[Ffmpeg-devel-irc] ffmpeg-devel.log.20111117

[burek]

[00:01] <ubitux> i wonder if the private fields comments is really worth since the struct is opaque
[00:03] <michaelni> documenting private fields is usefull for people that want to work on swr
[00:03] <ubitux> i meant the "/* private fields */" comment sorry
[00:04] <ubitux> in patch 7 (previously "//below are private")
[00:06] <michaelni> i see, well, indeed iam not sure if its better to keep or drop that comment
[00:38] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rbfca35114a 10ffmpeg/libavcodec/h264.c: 
[00:38] <CIA-41> ffmpeg: h264: rewrite has_b_frame calculation code,
[00:38] <CIA-41> ffmpeg: the previous implementation was too buggy.
[00:38] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:38] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r18b7f40007 10ffmpeg/libavcodec/h264.c: 
[00:38] <CIA-41> ffmpeg: h264: reset last_pocs on IDRs
[00:38] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:53] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * red83f84891 10ffmpeg/libavcodec/h264.c: 
[01:53] <CIA-41> ffmpeg: h264: minor simplification
[01:53] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:53] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r4f7ad4c3e4 10ffmpeg/libavcodec/h264.c: 
[01:53] <CIA-41> ffmpeg: h264: migrate poc gap based has_b_frames calculation into new has_b_frame calculation code.
[01:53] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:01] <CIA-41> ffmpeg: 03Nathan Caldwell 07master * r39616fc307 10ffmpeg/libavcodec/lagarith.c: 
[03:01] <CIA-41> ffmpeg: lagarith: Add correct line prediction for RGB
[03:01] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[03:01] <CIA-41> ffmpeg: 03Kostya Shishkov 07master * rffc638c283 10ffmpeg/libavcodec/lagarith.c: 
[03:01] <CIA-41> ffmpeg: lagarith: add some RGBA decoding support
[03:01] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[03:01] <CIA-41> ffmpeg: 03Mike Melanson 07master * r747283a078 10ffmpeg/libavcodec/cinepak.c: 
[03:01] <CIA-41> ffmpeg: cinepak: Add another special case so that it can handle the following file:
[03:01] <CIA-41> ffmpeg: http://samples.mplayerhq.hu/V-codecs/CVID/bad_cinepak_frame_size.mov
[03:01] <CIA-41> ffmpeg: This fix works around another work around which handles a different type
[03:01] <CIA-41> ffmpeg: of odd Cinepak data.
[03:01] <CIA-41> ffmpeg: Thanks to Matthew Hoops (clone2727 - gmail.com) for the sample and fix.
[03:01] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[03:01] <CIA-41> ffmpeg: 03Martin Storsjö 07master * r4a83541650 10ffmpeg/libavformat/utils.c: 
[03:01] <CIA-41> ffmpeg: avformat: Interpret times in ff_iso8601_to_unix_time as UTC
[03:01] <CIA-41> ffmpeg: This function is used in muxers for parsing the 'creation_time'
[03:01] <CIA-41> ffmpeg: metadata key, for converting it to a time value.
[03:01] <CIA-41> ffmpeg: This makes it match the behaviour of the exported 'creation_time'
[03:01] <CIA-41> ffmpeg: metadata from demuxers, where it is in UTC, too.
[03:02] <CIA-41> ffmpeg: mpegaudiodec: Don't use a nonexistent log context for av_dlog
[03:02] <CIA-41> ffmpeg: This fixes builds with DEBUG defined, broken since
[03:02] <CIA-41> ffmpeg: 1158745a2d12966c58e83b66243a42f20190812e.
[03:02] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[03:02] <CIA-41> (14 lines omitted)
[03:19] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r1d23e5246c 10ffmpeg/libavformat/rawdec.c: 
[03:19] <CIA-41> ffmpeg: mjpeg: support mpo
[03:19] <CIA-41> ffmpeg: Fixes stereoscopic_photo.mpo
[03:19] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:59] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r3e971fc6b5 10ffmpeg/libavcodec/h264.c: 
[05:59] <CIA-41> ffmpeg: h264: fix cur_chroma_format_idc with slice multithreading.
