <durandal_1707> it should not matter (really) its just no such filter use it
[17:43] <Daemon404> i see
[17:44] <Daemon404> im noticing our transcoder boxes randomly have hugn ffmpeg prcesses
[17:44] <Daemon404> waiting at pthread_join in lavfi
[17:44] <Daemon404> and its unreproducible
[17:44] <Daemon404> sounds like a race condition to me
[17:44] <Daemon404> and it's happening quite often
[17:44] <durandal_1707> well that filtered code is from *cough* libav
[17:44] <Daemon404> im pretty sure the lavfi threading is from ffmpeg
[17:45] <Daemon404> unless im mistaken
[17:45] <durandal_1707> that one
[17:45] <saste> Daemon404, kill the process and analyze the core, it is sometimes useful
[17:46] <durandal_1707> well, everybody ignores tsan reports
[17:46] <saste> yes and i suppose there are better tools to debug those problems nowadays, tsan or valgrind
[17:47] <durandal_1707> Daemon404: what filters such boxes use?
[17:47] <Daemon404> saste, you mean helgrind
[17:47] <Daemon404> durandal_1707, a variety
[17:47] <Daemon404> im looking into it...
[17:51] <durandal_1707> they didn't hung before? or this is first time threading in lavfi is used?
[17:52] <Daemon404> never used to happen
[17:54] <saste> Daemon404, lavfi slice threading was recently added
[17:54] <Daemon404> yeah
[17:54] <Daemon404> too bad we were using stripped binaries
[17:54] <Daemon404> the bt is fairly useless
[17:55] <Daemon404> #0  0x00000030b7c080ad in pthread_join () from /lib64/libpthread.so.0
[17:55] <Daemon404> #1  0x00007fc6c5c2b727 in ?? () from /usr/local/vimeo/lib/libavfilter.so.3
[17:55] <Daemon404> #2  0x00007fc6c5c2b910 in ?? () from /usr/local/vimeo/lib/libavfilter.so.3
[17:55] <Daemon404> #3  0x00007fc6c5c0c975 in avfilter_graph_alloc_filter ()
[17:57] <saste> weird
[17:57] <saste> what's doing thread #4?
[17:57] <Daemon404> not sure
[17:57] <Daemon404> will check
[17:57] <Daemon404> colleague is working on it currently
[17:58] <saste> i'd expect the join to be called only after all the filters were created
[17:58] <saste> and i meant thread #3 of course
[17:59] <Compn> Daemon404 : hey, can i get a report of rare / unusual codecs being uploaded by chance? i just need fourcc and isom really. samples are a plus but i understand if not.
[18:00] <Compn> and are vim** using mencoder like yout**** ?
[18:00] <Daemon404> no
[18:00] <Daemon404> C api + cli elsewhere
[18:00] <Daemon404> Compn, might be a pain, but sounds interesting
[18:00] <Daemon404> right now teh big ones are still G2M4 and canopus
[18:00] <Daemon404> with some cineform
[18:00] <Daemon404> (aka VC-5 / GoPro)
[18:01] <Compn> yeah, theres some cineform reports already
[18:01] <Compn> cfhd i think
[18:01] <Compn> g2m4 i thought was completed ... only g2m2 was left ?
[18:01] <Daemon404> er yeah i mean 2
[18:01] <Compn> ah
[18:02] <Daemon404> well also some otehr g2m
[18:02] <Compn> but yeah, i'd appreciate it if you could get some lists. no rush
[18:02] <Daemon404> mss2 still faisl sometimes
[18:02] <Daemon404> due to ffmpeg's asf demuxer
[18:02] <Compn> asf demuxer and rm demuxer need rewrites
[18:02] <Compn> or ... anyone to work on them :P
[18:02] <Compn> and samples
[18:03] <Daemon404> never get complaints about .rm
[18:03] <Daemon404> probably because all .rm content is pirated
[18:03] <Daemon404> like 100% of it ever
[18:03] <Compn> the newer stuff is fine i think
[18:03] <Compn> problem with older or mbr crap
[18:03] <Compn> newer stuff is just dvdrips. dont need mbr for it :P
[18:03] <durandal_1707> no audio codecs?
[18:04] <Compn> yea i could use audio codec list too. but those are much better with samples
[18:04] <Compn> without samples its guess work. a lot of the codecs are decode only, so its impossible for me to make samples
[18:07] <Daemon404> almost all failed audio codecs are various aac variants
[18:07] <Daemon404> like SSR
[18:07] <Daemon404> or ELD
[18:11] <saste> durandal_1707, why there are several versions of your patch?
[18:12] <saste> did i review the last one? what about the extended mode syntax?
[18:15] <durandal_1707> saste: it cant work as you can not set format in config_output
[18:17] <durandal_1707> saste: there are several versions because I'm doing things nobody have done before
[18:55] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:1c71f185add9: avcodec/tiff_common: Improve formating of long arrays
[19:10] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:fcd19d6c8db5: avcodec/jpeg2000dec: Zero code and quant style
[19:20] <durandal_1707> Daemon404: why you allocate coded_frame when you never use it?
[20:19] <llogan> someone asked me for help. version: SVN-r15261 from 2008.
[20:20] <JEEB> better than some 2006 version I guess :)
[20:21] <llogan> maybe some of the 41582 commits since could be useful
[20:22] <durandal_1707> llogan: via mail?
[20:22] <llogan> a forum
[20:23] <llogan> s/me//
[20:24] <llogan> i'm just always surprised (and amused) when I see such graybeards still being used.
