[Ffmpeg-devel-irc] ffmpeg-devel.log.20140907

[burek]

[00:48] <cone-784> ffmpeg.git 03Michael Niedermayer 07master:c2430304dfb3: avformat/swfdec: Do not change the pixel format
[02:06] <cone-784> ffmpeg.git 03Michael Niedermayer 07master:033a5334badd: avcodec/h264: Allow partial escaping
[03:14] <pross-au> rcombs: ... and limit ourselves to ~3 platforms!
[03:14] <pross-au> oops. up and enter
[03:14] <rcombs> pross-au: Great Plan"!
[03:15] <pross-au> still is
[04:19] <cone-784> ffmpeg.git 03Michael Niedermayer 07master:ab84effdeda9: ffmpeg: Copy extradata if it has been initialized later from the encoder
[11:32] <cone-540> ffmpeg.git 03Reimar Döffinger 07master:2231d5b67185: libswresample: Avoid needlessly large on-stack array.
[11:32] <cone-540> ffmpeg.git 03Reimar Döffinger 07master:284123d7fd8e: Remove pointless if.
[11:32] <cone-540> ffmpeg.git 03Reimar Döffinger 07master:2c5c37ade115: libswresample: move condition to start of loop.
[13:19] <cone-540> ffmpeg.git 03Stefano Sabatini 07master:4f5493fe2380: lavf/segment: add segment_format_options option
[13:19] <cone-540> ffmpeg.git 03Stefano Sabatini 07master:de1f0c94cdcb: doc/muxers/segment: reword text for examples, improve consistency
[13:32] <cone-540> ffmpeg.git 03Michael Niedermayer 07master:94c61cd9aede: avcodec/mpegvideo: free_context_frame() cannot fail, make it return void
[13:32] <cone-540> ffmpeg.git 03Michael Niedermayer 07master:2762323c3751: avcodec/mpegvideo: Use "goto fail" for all error paths in ff_mpv_common_frame_size_change()
[13:33] <cone-540> ffmpeg.git 03Michael Niedermayer 07master:0d0f7f0ba43f: avcodec/mpegvideo: check that the context is initialized in ff_mpv_common_frame_size_change()
[14:13] <ubitux> michaelni: re: 2762323c37511fbbc98b164c07620b9ebc59ec68, shouldn't you set err?
[14:13] <ubitux> like, && (err=av_image_check_size(...)) { ...
[14:31] <cone-540> ffmpeg.git 03Michael Niedermayer 07master:cfce6f7efd28: avcodec/mpegvideo: Set err on failure in ff_mpv_common_frame_size_change()
[14:37] <ubitux> michaelni: thx :)
[17:10] <cone-540> ffmpeg.git 03Michael Niedermayer 07master:7c1835c52a4b: avformat/m4vdec: Check for non startcode 00 00 00 sequences in probe
[17:30] <J_Darnley> WTF is goin on here?  How is gcc linking this?  There is no *_mmx2 function.
[17:39] <michaelni> J_Darnley, linking what ? what issue do you speak of ? fate looks fine
[17:39] <J_Darnley> No sorry
[17:40] <J_Darnley> My own code, I guess this was the wrong window
[18:35] <cone-540> ffmpeg.git 03Andreas Cadhalpun 07master:a5b20bf4b2f7: doc/filters.texi: improve the wording of the expr_int_format documentation
[19:37] <wm4> ubitux: so how shall we fabricate a utf-16 sample
[19:44] <ubitux> take an existing one
[19:44] <ubitux> typically the one from #3496
[20:03] <wm4> why does ffmpeg still support OS/2? http://odin.netlabs.org/en/site/index.xml
[20:03] <nevcairiel> because some weirdo sends patches when it breaks
[22:10] <ubitux> arg we still have that srt code
[22:10] <ubitux> we should really drop it at some point
[22:11] <Daemon404> i wonder what percentage of the bug tracker could be condensed into "i have a sample where analyseduration is too small by default. FIX FIX FIXFIX!!!!11on"
[22:13] <wm4> ubitux: what srt code?
[22:13] <ubitux> wm4: all the code paths handling the old timestamped packets
[22:13] <wm4> Daemon404: I don't get why mov needs that at all
[22:14] <ubitux> wm4: http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavformat/matroskaenc.c;h=367eb0e08238f73bef78645e938f73f0505a437e;hb=HEAD#l1657 these 2 functions for example
[22:14] <ubitux> or this timing parsing in codecs: http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/srtenc.c;h=89c26dcbf09326bc13451063e8c08dd8041ce549;hb=HEAD#l266
[22:14] <ubitux> all this stuff can be dropped nowadays thanks to CODEC_ID_SUBRIP
[22:15] <wm4> oh
[22:15] <ubitux> the main problem is how to keep cmd line compatibles
[22:15] <ubitux> typically, i'd like to keep -c:s srt working
[22:15] <ubitux> (re-mapped to subrip automatically)
[22:18] <J_Darnley> Can't you just add a "new" codec called "srt" with the struct filled with all subrip's functions?
[22:19] <ubitux> that codec already exists
[22:21] <ubitux> but yeah
[22:33] <rcombs> [insert similar set of ASS-related concerns]
[22:35] <ubitux> rcombs: i just sent a patch to remove some old ASS boilerplate code
[22:35] <rcombs> :D
[22:35] <ubitux> which wasn't in use since a while
[22:39] <wm4> now ffmpeg just needs to stop writing invalid ASS
[22:45] <ubitux> wm4: sure; gonna do it? :P
[22:45] Action: wm4 looks away
[22:46] <rcombs> got a sample of that issue?
[22:49] <ubitux> https://trac.ffmpeg.org/ticket/3207 ?
[22:50] <rcombs> sounds like a healthy dose of commas is in order
[22:52] <wm4> I'd really like to change libass so that it accepts standard ASS only and to piss off everyone, but I guess that's not viable
[22:52] <wm4> possibly too many files in the wild
[22:53] <ubitux> i don't mind giving hints to help fixing the problem in FFmpeg
[22:54] <ubitux> basically, you'd have to change the header in lavc/ass.c, adjust the default values in that same file, update the struct in lavc/ass_split.c
[22:55] <ubitux> then make sure the "Dialogue:" lines are properly written (see lavc/ass.c and lavf/assenc.c)
[22:55] <ubitux> written, and parsed :p
[22:58] <ubitux> the scenarios for testing includes ffmpeg -i in.ass -c copy out.ass, ffmpeg -i in.ass -c copy out.mkv, ffmpeg -i in.ass -c ass out.mkv, ffmpeg -i in.srt out.ass, ffmpeg -i in.ass out.srt
[22:58] <ubitux> hf...
[23:03] <cone-977> ffmpeg.git 03Reynaldo H. Verdejo Pinochet 07master:5e7524195c6a: ffserver: tests, force +global_header for ASF
[00:00] --- Mon Sep  8 2014