[05:59] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * r3a5fc38574 10ffmpeg/libswresample/audioconvert.h: swr: fix #endif comment ref.
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * rfc6351d019 10ffmpeg/libswresample/ (4 files): 
[10:21] <CIA-41> ffmpeg: swr: make audio convert code explicitely private.
[10:21] <CIA-41> ffmpeg: Only what's declared in libswresample/swresample.h is public.
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * r64db5a1a57 10ffmpeg/libswresample/ (resample.c swresample_internal.h): 
[10:21] <CIA-41> ffmpeg: swr: drop 'AV' prefix from ResampleContext.
[10:21] <CIA-41> ffmpeg: This type/struct is not part of the public API.
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * rc4deb90cfa 10ffmpeg/libswresample/ (rematrix.c resample.c swresample.c swresample_internal.h): swr: use "swri_" prefix instead of "swr_" for the private API.
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * r645c61fb2b 10ffmpeg/ (5 files in 2 dirs): swr: rename swr_alloc2() to swr_alloc_set_opts().
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * r4c0bad513a 10ffmpeg/libswresample/swresample_internal.h: swr: start doxycommenting the API.
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * r809631af1a 10ffmpeg/libswresample/swresample.h: swr: bump minor version.
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * r7cdfce4f63 10ffmpeg/libswresample/swresample_test.c: swr: fix "control reaches end of non-void function" compiler warning.
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * ra768b9d825 10ffmpeg/libswresample/swresample.c: swr: make swr_* functions match the prototypes.
[10:21] <CIA-41> ffmpeg: 03Clément BSsch 07master * raf1eb298c1 10ffmpeg/libswresample/ (resample.c swresample_internal.h): swr: use int16_t instead of short for resampling.
[11:00] <ubitux> michaelni: ATM, swr_alloc() seems to require a later swr_alloc_set_opts() only for one thing: the internal sample format (can easily be changed with the default value in AVOption)
[11:00] <ubitux> do you want to keep that dependency?
[11:07] <ubitux> mmh my bad they can actually set it manually
[11:20] <nevcairiel> Speaking about swr, are there any plans to let the application take full control over the matrix? As it is now its ok for downmixing, but any other tasks seem not supported. (unless I missed it.)
[11:24] <ubitux> upmixing seems to work too
[11:24] <ubitux> you can also use swresample to re-order the channels
[11:24] <ubitux> if you want to manually set the mixing level, the af_pan filter is currently under review
[11:26] <nevcairiel> Reading rematrix.c it seems to only map channels around that are missing in the output layout.
[11:29] <nevcairiel> It already has all the code, you just don't have enough control over the matrix
[11:31] <ubitux> i guess we could add a swr_set_matrixes()
[11:32] <ubitux> that makes me wonder if a swr_set_channel_mapping() won't be more appropriate than a parameters to swr_alloc_set_opts()
[14:07] <michaelni> ubitux, nevcairiel swr_set_matrix() / swr_set_mapping() sound like a good idea
[14:10] <ubitux> nevcairiel: i won't do the matrixes user control, so if you want to send a patch introducing swr_set_matrix() go ahead
[14:11] <ubitux> i'll then update the channel mapping for consistency
[14:12] <ubitux> michaelni: http://pastie.org/2877795  any reason not to do this?
[14:13] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * rd50122eafa 10ffmpeg/libavformat/mov.c: 
[14:13] <CIA-41> ffmpeg: Do not try to read total disc or track number if data atom is too short.
[14:13] <CIA-41> ffmpeg: Fixes ticket #659.
