[Ffmpeg-devel-irc] ffmpeg-devel.log.20111114

burek burek021 at gmail.com
Tue Nov 15 02:05:03 CET 2011


[00:47] <CIA-41> ffmpeg: 03Ronald S. Bultje 07master * r8283f90a52 10ffmpeg/libswscale/x86/scale.asm: 
[00:47] <CIA-41> ffmpeg: swscale: handle unaligned buffers in yuv2plane1
[00:47] <CIA-41> ffmpeg: The issue had been introduced in
[00:47] <CIA-41> ffmpeg: c435653627529e22d74214c2266f571255e404d6
[00:47] <CIA-41> ffmpeg: Signed-off-by: Luca Barbato <lu_zero at gentoo.org>
[00:47] <CIA-41> ffmpeg: 03Luca Barbato 07master * r84fb63ed23 10ffmpeg/doc/t2h.init: 
[00:47] <CIA-41> ffmpeg: doc: prettyfy the texi2html documentation
[00:47] <CIA-41> ffmpeg: make it use the website css and start structuring it so it is consistent
[00:47] <CIA-41> ffmpeg: 03Luca Barbato 07master * r58b68d6b36 10ffmpeg/doc/t2h.init: 
[00:47] <CIA-41> ffmpeg: texi2html: remove stray \n
[00:47] <CIA-41> ffmpeg: Single-quoted strings are printed verbatim in perl.
[00:47] <CIA-41> ffmpeg: 03Martin Storsjö 07master * r6aa0b98fb2 10ffmpeg/libavformat/ (avformat.h avio.h url.h): (log message trimmed)
[00:47] <CIA-41> ffmpeg: avio: Add AVIOInterruptCB
[00:47] <CIA-41> ffmpeg: This is a better io interrupt callback function, which has an
[00:47] <CIA-41> ffmpeg: opaque parameter, which is given to the interrupt callback.
[00:47] <CIA-41> ffmpeg: This allows callers to precisely cancel IO for one single
[00:47] <CIA-41> ffmpeg: AVFormatContext, without interrupt other ones in the same
[00:47] <CIA-41> ffmpeg: process.
[00:47] <CIA-41> ffmpeg: 03Martin Storsjö 07master * rc4a090ddb5 10ffmpeg/libavformat/ (avio.c url.h): 
[00:47] <CIA-41> ffmpeg: avio: Add an internal utility function for checking the new interrupt callback
[00:47] <CIA-41> ffmpeg: Signed-off-by: Anton Khirnov <anton at khirnov.net>
[00:47] <CIA-41> (22 lines omitted)
[03:29] <michaelni> Daemon404, more vble pacthes for you to review on ffmpeg-dev :)
[03:33] <Daemon404> michaelni, ill take a look in 1-2 hrs. i have a rather busy night tonight
[03:33] <Daemon404> and you should probably sleep
[03:33] <michaelni> :)
[03:33] <michaelni> no hurry, ill look at your review tomorrow
[03:33] <Daemon404> kk
[05:56] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rb1b0fd2790 10ffmpeg/libavfilter/vf_pad.c: 
[05:56] <CIA-41> ffmpeg: pad: fix format string length
[05:56] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:56] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r70eb58cd92 10ffmpeg/libavfilter/vsrc_mandelbrot.c: 
[05:56] <CIA-41> ffmpeg: mandelbrot: avoid INFINITY as openbsd seems not to like it.
[05:56] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:04] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r73af8ea356 10ffmpeg/libavfilter/vsrc_mandelbrot.c: 
[06:04] <CIA-41> ffmpeg: mandelbrot: add math.h for log2()
[06:04] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[07:26] <CIA-41> ffmpeg: 03Clément BSsch 07master * reef3ea8c71 10ffmpeg/libavformat/ (act.c bintext.c bit.c mxfdec.c): Rename remaining ByteIOContext to AVIOContext.
[07:26] <CIA-41> ffmpeg: 03Clément BSsch 07master * r39edfcc9e8 10ffmpeg/libavformat/ (act.c bintext.c): Rename remaining url_fseek to avio_seek.
[07:26] <CIA-41> ffmpeg: 03Clément BSsch 07master * rf3c47e4100 10ffmpeg/libavformat/bintext.c: Rename remaining url_fsize to avio_size.
[07:26] <CIA-41> ffmpeg: 03Clément BSsch 07master * r53d70a334a 10ffmpeg/libavformat/ (act.c asfdec.c bintext.c g723_1.c): Rename remaining get_buffer to avio_read.
[07:26] <CIA-41> ffmpeg: 03Clément BSsch 07master * r055cda8acc 10ffmpeg/libavformat/ (act.c bintext.c g723_1.c mov.c): Rename remaining get_byte to avio_r8.
[07:26] <CIA-41> ffmpeg: 03Clément BSsch 07master * rb04ee90043 10ffmpeg/libavformat/bintext.c: Rename remaining av_metadata_set2 to av_dict_set.
