[Ffmpeg-devel-irc] ffmpeg-devel.log.20151217

[burek]

[00:08:35 CET] <michaelni> kierank, a349a10edf84982cf1cf3b8747a6a2e7ca733083 seems breaking fate-dirac here
[00:09:49 CET] <kierank> lemme see
[00:09:55 CET] <michaelni> ffplay fate-suite//dirac/vts.profile-main.drc results in a mostly gray image here
[00:12:44 CET] <kierank> very strange
[00:12:52 CET] <kierank> since that file shouldn't be affected
[00:14:02 CET] <michaelni> no new valgrind warnings for fate-dirac
[00:21:05 CET] <kierank> ah it's a merge issue
[00:24:41 CET] <kierank> fix coming up
[00:29:37 CET] <kierank> michaelni: can you confirm patches fixes issue
[00:31:08 CET] <fritsch> https://dl.dropboxusercontent.com/u/55728161/yBsn6BY8BFIXBwWIsEyYIvxlNq3.jpg <- ffmpeg has severe issues to determine this one as jpg
[00:31:37 CET] <fritsch> when reading from memory with avformat_open_input
[00:32:08 CET] <fritsch> only thing that helps, forcing mjpeg as AVInputFormat param
[00:41:15 CET] <michaelni> kierank, seems fixing it
[00:55:40 CET] <kierank> michaelni: ok will push
[00:56:18 CET] <cone-322> ffmpeg 03Kieran Kunhya 07master:25f6ccccd690: diracdec: Fix codeblock parameters reading
[01:02:41 CET] <kierank> atomnuker: fuzzing hq now
[01:27:13 CET] <Daemon404> t
[01:30:03 CET] <TD-Linux> I run the predecessor to the Archer C7 with openwrt and it's great btw. No beamforming like some of the fancy new ones though
[02:38:23 CET] <cone-322> ffmpeg 03Michael Niedermayer 07master:6ea7dd25c773: avformat/swfenc: Check frama rate before writing it
[02:38:25 CET] <cone-322> ffmpeg 03Moritz Barsnick 07master:9d1fb9ef313e: ffmpeg: add progress speed to status line and report
[10:24:16 CET] <kierank> atomnuker: manymany crashes
[11:59:59 CET] <ubitux> dammit this mats is one hell of a toxic guy
[12:00:50 CET] <nevcairiel> his first interaction with the project was to call all developers assholes
[12:00:52 CET] <nevcairiel> so there is that
[12:01:16 CET] <nevcairiel> that some people even corresponded with him after that is beyond me
[12:40:56 CET] <ubitux> michaelni: i'm not sure how to proceed in order to test that slicing aside from a threading context
[12:41:09 CET] <ubitux> (yeah i still didn't get it)
[12:58:31 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:741b494fa8cd: h264: eliminate default_ref_list
[12:58:31 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:bc66451e5e90: Merge commit '741b494fa8cd28a7d096349bac183893c236e3f9'
[13:00:23 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:792b9c9dfcf4: h264: set frame_num in start_frame(), not decode_slice_header()
[13:00:24 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:c6f1f334cbc5: Merge commit '792b9c9dfcf44b657d7854368d975b5ca3bc22ca'
[13:09:06 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:b09ad37c8384: h264: derive the delay from the level when it's not present
[13:09:07 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:10e55bd658b0: Merge commit 'b09ad37c83841c399abb7f2503a2ab214d0c2d48'
[13:13:53 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:31c51f7441de: avpacket: add a function for wrapping existing data as side data
[13:13:54 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:30833d121e24: Merge commit '31c51f7441de07b88cfea2550245bf1f5140cb8f'
[13:18:53 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:84adab333cdd: lavc: add stream-global packet side data
[13:18:54 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:a7d5b9f1c3ca: Merge commit '84adab333cddeefc3cfd843089dee23f58bd372c'
[13:24:09 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:e63e3797a1ed: avconv: pass the global codec side data to the muxer
[13:24:10 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:be52b95d311b: Merge commit 'e63e3797a1ed9346f529848e6ba3d27fd2d2cc8d'
[13:28:48 CET] <michaelni> ubitux, a simple for () that passes a frame a slice t a time, this should be really simple (of course makes only sense for testing, nothung is gained by passing a frame that way)
[13:29:47 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:f0b769c16daa: lavc: add a packet side data type for VBV-like parameters
[13:29:48 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:dd6ee019ea82: Merge commit 'f0b769c16daafa64720dcba7fa81a9f5255e1d29'