[14:21] <michaelni> ubitux change looks ok, code before looks ok too :)
[14:22] <ubitux> i think it's better to show how to set that value in case user want to avoid calling swr_alloc_set_opts() but just swr_alloc()
[14:28] <michaelni> yes
[14:43] <ubitux> mmh you meant in/out ch count could be done as well too? indeed&
[15:08] <nevcairiel> ubitux: ok, I'll come up with some patch for review, not sure when I'll get around to it though.
[15:09] <ubitux> i locally made the swr_set_channel_mapping() one
[15:10] <ubitux> when my pending swr work is approved i'll send the next related patch set :)
[15:15] <ubitux> nevcairiel: also not you might need to introduce a SWR_FLAG_REMATRIX to force rematrix
[15:16] <ubitux> note*
[16:31] <CIA-41> ffmpeg: 03Michael Bradshaw 07master * r31ec694771 10ffmpeg/libavcodec/ (Makefile libopenjpeg.c): 
[16:31] <CIA-41> ffmpeg: libopenjpeg: rename decoder source file.
[16:31] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:58] <CIA-41> ffmpeg: 03Clément BSsch 07master * racd56efbeb 10ffmpeg/libswresample/swresample.c: swr: correctly raise rematrix AVERROR.
[17:58] <CIA-41> ffmpeg: 03Clément BSsch 07master * r68d3144e89 10ffmpeg/libswresample/swresample.h: swr: doxycomment public API.
[17:58] <CIA-41> ffmpeg: 03Clément BSsch 07master * rf03afd5d54 10ffmpeg/libswresample/swresample_internal.h: swr: surround mixing is obviously used, fix comment.
[17:58] <CIA-41> ffmpeg: 03Clément BSsch 07master * r1dbd433f0d 10ffmpeg/libswresample/swresample.c: swr: use av_opt_set_int() instead of manual assignments in swr_alloc_set_opts().
[18:14] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r5a2b3f3a52 10ffmpeg/libavcodec/aacdec.c: 
[18:14] <CIA-41> ffmpeg: aacdec: Avoid unneeded reinit due to ADTS headers
[18:14] <CIA-41> ffmpeg: Fixes Ticket 662
[18:14] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:32] <CIA-41> ffmpeg: 03Tim Nicholson 07master * rd3baced24f 10ffmpeg/doc/ffmpeg.texi: doc: reword and clarify default stream selection.
[18:32] <CIA-41> ffmpeg: 03Clément BSsch 07master * r0bfc03df3a 10ffmpeg/libswresample/swresample.c: swr: move code to avoid resample forward declaration.
[18:32] <CIA-41> ffmpeg: 03Clément BSsch 07master * reafa2b6005 10ffmpeg/ (5 files in 2 dirs): swr: introduce a public function to customize the channel mapping.
[20:33] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r30ca700ba1 10ffmpeg/libavformat/ac3dec.c: 
[20:33] <CIA-41> ffmpeg: ac3probe: Detect Sonic Foundry Soft Encode AC3 as raw AC3.
[20:33] <CIA-41> ffmpeg: Our ac3 code chain can handle it fine.
[20:33] <CIA-41> ffmpeg: More ideal would be to write a demuxer that actually extracts what can be from the additional
[20:33] <CIA-41> ffmpeg: headers and uses it for whatever it can be used for.
[20:33] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:33] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r4fef94c6d2 10ffmpeg/libswresample/swresample_internal.h: 
[20:33] <CIA-41> ffmpeg: swr: doxument matrix fields.
[20:33] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:35] <zayac> Does anyone know where mpeg format full description could be found?
[20:36] <ubitux> michaelni: thx for 4fef94c6
[20:37] <ubitux> zayac: you mean the official specs?
[20:38] <ubitux> look for is138181.pdf and is138182.pdf on your favorite search engine
[20:38] <ubitux> you are supposed to pay for it though
[20:56] <zayac> isn;t mpeg an open format? Why there is lack of description available for public?
