[Ffmpeg-devel-irc] ffmpeg-devel.log.20160613

burek burek021 at gmail.com
Tue Jun 14 02:05:03 CEST 2016

[00:25:38 CEST] <ubitux> Illya: i pushed the patch :p
[00:30:40 CEST] <Illya> ubitux: you're Clement Boesch, right?
[02:27:04 CEST] <cone-651> ffmpeg 03Michael Niedermayer 07master:4a8ed314fb38: avcodec/h264_slice: Add casts to suppress warnings
[02:27:04 CEST] <cone-651> ffmpeg 03Ilya87 07master:be37a6697140: Timeline function for the eq filter
[05:50:26 CEST] <cone-034> ffmpeg 03Matt Oliver 07master:37787f261639: lavf/os_support.h: Fix for unicode filenames on windows.
[05:50:26 CEST] <cone-034> ffmpeg 03Matt Oliver 07master:5ca44ebd9993: lavu/intmath.h: fix compilation with msvc10.
[08:01:20 CEST] <cone-034> ffmpeg 03Muhammad Faiz 07master:b8c6e5a66183: swresample: add exact_rational option
[08:01:21 CEST] <cone-034> ffmpeg 03Muhammad Faiz 07master:241e25052002: fate: add libswresample exact_rational test
[08:18:00 CEST] <cone-034> ffmpeg 03Muhammad Faiz 07master:0b8dcb3349a2: doc/resampler: add exact_rational option
[08:29:04 CEST] <ubitux> Illya: yes
[10:01:03 CEST] <ubitux> nevcairiel: 22:57 <maru> apparenlt ca2f19b9cc37b breaks videtoolbox support
[10:01:08 CEST] <ubitux> does this makes sense to you?
[10:02:08 CEST] <nevcairiel> no idea, the vt code is crap, and if it relies on internals of the h264 decoder, it deserves to break
[10:02:32 CEST] <ubitux> i see 617cd45ddc9b0fed4e3b4bcb8a51b37fc8565ee8
[10:02:39 CEST] <ubitux> maybe sth would be needed for vt as well
[10:03:07 CEST] <ubitux> but i don't see any similar pattern
[10:07:34 CEST] <nevcairiel> i dont know how the vt code works, it looks unlike any other hwaccel, and i have no means to test any of it, so delegate to someone that does :)
[10:19:46 CEST] <nevcairiel> its probably related to what buffers it gives to hwaccel->start_frame, vt is the only thing that uses those parameters, everything else uses the slice c allback
[10:21:54 CEST] <nevcairiel> although i'm not sure that data changed...
[10:22:12 CEST] <j-b> 'morning
[11:23:02 CEST] <cone-034> ffmpeg 03Muhammad Faiz 07master:66023612955d: avfilter/graphparser: remove '\n' from parse_filter
[11:29:22 CEST] <ubitux> btw, i'm merging the next one (sei h264) today
[11:29:35 CEST] <ubitux> but i probably won't be done at the end of the day
[11:30:05 CEST] <nevcairiel> from a quick look any that remain probably shouldnt be as bad as the ps one
[11:31:23 CEST] <ubitux> we'll see
[11:40:10 CEST] <ubitux> i have to make 8710a634a fit in though
[12:10:45 CEST] <nevcairiel> thats just one more type of SEI info to move into the new struct
[12:36:02 CEST] <ubitux> for each [VOTE] we probably need to remind who can vote
[12:36:30 CEST] <ubitux> and all the people not part of that commitee should probably just not be part of the debate at all
[12:55:44 CEST] <nevcairiel> i dont even know who that latest guy is that commented
[12:56:47 CEST] <kierank> ffmpeg makes north korea look democratic
[13:10:20 CEST] <cone-034> ffmpeg 03Michael Niedermayer 07master:60de31e98c16: avformat/utils: Export coded dimensions unconditionally
[13:31:12 CEST] <ubitux> well dprk has "democratic" in its name
[13:37:47 CEST] <BBB> michaelni: ok, nicolas seems to have broken the 4-4 tie, are you ok with the result?
[14:35:34 CEST] <michaelni> BBB, absolutely yes
[15:00:51 CEST] <ubitux> meh, nice mess with fpa
[15:17:06 CEST] <ubitux> q: in ffmpeg we have frame_packing_arrangement_cancel_flag (following the specs), and in libav they use "present" (semantic being !frame_packing_arrangement_cancel_flag)
[15:17:16 CEST] <ubitux> what do you guys prefer?
[15:17:53 CEST] <nevcairiel> keep ours and just move it to the sei struct, imho
[15:18:23 CEST] <ubitux> actually we also have a sei_frame_packing_present
[15:18:25 CEST] <ubitux> meh.
[15:18:47 CEST] <ubitux> ok that was introduced because of a merge
[15:18:56 CEST] <nevcairiel> is it actually read or set anywhere?