[20:37] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:6565b1ad97fe: avcodec/h264_sei: make SEI type truncated message more informative
[20:44] <durandal_1707> wtf people build ffmpeg with caca support...
[20:44] <durandal_1707> that thing was joke
[20:51] <wm4> :)
[20:54] <Compn> just think of caca as an interesting video filter
[20:54] <Compn> mplayer -vo matrixview is way more fun :P
[20:54] <Compn> http://pigeond.net/mplayer/mplayer-matrixview.html
[20:55] <wm4> when I deleted that from mplayer2, 1 person complained
[20:55] <Compn> hahaha
[20:55] <Compn> didnt know it had any users at all
[20:55] <Compn> at least we know it has one :)
[21:16] <llogan> ubitux: i'm going to use your fate ffmpeg tip: "Is it OK for me to use FFmpeg git/HEAD? Let's check how green http://fate.ffmpeg.org is."
[21:16] <ubitux> :)
[21:20] <cone-480> ffmpeg.git 03Diego Biurrun 07master:be0b4c70ec40: atrac3: Replace a silly counter variable name with plain 'j'
[21:20] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:c2df95978b19: Merge commit 'be0b4c70ec40d7f9ac8e416a4379d4a387421184'
[21:28] <cone-480> ffmpeg.git 03Maxim Poliakovski 07master:4be292dac7e7: atrac: Move doxygen comments to the header
[21:28] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:16bb8faac6d5: Merge commit '4be292dac7e7f0586053fcced8d530ccf4ebef29'
[21:31] <durandal_1707> when next release is planned?
[21:34] <cone-480> ffmpeg.git 03Diego Biurrun 07master:5ce04c14dd3d: Use correct Doxygen syntax
[21:34] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:2ece7d94bc35: Merge commit '5ce04c14dd3dd3670cbdba82275a3a72c716ec6f'
[21:34] <durandal_1707> huh someone sent mail to ffmpeg-user at mplayerhq.hu
[21:38] <ubitux> llogan: you could also twitt about VP9
[21:39] <durandal_1707> and atrac+,caca,exif,sftp,opus......
[21:53] <cone-480> ffmpeg.git 03Maxim Poliakovski 07master:79cbac8cd456: atrac3: Generalize gain compensation code
[21:54] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:4a63c69faac6: Merge remote-tracking branch 'qatar/master'
[22:00] <cone-480> ffmpeg.git 03Paul B Mahol 07master:5f99e6d632da: avfilter/vf_pullup: do not ignore frame field order
[22:13] <wm4> ubitux: I still think AVPacket is the best "container" for most subtitle data (associated with a codec id)
[22:13] <wm4> ubitux: and for bitmap sub decoder output, maybe an AVFrame
[22:14] <ubitux> huh?
[22:14] <wm4> with a an additional list of bounding rectangles
[22:14] <ubitux> aren't you mixing demuxed and decoded subtitles?
[22:14] <wm4> there's not necessarily a difference
[22:14] <wm4> your text sub "decoders" just convert from one format to another
[22:15] <wm4> maybe they should be bitstream filters or so
[22:15] <Daemon404> why wouldnt bitmap subs just be an RGBA overlay
[22:15] <Daemon404> why must they be special?
[22:16] <wm4> they could be, but then they must come with one or more bounding box
[22:16] <Daemon404> right... streaming
[22:16] <ubitux> wm4: no i think that's wrong, sorry
[22:16] <ubitux> wm4: because it breaks remuxing
[22:16] <ubitux> i see no benefit in moving the decoding into the demuxers
[22:17] <wm4> did I say that
[22:17] <ubitux> then i misunderstood your point
[22:17] <wm4> I'm just wondering: why do your text sub decoders not output ASS packets?
[22:18] <wm4> because that's what they do
[22:18] Action: ubitux is lost
[22:22] <wm4> actually, movtext to text is a bitstream filter
[22:22] <wm4> it allows both directions, too
[22:24] <ubitux> i don't know if the bitstream filter is still relevant
[22:25] <ubitux> but anyway, i'm not sure what problem you are trying to solve here
[22:44] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:b72040daad03: ffplay: avoid direct access to max_lowres use av_codec_get_max_lowres()
[22:44] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:c7fe2a3733ba: ffplay: avoid direct access to lowres use av_codec_g/set_lowres()
[22:45] <llogan> ubitux: ok
[22:46] <llogan> got a text you want? "FFmpeg can now decode VP9."
[22:48] <ubitux> "VP9 bitches!"
[22:49] <ubitux> llogan: be more enthousiastic :)
[22:49] <ubitux> "VP9 decoder in FFmpeg!"
[22:49] <ubitux> or whatever feature you want
[22:49] <ubitux> twitter is supposed to be a torrent of transient shit, don't worry too much about it imo
[22:58] <llogan> hmmm...i do like the bitches one, but i'll add the !
[22:59] <mateo`> Fast & Furious VP9 decoder NOW in FFmpeg ! (+ a maximum of flashing gif around)
[23:00] <ubitux> not that fast yet
[23:00] <ubitux> :(
[23:00] <llogan> how about a link to the commit?
[23:00] <llogan> or is that too much for twitusers?
[23:00] <ubitux> watch the comments if anyone wants it
[23:00] <mateo`> ubitux: but furious at least :D
[23:00] <llogan> VP9!!!!111~``1!!`
[23:01] <mateo`> enough furious to make the libav guys cry
[23:01] <ubitux> VP9!!111!ONE1!!ELEVEN
[00:00] --- Fri Oct  4 2013