[07:26] <CIA-41> ffmpeg: 03Clément BSsch 07master * r7b79305a47 10ffmpeg/libavformat/ (act.c bintext.c): Rename remaining url_fskip to avio_skip.
[07:26] <CIA-41> ffmpeg: 03Clément BSsch 07master * r49651063da 10ffmpeg/libavformat/bintext.c: Use AVIOContext.seekable instead of deprecated url_is_streamed.
[07:26] <CIA-41> ffmpeg: 03Clément BSsch 07master * r63b796b0dd 10ffmpeg/libavformat/ (g723_1.c mpegts.c): Rename remaining url_ftell to avio_tell.
[07:27] <CIA-41> ffmpeg: 03Clément BSsch 07master * r07c7ffc04b 10ffmpeg/libavformat/act.c: Rename remaining get_le32 to avio_rl32.
[07:27] <CIA-41> ffmpeg: 03Clément BSsch 07master * rc142e2a00f 10ffmpeg/libavformat/ (act.c bintext.c): Rename remaining get_le16 to avio_rl16.
[10:47] <ubitux> does anyone know an audio container in which we could randomly put extradata in the stream?
[11:49] <j-b> :win 1
[11:52] <av500> j-b: no, you lose
[11:55] <j-b> :)
[11:55] <j-b> av500: what's up?
[11:56] <av500> not much
[13:59] <spaam> av500: any new embedded stuff you working with ? 
[14:39] <av500> spaam: yes, I make LEDs blink
[14:40] <spaam> av500: sending message using it ? 
[17:46] <ubitux> michaelni: asm optimiztaions  small typo in $subj (vble patch)
[17:48] <michaelni> ubitux, https://ffmpeg.org/trac/ffmpeg/ticket/523 (is this caused by the remap code?)
[17:48] <michaelni> and ill fix the typo before commit 
[17:48] <michaelni> of course
[17:49] <michaelni> thx for pointing me to it
[17:49] <ubitux> i'm going to check
[18:05] <ubitux> michaelni: indeed, it looks like a regression after map_channel was introduced
[18:10] <ubitux> i wonder what's wrong&
[18:17] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r6718536ba5 10ffmpeg/libavcodec/diracdec.c: 
[18:17] <CIA-41> ffmpeg: diracdec: fix empty data units.
[18:17] <CIA-41> ffmpeg: Fixed Ticket636
[18:17] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:22] <ubitux> ok so before mapch, we have an init with in ch count=1, and in ch count=2 in swr_init (so it triggers the in ch layout reset to 0 condition because in ch layout is mono)
[18:22] <ubitux> while after the mapch, the (in/used) ch count is still set to 1
[18:23] <ubitux> (in swr_init)
[18:23] <ubitux> so the ch count mismatch condition is not triggered
[18:27] <ubitux> -    if(!s->used_ch_count)
[18:27] <ubitux> +    if(!s->used_ch_count || (!s->channel_map && s->used_ch_count != s->in.ch_count)) s->used_ch_count= s->in.ch_count;
[18:27] <ubitux> this is a quick and dirty hack
[18:27] <ubitux> afk for a while, i'm looking at it when i'm back
[18:55] <ubitux> ok got a better fix in mind
[19:11] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * recf5e70548 10ffmpeg/libavcodec/vble.c: 
[19:11] <CIA-41> ffmpeg: vble: use dsp.add_hfyu_median_prediction() this allows asm optimizations to be used.
[19:11] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:11] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r75c0ddc9c2 10ffmpeg/libavcodec/vble.c: 
[19:11] <CIA-41> ffmpeg: vble: move 2nd pass bitreading into vble_restore_plane()
[19:11] <CIA-41> ffmpeg: this improves data cache hits and speeds up grayscale only decoding
[19:11] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:31] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r1fea432b7a 10ffmpeg/libavformat/mp3enc.c: 
[21:31] <CIA-41> ffmpeg: mp3enc: simplify mp3_write_xing()
[21:31] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:31] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r40176fc314 10ffmpeg/libavformat/mp3enc.c: 
[21:31] <CIA-41> ffmpeg: mp3enc: Try to use a matching bitrate in the xing header.
[21:31] <CIA-41> ffmpeg: Might Fix Ticket431
[21:31] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:32] <ubitux> michaelni: actually, i can't think of anything better than: http://blackhole.ubitux.fr/0001-swr-set-and-reset-used_ch_count-the-right-way.patch
[21:39] <michaelni> ubitux, 
[21:39] <michaelni> how does "!s->channel_map && s->used_ch_count != s->in.ch_count"
[21:39] <michaelni> happen ?
[21:48] <paulbrookfield> Anyone around for a question about multiple connections in a protocol?