[15:18:58 CEST] <nevcairiel> if not just drop it
[15:19:30 CEST] <ubitux> libav moved from sei_frame_packing_present to frame_packing.present
[15:19:50 CEST] <ubitux> we have both sei_frame_packing_present and frame_packing_arrangement_cancel_flag
[15:21:01 CEST] <ubitux> well i'll pick a random choice in my next whim
[15:42:08 CEST] <Illya> ubitux: ok :) Regarding the libopenmpt patch, I've updated it to include your suggestions http://sprunge.us/BYLN except for the pkt->size one as the ret then becomes the number of frames, and I'm not sure the sizes would be consistent afterwards (also because the terminalogy of frames vs samples confuses me), and AFAIK there is no way to know the channel layouts. I have also included something to change log level
[15:42:08 CEST] <Illya>  somewhat based on the log message
[15:46:54 CEST] <ubitux> Illya: just resubmit on the ml (and better answer here forthe MOD exclusive aspect)
[15:47:29 CEST] <ubitux> michaelni: can you wait for the end of the h264 batch merge before the release?
[15:47:45 CEST] <ubitux> since i don't like having a release with half of it
[16:00:57 CEST] <michaelni> ubitux, sure
[16:01:12 CEST] <ubitux> thanks
[17:01:22 CEST] <ubitux> http://sprunge.us/VVPc 
[17:01:26 CEST] <ubitux> this is broken
[17:01:28 CEST] <ubitux> right?
[17:02:02 CEST] <ubitux> if (green_metadata_type==0) { if (green_metadata_type==2) { ... } }
[17:05:42 CEST] <ubitux> i suppose the second if should be sei_green_metadata.period_type
[17:11:01 CEST] <nevcairiel> this green shit was always rather flimsy
[17:11:04 CEST] <ubitux> ffs all this shit is just for logging
[17:11:09 CEST] <nevcairiel> just migrate as is
[17:11:26 CEST] <ubitux> logging using avctx->debug which i do not have access to anymore
[17:12:07 CEST] <ubitux> that's some damn monkey code there
[17:12:38 CEST] <ubitux> pretty sure the guy is parsing the ffmpeg output
[17:17:30 CEST] <nevcairiel> there is actually a new debug flag for this thing?
[17:39:28 CEST] <ubitux> yes
[17:39:30 CEST] <ubitux> "green_metadata"
[17:41:35 CEST] <nevcairiel> should purge this =p
[17:42:59 CEST] <ubitux> it's awful
[18:20:22 CEST] <cone-356> ffmpeg 03Thomas Volkert 07master:668fb1cb6a4e: rtpenc: packetizer for VP9 RTP payload format (draft v2)
[18:38:25 CEST] <DSM_> durandal_170: minterpolate works now. obmc
[18:39:55 CEST] <DSM_> using MVs exported by mestimate filter through frame sidedata
[18:40:43 CEST] <cone-356> ffmpeg 03Thomas Volkert 07master:1f872332c47f: rtpdec_vc2hq: avoid magic number
[18:42:00 CEST] <durandal_170> DSM_: how many frames it produces?
[18:43:32 CEST] <DSM_> durandal_170: depends on fps we set
[18:44:17 CEST] <DSM_> i usually test with x2 the orig fps
[19:18:45 CEST] <durandal_170> DSM_: what algo in mestimate?
[19:32:37 CEST] <DSM_> durandal_170: ebma in mestimate
[20:29:27 CEST] <cone-356> ffmpeg 03Michael Niedermayer 07master:4879841d1529: avformat/dump: Show coded dimensions again
[21:23:59 CEST] <cone-356> ffmpeg 03Thomas Bernard 07master:836f3555a8de: avformat/au: Read MetaData from AU Sun audio file header
[22:12:22 CEST] <cone-356> ffmpeg 03Jérôme Martinez 07master:878c3a36451e: avcodec/ffv1dec: fix some unsupported pix_fmt
[22:41:41 CEST] <cone-356> ffmpeg 03Clément BSsch 07master:7a2a8b541afa: lavc/h264_sei: fix green metadata typo
[22:45:01 CEST] <cone-356> ffmpeg 03Marton Balint 07master:4bee44007a93: avformat/mxfdec: check if source_package is NULL
[22:45:02 CEST] <cone-356> ffmpeg 03Marton Balint 07master:301ead131381: avformat/mux: call deinit if write_header fails
[22:45:03 CEST] <cone-356> ffmpeg 03Marton Balint 07master:9da27fb57958: avformat/mux: factorize header writing code
[22:53:35 CEST] <rcombs> QTKit is removed in macOS 10.12
[22:54:06 CEST] <rcombs> or at least, large swaths of it are
[00:00:01 CEST] --- Tue Jun 14 2016

More information about the Ffmpeg-devel-irc mailing list