[Ffmpeg-devel-irc] ffmpeg-devel.log.20140413

[burek]

[00:11] <kierank> ubitux: for clarity i think
[00:12] <kierank> it's explaining in pseudocode what the instruction does
[00:12] <kierank> so it can do what it likes
[00:13] <ubitux> ok
[00:48] <cone-727> ffmpeg.git 03Katerina Barone-Adesi 07master:d4a8a86c6863: doc: Libavfilter English cleanup
[00:48] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:275155081f2e: Merge commit 'd4a8a86c6863ecea208fa2f774690b6df4b5bc71'
[01:19] <cone-727> ffmpeg.git 03Lukasz Marek 07master:c1c3c3611ef1: lavd: no default device unless explicitly set
[01:19] <cone-727> ffmpeg.git 03Lukasz Marek 07master:afc898e7ce00: lavd/fbdev_enc: implement get_device_list
[01:19] <cone-727> ffmpeg.git 03Lukasz Marek 07master:defb87eef4dd: lavd/qtkit: add device category
[01:19] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:d3284a62082b: Merge remote-tracking branch 'lukaszmluki/master'
[01:36] <cone-727> ffmpeg.git 03Justin Ruggles 07master:60c4660ba035: swscale: fix an implementation-defined unsigned-to-signed conversion
[01:36] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:28875c418886: Merge commit '60c4660ba035bbfbcc84ac34129ce40e037c70ad'
[01:41] <cone-727> ffmpeg.git 03Justin Ruggles 07master:20c38c9c18b7: swscale: fix some undefined signed left shifts
[01:41] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:57570052e244: Merge commit '20c38c9c18b75e62f6c489d9b95bea88efdf3915'
[01:46] <cone-727> ffmpeg.git 03Justin Ruggles 07master:d9a542ace1b9: swscale: Set alpha to opaque for internal palettes.
[01:46] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:69505a89f1ed: Merge commit 'd9a542ace1b9ca5c0ea6d758127d03d4124b7801'
[01:56] <cone-727> ffmpeg.git 03Jean First 07master:bf2064f046af: tiffdec: rename variables for consistency and fix variable shadowing
[01:56] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:6145b1d9c11b: Merge commit 'bf2064f046af64c59a416c814474a39b0a457569'
[02:05] <cone-727> ffmpeg.git 03Justin Ruggles 07master:58bc38a5f224: tiffdec: use correct data type for palette entries and set alpha to 0xFF
[02:05] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:b394b9013ee4: Merge commit '58bc38a5f224d29b79338200459792c765c25fd5'
[02:06] <cone-727> ffmpeg.git 03Paul B Mahol 07master:de9d2705f61e: bytestream: add functions for accessing size of buffer
[02:06] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:d06a7530f7ea: Merge commit 'de9d2705f61ef569487ec5f8974a9c7ce34ec783'
[02:13] <cone-727> ffmpeg.git 03Justin Ruggles 07master:5748faf291fe: bytestream: add bytestream2_copy_buffer() functions
[02:13] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:31876ea62006: Merge commit '5748faf291fec297ef25d81962b52b3438f54278'
[03:27] <cone-727> ffmpeg.git 03Justin Ruggles 07master:0a467a9b594d: tiffdec: use bytestream2 to simplify overread/overwrite protection
[03:27] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:94cf83ff91b2: Merge commit '0a467a9b594dd67aa96bad687d05f8845b009f18'
[03:36] <cone-727> ffmpeg.git 03Justin Ruggles 07master:fdbe18b7b054: tiffdec: use a single strip if RowsPerStrip is 0
[03:36] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:75b4e4013796: Merge commit 'fdbe18b7b0545f9b5923d25a5433e4f735719ecc'
[03:51] <cone-727> ffmpeg.git 03Justin Ruggles 07master:a9b046fb0ae5: tiff: use a better name and enum values for PhotometricInterpretation
[03:51] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:b9b2f9d218ba: Merge commit 'a9b046fb0ae5395c4d1d8a82cd0d3e354b5034ed'
[04:00] <cone-727> ffmpeg.git 03Justin Ruggles 07master:345a96c327e8: tiffdec: remove an unneeded variable
[04:00] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:a64d6d515218: Merge commit '345a96c327e8f2a8077189af9f5e8d2b3f4ad5fe'
