[Ffmpeg-devel-irc] ffmpeg-devel.log.20130914

[burek]

[00:00] <durandal_1707> ubitux: you stopped creating new filters?
[00:01] <ubitux> i'm not really working on new filters currently, right
[00:15] <durandal_1707> what happened with lua scripting: where certain filtering code would be available from lua?
[00:42] <cone-25> ffmpeg.git 03Michael Niedermayer 07master:108b91c3fe61: avcodec/mpeg12dec: clear mpeg_enc_ctx_allocated when contexts are deallocated
[04:00] <cone-440> ffmpeg.git 03Alexey Balekhov 07master:90ca81426736: ffserver: allow choosing the pixel format
[04:16] <cone-440> ffmpeg.git 03Piotr Bandurski 07master:8728360b5664: avcodec/mationpixels: check for malloc failures
[10:56] <cone-787> ffmpeg.git 03Luca Barbato 07release/0.10:c69b4bdf5895: avidec: Let the inner dv demuxer take care of discarding
[10:56] <cone-787> ffmpeg.git 03Luca Barbato 07release/0.10:139f352daf84: wtv: Mark attachment with a negative stream id
[10:56] <cone-787> ffmpeg.git 03Michael Niedermayer 07release/0.10:d06dda65b499: Merge commit '139f352daf84e005824562e0e0f36e06ac60ee36' into release/0.10
[11:06] <cone-787> ffmpeg.git 03Michael Niedermayer 07release/0.10:42ad4178fd2d: avio: Handle AVERROR_EOF in the same way as the return value 0
[11:06] <cone-787> ffmpeg.git 03Michael Niedermayer 07release/0.10:fe8b5a37d585: rmdec: Use the AVIOContext given as parameter in rm_read_metadata()
[11:06] <cone-787> ffmpeg.git 03Luca Barbato 07release/0.10:c6942a4b0374: vqavideo: check the version
[11:06] <cone-787> ffmpeg.git 03Luca Barbato 07release/0.10:26589aa81028: westwood_vqa: do not free extradata on error in read_header
[11:06] <cone-787> ffmpeg.git 03Michael Niedermayer 07release/0.10:3f4fb49d1b14: Merge commit '26589aa81028f42c763c5581a1486a271799890b' into release/0.10
[11:30] <cone-787> ffmpeg.git 03Luca Barbato 07release/0.10:7296ee7af142: qdm2: check and reset dithering index per channel
[11:31] <cone-787> ffmpeg.git 03Luca Barbato 07release/0.10:09a098fb8bc7: atrac3: fix error handling
[11:31] <cone-787> ffmpeg.git 03Michael Niedermayer 07release/0.10:e27bc85b31ee: Merge commit '09a098fb8bc7d70a0258b3e8d658833c8d11debe' into release/0.10
[11:36] <cone-787> ffmpeg.git 03Luca Barbato 07release/0.10:e06623c48030: atrac3: set the getbits context the right buffer_end
[11:36] <cone-787> ffmpeg.git 03Michael Niedermayer 07release/0.10:cb198f945fdc: Merge commit 'e06623c48030530b50f00cee26f1a3a33c4b4d4b' into release/0.10
[12:52] <cone-787> ffmpeg.git 03Paul B Mahol 07master:8134edf3eb1a: avcodec/cdxl: use av_freep()
[13:22] <ubitux> BBB: ping
[13:29] <durandal_1707> what are good fieldmatch arguments that works with telecine filter?
[13:35] Action: ubitux has no idea
[13:47] <durandal_1707> can build of libs be separated?, so that if one adds new filter he does not need to rebuild whole f*king tree
[14:12] <cone-787> ffmpeg.git 03Paul B Mahol 07master:7d8939a0ff71: avfilter/avf_avectorscope: improve description
[14:13] <ubitux> vp8 idct is only on 4x4 blocks?