[20:56] <nevcairiel> its anything but a open format really. :p
[21:02] <Daemon404> uh
[21:02] <Daemon404> "mpeg" isnt sufficiently descriptive enough to tell us what format you indeed mean
[21:02] <Daemon404> MPEG are a group of people
[21:03] <nevcairiel> I was assuming he meant PS or TS
[21:04] <Daemon404> well, when some people say "mpeg format" they dont even mean the file format, but the mpeg-2 codec
[21:04] <ubitux> Daemon404: are you ok with libutvideo in K&R?
[21:04] <Daemon404> ubitux, yes, but you cant use named struct init
[21:04] <Daemon404> thats the only bit
[21:04] <ubitux> sure i drop that patch
[21:04] <ubitux> i was refering to patch 2/3
[21:05] <zayac> ok. I am looking for a description of mpeg-1 codec format, as I'm trying to do my own stuff with it in ffmpeg
[21:05] <Daemon404> ubitux, well patch 2 was k&r, and 3 was the named struct init
[21:05] <Daemon404> i think 1 was actually vba
[21:05] <Daemon404> iirx
[21:05] <Daemon404> iirc*
[21:06] <zayac> a code itself is rather descriptive, however I don't mind reading some specification papers
[21:06] <Daemon404> er, vda.
[21:06] <CIA-41> ffmpeg: 03Clément BSsch 07master * r3495817468 10ffmpeg/libavcodec/libutvideo.cpp: libutvideo: use K&R style.
[21:07] <Daemon404> as a side note
[21:07] <Daemon404> K&R switch statements... i hate them
[21:07] <ubitux> Daemon404: yes, but you didn't say anything about patch2
[21:07] <ubitux> Daemon404: you mean the non-extra indent?
[21:07] <Daemon404> yes.
[21:07] <Daemon404> i think it's ugly a uninformtive
[21:07] <Daemon404> :P
[21:07] <Daemon404> and*
[21:08] <Daemon404> hmm i have a reply sitting in my drafts with a 2 am timestamp
[21:08] <Daemon404> derp
[21:08] <ubitux> :)
[21:08] <ubitux> anyway, as you just see i pushed it
[21:08] <Daemon404> yes
[21:09] Action: Daemon404 is sort of busy with :life: atm
[21:09] <Daemon404> libutvideoenc is done tho, and waiting on upstream
[21:09] <Daemon404> so is part of ylc.c
[21:09] <ubitux> Daemon404: btw, the av_malloc and UtVideoContext casts are not needed (even in c++ afaik)
[21:10] <Daemon404> hmm?
[21:10] <Daemon404> oh teh casts
[21:10] <Daemon404> i got gcc warnings i believe
[21:10] <ubitux> seriously?
[21:10] <Daemon404> about void * and UtVideoContext mismatches
[21:10] <Daemon404> *
[21:11] <Daemon404> well, g++.
[21:56] <Compn> 3GP_Converter051103 www.nurs.or.jp/~calcium FFmpeg2.40b((49<<16)+(1<<8)+0)
[21:56] <Compn> remind me to look that up
[21:58] <JEEB> woah, mobilehackerz' thing :3
[21:58] <JEEB> I used to use that thing to script some stuff (the main application, not the actual tools it came with that the default scripts used)
[22:02] <Compn> oh that one
[22:02] <Compn> must be popular 
[22:03] <JEEB> it probably still is a popular frontend to make relatively-automated encodes for mobile stuff such as phones etc.
[22:03] <JEEB> their wiki is the best knowledgebase for Japanese devices so far
[22:03] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r5bccb2d096 10ffmpeg/libswscale/utils.c: 
[22:03] <CIA-41> ffmpeg: sws: handle jpeg pixel formats in sws_init_context()
[22:03] <CIA-41> ffmpeg: Fixes Ticket576
[22:03] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:03] <JEEB> http://mobilehackerz.jp/contents/3GPConv
[23:21] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r32aa9590ac 10ffmpeg/libavfilter/vf_scale.c: 
[23:21] <CIA-41> ffmpeg: vf_scale: Fix slice position assert to take vertical subsampling into account.
[23:21] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:00] --- Fri Nov 18 2011