[22:05] <ubitux> michaelni: mmh forget that, this should be better: http://blackhole.ubitux.fr/0001-swr-set-the-default-used_ch_count-value-to-zero.patch
[22:08] <ubitux> maybe i should give more insight in the description
[22:11] <ubitux> i'll push and update the ticket if you're ok
[22:13] <ubitux> also, should we add this file to a regtest?
[22:16] <michaelni> ubitux, patch LGTM 
[22:16] <ubitux> ok
[22:16] <ubitux> thx
[22:17] <michaelni> about fate, if you want ...
[22:17] <CIA-41> ffmpeg: 03Clément BSsch 07master * rc8136ebd6a 10ffmpeg/libswresample/swresample.c: 
[22:17] <CIA-41> ffmpeg: swr: set the default used_ch_count value to zero.
[22:17] <CIA-41> ffmpeg: This allows an appropriate behaviour when it is not specified (fallback
[22:17] <CIA-41> ffmpeg: to the number of input channels).
[22:17] <CIA-41> ffmpeg: Fixes Ticket523
[22:20] <ubitux> michaelni: my concern is about the ac3 channel layout guess, which need to always be wrong in the future or else the test would become void
[22:56] <dilaroga> michaelni: patch 3/7 is in moderation queue, i sent it with another email by mistake
[22:57] <michaelni> dilaroga, i was already wondering where it was ...
[22:58] <j-b> dilaroga: ah, the answer is here :à
[22:58] <j-b> dilaroga: \o/
[22:58] <j-b> dilaroga: good job!
[22:58] <j-b> dilaroga: VLC 1.2 with VDA, then?
[22:59] <michaelni> pasteeater, switching the ML moderation notify mails off is not as nice as i initially thought
[23:01] <dilaroga> michaelni: should i re-send it ?
[23:02] <michaelni> no
[23:02] <michaelni> i just let it through
[23:03] <michaelni> dilaroga, btw, as we already speak, any reason why you send patch instead of just pushing to your repo and asking me to merge ?
[23:03] <michaelni> patchES
[23:06] <dilaroga> I prefer that they are reviewed before pushing them :)
[23:07] <michaelni> of course but you are the VDA expert, its unlikely others will spot things you missed
[23:07] <michaelni> not counting whitespace errors and english typos
[23:09] <j-b> being a VDA expert doesn't make you a h264 expert, if I may say so...
[23:09] <j-b> I wrote a CrystalHD decoder. I still don't understand correctly h264
[23:10] <dilaroga> j-b: you're right :)
[23:11] <michaelni> dilaroga, but if you prefer, i could also just approve patches next time, and you can then push them to your repo once you got enough reviews and then tell me to merge ?
[23:12] <michaelni> j-b, dilaroga btw ,theres nothing that requires h264 knowledge in todays patchset :)
[23:14] <dilaroga> michaelni: ok for me for approving the patches and pushing after to my public repo
[23:15] <michaelni> dilaroga, ok, then should i wait with todays too ? they are rather trivial and just passed fate tests :)
[23:15] <michaelni> or should i push ?
[23:16] <dilaroga> michaelni: i'll push them now ;)
[23:18] <dilaroga> michaelni: done
[23:19] <michaelni> dilaroga, should i merge them or wait ?
[23:19] <dilaroga> michaelni: it's ok for me, you can merge them
[23:23] <CIA-41> ffmpeg: 03Sebastien Zwickert 07master * r9837442a39 10ffmpeg/libavcodec/vda.h: vda: undef Picture.
[23:23] <CIA-41> ffmpeg: 03Sebastien Zwickert 07master * rb46134f659 10ffmpeg/libavcodec/ (vda.c vda.h vda_h264.c vda_internal.h): vda: reordering headers inclusion.
[23:23] <CIA-41> ffmpeg: 03Sebastien Zwickert 07master * ra40867f27b 10ffmpeg/configure: configure: add vda dependencies.
[23:23] <CIA-41> ffmpeg: 03Sebastien Zwickert 07master * r6e1e49385e 10ffmpeg/libavcodec/vda.c: vda: cosmetic.
[23:23] <CIA-41> ffmpeg: 03Sebastien Zwickert 07master * rc7fa3eeab9 10ffmpeg/libavcodec/vda.c: vda: removes useless casts.
[23:23] <CIA-41> ffmpeg: 03Sebastien Zwickert 07master * r81852ef5d6 10ffmpeg/libavcodec/vda_h264.c: vda: get correctly the frame from h264 context.
[23:23] <CIA-41> ffmpeg: 03Sebastien Zwickert 07master * r4590e161e6 10ffmpeg/libavcodec/vda.c: vda: check allocation result.
[23:25] <michaelni> dilaroga, as you are a maintainer now, you get op here :)
[23:26] <dilaroga> michaelni: oh! when my mom will see that ! :) Thanks you
[23:59] <ubitux> dilaroga: could you use K&R coding style?
[00:00] --- Tue Nov 15 2011


More information about the Ffmpeg-devel-irc mailing list