[14:21] <JEEB> heh, seems like currently there's a single bit that is completely different between 14496-15 and DivX's HEVC-in-Matroska extradata
[14:21] <JEEB> 14496-15 specifies a reserved bit with the value of 0
[14:21] <JEEB> DivX specifies and writes a reserved bit with the value of 1
[14:25] <nevcairiel> luckily noone cares about reserved bits :P
[14:26] <JEEB> :)
[14:26] <ubitux> from a certain point of view, it's 100% different
[14:26] <JEEB> I wonder where they got the '1'b there
[14:27] <JEEB> GPAC doesn't seem to write a value of 1 there either
[14:27] <nevcairiel> its a bit odd that half the reserved bits in the spec are 0, and the other half are 1
[14:27] <JEEB> yes
[14:28] <JEEB> anyways, wrote an overview of their current spec page @ http://lists.matroska.org/pipermail/matroska-devel/2013-September/004575.html
[17:57] <cone-787> ffmpeg.git 03Piotr Bandurski 07master:bb5e7d3b5ad2: avcodec/mationpixels: free memory after malloc failures
[17:57] <cone-787> ffmpeg.git 03Piotr Bandurski 07master:1a3d142f1f1e: avcodec/bfi: check for malloc failure
[18:17] <cone-787> ffmpeg.git 03Paul B Mahol 07master:6a70feadb013: avformat/nutenc: use av_calloc()
[18:17] <cone-787> ffmpeg.git 03Paul B Mahol 07master:f2afe96181a8: avformat/nutenc: use av_reallocp_array()
[19:38] <wm4> durandal_1707: have you tried to find out yet why abuffersink always picks U8, even if there are better formats?
[19:39] <wm4> looking at the code, I don't actually see anything unusual
[19:39] <wm4> it should use the normal lavfi format negotiation
[20:15] <durandal_1707> wm4: you use abuffersink?
[20:20] <durandal_1707> i do not think its abuffersink job to pick best format
[21:15] <durandal_1707> saste: how is lavfi scripting going?
[21:31] <durandal_1707> wm4: let me try if i explicitly set outlink format
[21:36] <durandal_1707> huh, pullup leaks
[21:44] <durandal_1707> *mp=pullup
[22:07] <wm4> durandal_1707: well, you can set a list of formats on abuffersink
[22:07] <wm4> durandal_1707: which should be similar to a filter reporting supported formats
[22:08] <durandal_1707> ffmpeg have some logic othewise it would have same bug
[22:14] <wm4> *shrug*
[22:14] <wm4> anyway, U8 was picked even if it was at the end of the format list
[22:15] <wm4> and having flt as only item in the list worked too
[22:15] <wm4> also, aecho doesn't even support u8...
[22:17] <durandal_1707> but u8p
[22:17] <durandal_1707> oh not at all
[22:23] <ubitux> beastd: so, did you push the backport patches?
[22:25] <beastd> ubitux: yes, i wanted to do so
[22:29] <beastd> if you do not object i will push the patch skipping empty lines at the beginning and leave out the last one adding CR only EOL support
[22:29] <ubitux> sure just push whatever you want
[22:29] <ubitux> i gave my opinion, rest is up to you
[22:33] <beastd> ubitux: you are the author of the patches in question and you coded most of the text subtitle related code in FFmpeg, so I sure did not want to override you
[22:34] <ubitux> i don't think it's important, and i trust your decision
[22:34] <ubitux> you seem to overthink this a bit too much ;)
[22:38] <beastd> not exactly :)  it is just i did it that way and tested that version. changing stuff is more work for me edit/compile/test... dropping patches from the end of the patchset is easy.
[22:38] <beastd> but after having talked to you know i think you are ok with it. from the ML i couldn't really figure
[22:40] <beastd> Still need to check if it useful for/applies to 1.2 branch too
[23:27] <cone-787> ffmpeg.git 03Dave Rice 07master:2fcef19c3830: Adding a line to Changelog about FFV1 version 1.3 now being stable
[23:34] <crtmpserver> aloha to all!
[23:35] <crtmpserver> philosophical question: one kqueue multiple threads or each thread with its own kqueue?
[00:00] --- Sun Sep 15 2013