[04:03] <cone-727> ffmpeg.git 03Jean First 07master:e75ef2b7f48b: tiffdec: support predictor type 2 with RGB48 pixel format
[04:03] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:d7644e0fdb5c: Merge commit 'e75ef2b7f48b96a9b6c8646058713899d5ea5731'
[04:13] <cone-727> ffmpeg.git 03Tim Walker 07master:c8cf461c19e8: dcadec: Do not decode the XCh extension when downmixing to stereo
[04:13] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:952614b8996e: Merge commit 'c8cf461c19e8e35df4b7364d9b90aa42f1ab4560'
[04:50] <cone-727> ffmpeg.git 03Tim Walker 07master:801c39e1e305: dcadec: Use correct channel count in stereo downmix check
[04:50] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:a676e48d394c: Merge commit '801c39e1e3058fc4ba822bfb5d8612d777111e32'
[04:50] <cone-727> ffmpeg.git 03Michael Niedermayer 07master:1dc2d4a8efd1: avcodec/tiff: remove unused variables
[05:42] <cone-727> ffmpeg.git 03Lukasz Marek 07master:88c4e5a049aa: lavd/avdevice: fix buildbreak when included to C++ code
[11:51] <wm4> this is the display matrix API which collects feedbacks and requests from
[11:51] <wm4>     our downstreams users (such as VLC, mpv and ffmpeg) and developers.
[11:51] <wm4> trolling 2k
[11:52] <ubitux> & :/
[11:58] <ubitux> wm4: any more comment on the subtitles fonts patch?
[12:01] <wm4> no
[12:03] <ubitux> ok thx
[12:22] <Compn> wm4 : except i havent seen anyone on ffmpeg being upset by the downstream troll. its ineffective
[12:22] <nevcairiel> they keep trying anyway
[12:23] <bencoh> :/
[12:23] <Compn> i think the qatar thing annoys a few people :d
[13:08] <cone-719> ffmpeg.git 03Facundo Gaich 07master:7e6b3ad6930c: lavfi/subtitles: load attached fonts to libass.
[13:23] <pross-au> channel topic suggestion. *2.2.1* has been released
[13:23] <ubitux> you are +o
[13:25] <pross-au> thanks!
[13:43] <ubitux> wm4: still looking forward utf16 support btw :)
[13:43] <wm4> oh yeah I should do that
[14:13] <cone-719> ffmpeg.git 03Anton Khirnov 07master:f7c5fd81516f: resample: implement flushing
[14:13] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:cbfa71d689a1: Merge commit 'f7c5fd81516fb7b186c64c66dc080e1139586226'
[14:29] <cone-719> ffmpeg.git 03Anton Khirnov 07master:13164e479f1c: FATE: add tests for more resampling modes
[14:29] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:c955525f724d: Merge commit '13164e479f1c87829096c19003d2cc658ccc1fd4'
[14:34] <cone-719> ffmpeg.git 03Anton Khirnov 07master:ac976ed91e32: lavr: allocate the resampling buffer with a positive size
[14:34] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:c11aa9d29aae: Merge commit 'ac976ed91e323754e9a84509873ebdb437372797'
[14:46] <cone-719> ffmpeg.git 03Alessandro Ghedini 07master:aaab192df24a: af_volume: implement replaygain clipping prevention
[14:46] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:878f8b0d26e9: Merge commit 'aaab192df24a90f4450285cfb73b395cf495b462'
[14:48] <cone-719> ffmpeg.git 03James Almer 07master:0f524b6c6909: x86/synth_filter: remove the fma3 version ifdefs
[14:48] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:b6c1e80bf4a9: Merge commit '0f524b6c690962d7ebbd59ded74288e1e9b0f009'
[14:54] <cone-719> ffmpeg.git 03Peter Ross 07master:d95c9679f727: avformat/iff: remove unused code
[14:54] <cone-719> ffmpeg.git 03Peter Ross 07master:e311b7446a8c: avformat/iff: indent
[15:34] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:ca6dd53a734d: ffprobe: do not lose non ascii characters in non utf8 strings
[16:56] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:a31547ce2e5c: avutil/avstring: do not lose ascii characters when decoding non utf-8 with av_utf8_decode()
[18:13] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:289f02f923ac: avformat/utils: Inject global side data into first packet.