[13:31:41 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:11c9bd633f63: libopenh264enc: export CPB props side data
[13:31:42 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:f49264a1c591: Merge commit '11c9bd633f635f07a762be1ecd672de55daf4edc'
[13:33:57 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:03afb62e8351: libvpxenc: export CPB props side data
[13:33:58 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:d6322710c5ec: Merge commit '03afb62e83516141ba999536fc97575faefb98af'
[13:35:10 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:732a37d1466d: libx264: export CPB props side data
[13:35:11 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:b77061b5cace: Merge commit '732a37d1466d45b3812509d68c82e783530e291a'
[13:35:49 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:3f5c99fcbb2c: mpegvideo_enc: export CPB props side data
[13:35:50 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:b799619f48ba: Merge commit '3f5c99fcbb2c366d7bdef8500c19f43a33bdb6b9'
[13:38:02 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:2507b5dd6748: mpegvideo_enc: export vbv_delay in side data
[13:38:04 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:31ae2308b3f6: Merge commit '2507b5dd674834be7261772996f47ae3b95cca69'
[13:42:48 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:1520c6ff05d8: nvenc: export CPB props side data
[13:42:49 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:5fc17edc7dce: Merge commit '1520c6ff05d835da4b793318fc88bbbc129c86a1'
[13:43:11 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:825900248b40: qsvenc: export CPB props side data
[13:43:12 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:b2d8b91cf0a2: Merge commit '825900248b4053515803152d3165efdb034b660b'
[13:55:23 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:5845a8273e46: movenc: use the CPB props side data
[13:55:24 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:f95385dd5806: Merge commit '5845a8273e4694e0254ad728970b82bb64fd8bc0'
[13:57:53 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:8bcadaacc2b8: mpegenc: use the CPB props side data
[13:57:54 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:079b5d4ef888: Merge commit '8bcadaacc2b8dc3c5d6569835a5ca20e62d3efca'
[14:10:28 CET] <J_Darnley> heh, a reply to a mailing list email: can you give me your email?
[14:49:08 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:e02de9df4b21: lavc: export Dirac parsing API used by the ogg demuxer as public
[14:49:09 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:efcc8fddd6b7: Merge commit 'e02de9df4b218bd6e1e927b67fd4075741545688'
[14:52:09 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:b8b21dee0f0b: mxfenc: always assume long gop
[14:52:10 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:bd88914b15e7: Merge commit 'b8b21dee0f0b4bcc1888067c497d6e838859d2cd'
[14:52:19 CET] <cone-926> ffmpeg 03Ganesh Ajjanagadde 07master:8df8f0c70c9c: movenc-test: add a missing va_end call
[14:52:20 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:fd6c5285f97f: Merge commit '8df8f0c70c9c2085d9ca186b7cbf43656c4412eb'
[14:52:29 CET] <cone-926> ffmpeg 03Ganesh Ajjanagadde 07master:d5f5c90be9ec: rtsp: free opts dictionary on failure of getnameinfo
[14:52:30 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:ea1000b9fe68: Merge commit 'd5f5c90be9ec120f1320dcdd808b7bd917dee47c'
[14:52:42 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:68e547ae8b45: avpacket: use ERANGE instead of EOVERFLOW
[14:52:43 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:74b8fa103d63: Merge commit '68e547ae8b455e5e2b60839f35c359d77a6d94bc'
[14:53:54 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:d6dc5d15af0d: aacdec: fix aac_static_table_init() prototype
[14:53:55 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:49507f1d1731: Merge commit 'd6dc5d15af0d8617611281a34a2c3f9ced149ccf'
[15:04:46 CET] <cone-926> ffmpeg 03Anton Khirnov 07master:f1ccd0768014: h264: do not call frame_start() for missing frames
[15:04:47 CET] <cone-926> ffmpeg 03Hendrik Leppkes 07master:7b880464cb31: Merge commit 'f1ccd076801444ab7f524cb13e0886faaf10fd50'
[15:05:37 CET] <kierank> Not merged, FFmpeg does a lot more in frame_start to setup missing frames
[15:05:37 CET] <kierank> as well (like coloring them), and the overhead of the other setup is
[15:05:37 CET] <kierank> minimal.