[18:13] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:c0d62d999d0e: avformat/utils: factorize if()
[18:14] <wm4> michaelni: uh
[18:14] <wm4> michaelni: you just broke the api
[18:15] <michaelni> wm4, ?
[18:15] <michaelni> please explain
[18:15] <wm4> it was decided that if side data remains constant over the whole file, it should be added to AVStream only
[18:15] <wm4> and avplay AFAIK has something special to copy the AVStream side data to the first packet
[18:17] <michaelni> "it was decided" <-- dont remember the mail on ffmpeg-devel
[18:17] <wm4> this wasn't discussed on ffmpeg-devel at all
[18:18] <michaelni> ok then it hanst been decided
[18:18] <michaelni> ffmpeg API must be discussed on ffmpeg-devel
[18:18] <wm4> ah it wasn't added to avplay: https://git.libav.org/?p=libav.git;a=commit;h=59444c76e6d43529a12dbd80b6dd29c6ba4079a9
[18:18] <wm4> just avconv
[18:19] <michaelni> yes and ffmpeg as i merged that
[18:19] <michaelni> and iam sure not fond of that copying
[18:19] <michaelni> but IMHO as every application that uses avfilter pretty much needs it
[18:19] <michaelni> it would be better in avformat
[18:19] <michaelni> instead of in every app
[18:20] <wm4> you assume they even pass through side data
[18:20] <michaelni> i guess i shouldnt assume that but some filters need it
[18:20] <michaelni> like af_colume
[18:20] <michaelni> af_Volume
[18:21] <michaelni> also iam happy to revert if someone has a better idea
[18:21] <wm4> the reason it wasn't added automatically to the first packet is because it's not really clear what the first packet is
[18:21] <michaelni> its also not clear in ffmpeg
[18:21] <wm4> the first packet that is demuxed? after each seek? what if the application internally disables and then enables the stream again?
[18:22] <michaelni> what if the output from ffmpeg is broacast and the user starts viewing in the middle
[18:22] <wm4> and application could throw away the lavfi filter graph when disabling the stream, and build it again when reenabing the stream
[18:22] <wm4> same on seek (even ffplay recreates the lavfi graph on seek)
[18:23] <wm4> so the only reasonable way to auto-add it is probably adding it to _every_ packet
[18:23] <wm4> but that seems wasteful
[18:30] <michaelni> i agree but i think the copying itself is the problem and its better to copy in avformat until some other solution exists than in every app
[18:30] <michaelni> ill fix the seek case in a moment
[18:32] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:82796aacab1b: avformat/utils: reinject global side data on seeks
[18:34] <michaelni> the copying i avformat also should be enough for many apps, so only complex apps (which for example enable streams in the middle) would need to copy
[18:37] <wm4> also, the idea was that some side data types happen on _change_
[18:37] <wm4> so your changes still break the api
[18:42] <michaelni> is that documented and if so where ?
[18:44] <michaelni> also i was just trying to make the life of application devels easier by factorizing code that apps would need into avformat
[18:44] <michaelni> if app devels dont want that i can revert
[18:46] <wm4> michaelni: it's documented in the AVStream.side_data doxygen
[18:47] <wm4> it explicitly says that side data listed in AVStream never appears in the AVPackets
[18:47] <wm4> I can see that automatic injection might be helpful
[18:47] <wm4> but then it should be an option that is off by default
[19:36] <michaelni> wm4, posted patch that implements that
[21:29] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:eccec203978e: avformat/mp3enc: drop redundant and uninitialized variable
[23:45] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:5ec81929497c: avcodec/fft-test: use av_malloc_array()
[23:45] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:d42a61922be7: avcodec/psymodel: use av_malloc(z)_array()
[23:45] <cone-719> ffmpeg.git 03Michael Niedermayer 07master:2ebacb0fc444: avformat/network: use av_malloc_array()
[00:00] --- Mon Apr 14 2014