[15:05:38 CET] <kierank> <3
[15:05:49 CET] <kierank> nevcairiel <3
[15:07:12 CET] <nevcairiel> its not my fault libav removes half their features instead of fixing them, like ER
[15:07:36 CET] <kierank> I know
[15:07:49 CET] <nevcairiel> also, the ER works really quite well these days
[15:07:56 CET] <kierank> it's wonderful that you're taking the time to make sure merges from libav don't break h264
[15:07:58 CET] <kierank> highly appreciated
[15:08:08 CET] <kierank> that said I will try and fuzz over the weekend
[15:08:21 CET] <nevcairiel> apparently the other merge earlier broke something
[15:08:37 CET] <nevcairiel> but the code is weird
[15:08:47 CET] <kierank> we actually get very good reports about ffmpeg ER
[15:08:55 CET] <kierank> i.e it's shockingly good
[15:09:24 CET] <nevcairiel> yeah, I have a few samples where other decoders fail an entire slice entirely, while with ffmpeg you dont even see that something is broken
[15:09:26 CET] <nevcairiel> its fascinating
[15:09:34 CET] <nevcairiel> (other then from log output)
[15:09:47 CET] <nevcairiel> it used to be worse, but michaelni fixed it a while ago
[15:10:16 CET] <kierank> people have done packet loss shootouts vs hardware decoders and ours comes out with lower latency and with better ER
[15:49:07 CET] Action: kierank hopes ticket2371 is not the sample he thinks it is
[15:49:31 CET] <kierank> oh it's another
[16:21:57 CET] <durandal_1707> Compn: that mplayer ambisonic filter is of bad quality
[16:31:26 CET] <durandal_1707> shame that AVChannelLayout is so limited and does not support more generic stuff
[16:32:07 CET] <fritsch> durandal_1707: what is missing for you?
[16:32:22 CET] <fritsch> custom masking is possible iirc
[16:32:39 CET] <durandal_1707> cube square ...
[16:33:45 CET] <durandal_1707> eg every chlayout should have info about speaker position in space
[16:37:11 CET] <fritsch> durandal_1707: ah okay, add some extradata?
[16:37:56 CET] <durandal_1707> want to support ambisonic 3d outputs
[16:38:29 CET] <durandal_1707> and more silly speakers setup
[16:38:41 CET] <durandal_1707> 22.1
[16:39:08 CET] <kierank> object based audio too
[16:39:16 CET] <kierank> someone should implement dolby atmos
[16:39:27 CET] <kierank> it's the only spec i've ever seen that threatens you for implementing it
[16:40:10 CET] <ubitux> > Oh, and it's in zip format this time, since I got complaints from using tar.
[16:40:12 CET] <ubitux> lol
[16:53:10 CET] <durandal_1707> kierank: samples?
[16:53:26 CET] <kierank> there's one on trac
[16:53:31 CET] <kierank> from a bluray
[16:53:54 CET] <nevcairiel> does the spec contain enough info to do so?
[16:54:09 CET] <durandal_1707> it is in own codec or?
[16:55:01 CET] <nevcairiel> the bitstream guide seems to outline the bitstream syntax, but not how to decode it
[16:56:45 CET] <nevcairiel> although it doesnt seem very complex
[16:57:13 CET] <ubitux> lol durandal_1707 
[16:57:22 CET] <ubitux> (rar)
[17:14:15 CET] <cone-926> ffmpeg 03Ganesh Ajjanagadde 07master:39f8c7961dcd: lavu/random_seed: use arc4random() when available
[17:14:16 CET] <cone-926> ffmpeg 03Ganesh Ajjanagadde 07master:07a8fbaa5532: lavc/nellymoserenc: avoid wasteful pow
[17:25:21 CET] <Mavrik> Uhm, are avfilter filterchains thread safe across threads? E.g. can you write in it in one thread and read from it in another?
[17:25:45 CET] <nevcairiel> probably not
[17:26:03 CET] <Mavrik> Hrmf, that probably explains those strange crashes then :)
[17:33:42 CET] <durandal_1707> Mavrik: what's crashing?
[17:35:22 CET] <Mavrik> I'm feeding a filter chain in one thread and doing av_buffersink_get_frame on another thread and getting some strange crashes (e.g. crashes in functions that copy frame side data).
[17:35:56 CET] <Mavrik> So i was just wondering if just using a filter chain across two threads is dumb or should I look into my code more :)
[17:45:36 CET] <nevcairiel> I don't think thats supposed to work honestly
[17:47:16 CET] <durandal_1707> Mavrik: that can't work, why you need threads for what?
[17:48:54 CET] <Mavrik> Hmm, I don't really, will rewrite, thanks.
[17:51:56 CET] <cone-926> ffmpeg 03rsn8887 07master:02e9ba42eb36: avformat/xmv: Add *.xmv to the recognized extensions for the XMV format.
[17:57:29 CET] <durandal_1707> aww, missing real name, someone will sue us
[18:00:28 CET] <atomnuker> he's the noblest of all: the anonymous hero who dedicates his work to the people
[18:01:30 CET] <durandal_1707> the Elvis
[18:03:27 CET] <Daemon404> atomnuker, until one needs to find anonymous hero to ask wtf some piece of code is doing ;)
[18:18:43 CET] <durandal_1707> so anybody tried sofalizer? Asking again...
[18:29:58 CET] <durandal_1707> its time for new release?
[18:30:12 CET] <durandal_1707> 3.0?
[18:46:02 CET] <durandal_1707> nevcairiel, michaelni ^
[19:09:55 CET] <jamrial> it's been three months, yeah. but i think we should wait a bit since we're merging new apis from libav
[19:10:43 CET] <BtbN> There were some major API and ABI breaks, a bit more time won't hurt.
[19:11:18 CET] <cone-926> ffmpeg 03Michael Niedermayer 07master:ffad6f6b8962: avcodec/diracdec: fix aspect ratio (it was lost after efcc8fddd6b7d1f931ff349e195d78c3c943d7fd)
[19:11:19 CET] <cone-926> ffmpeg 03Michael Niedermayer 07master:e839db228844: avformat/oggparsedirac: Export sample aspect ratio
[19:11:20 CET] <cone-926> ffmpeg 03Michael Niedermayer 07master:73840bbe4eea: avcodec/diracdec: Check ff_set_dimensions() for failure
[19:11:21 CET] <cone-926> ffmpeg 03Michael Niedermayer 07master:c67cf84d7c65: avfilter/af_sofalizer: Fix occured typo
[19:56:32 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07master:9637c2531f7e: sonic: make sure num_taps * channels is not larger than frame_size
[21:14:53 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:76af12f5429b: ffmdec: reject zero-sized chunks
[21:14:53 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:507e0314c0d9: aaccoder: prevent crash of anmr coder
[21:14:55 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:dcd837e41c83: aacenc: update max_sfb when num_swb changes
[21:14:56 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:cb44683a8c5c: aacsbr: ensure strictly monotone time borders
[21:14:57 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:7e94ea3dd1e5: sbrdsp_fixed: assert that input values are in the valid range
[21:14:58 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:b9087aa65167: sbr_qmf_analysis: sanitize input for 32-bit imdct
[21:14:59 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:a79a5c32f924: golomb: always check for invalid UE golomb codes in get_ue_golomb
[21:15:00 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:0b24a0e0f12f: ffm: reject invalid codec_id and codec_type
[21:15:01 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:796f1a24f59b: opus_silk: fix typo causing overflow in silk_stabilize_lsf
[21:15:02 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.8:748d5fa2edfe: sonic: make sure num_taps * channels is not larger than frame_size
[21:27:28 CET] <cone-926> ffmpeg 03Michael Niedermayer 07master:95b59bfb9d9e: Revert "avcodec/aarch64/neon.S: Update neon.s for transpose_4x4H"
[21:36:36 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.7:e56d32203289: ffmdec: reject zero-sized chunks
[21:36:37 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.7:c78d268e9fa8: aaccoder: prevent crash of anmr coder
[21:36:38 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.7:435b0ab96977: golomb: always check for invalid UE golomb codes in get_ue_golomb
[21:36:39 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.7:8b30abd0201a: ffm: reject invalid codec_id and codec_type
[21:36:40 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.7:ea4b99f82cbe: opus_silk: fix typo causing overflow in silk_stabilize_lsf
[21:36:41 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.7:34b9c7612d1a: sonic: make sure num_taps * channels is not larger than frame_size
[22:04:43 CET] <cone-926> ffmpeg 03Michael Niedermayer 07release/2.8:e4a6a8553ed8: avfilter/vf_mpdecimate: Add missing emms_c()
[22:04:44 CET] <cone-926> ffmpeg 03Michael Niedermayer 07release/2.8:1dddd5371a6a: Revert "avcodec/aarch64/neon.S: Update neon.s for transpose_4x4H"
[22:06:18 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.6:e1b38b36853b: ffmdec: reject zero-sized chunks
[22:06:19 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.6:4733060ef394: aaccoder: prevent crash of anmr coder
[22:06:20 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.6:1828abb7449a: golomb: always check for invalid UE golomb codes in get_ue_golomb
[22:06:21 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.6:f7b4e4625120: ffm: reject invalid codec_id and codec_type
[22:06:23 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.6:d8c70ebd2c53: opus_silk: fix typo causing overflow in silk_stabilize_lsf
[22:06:23 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.6:ff663ae464c1: sonic: make sure num_taps * channels is not larger than frame_size
[22:06:24 CET] <cone-926> ffmpeg 03Michael Niedermayer 07release/2.6:7db1014b314a: avcodec/dirac_parser: Check that there is a previous PU before accessing it
[22:06:25 CET] <cone-926> ffmpeg 03Andreas Cadhalpun 07release/2.6:a90967013b1c: mjpegdec: consider chroma subsampling in size check
[22:06:26 CET] <cone-926> ffmpeg 03Michael Niedermayer 07release/2.6:43bd666fb4d4: avcodec/mpeg4videodec: Check available data before reading custom matrix
[23:10:41 CET] <cone-926> ffmpeg 03Ganesh Ajjanagadde 07master:97d2c2d678f6: lavc/opus_celt: replace pow by exp2
[00:00:00 CET] --- Fri Dec 18 2015