[Ffmpeg-devel-irc] ffmpeg-devel.log.20120908

burek burek021 at gmail.com
Sun Sep 9 02:05:02 CEST 2012


[00:04] <saste> alternatively we could just copy min(dst_linesize, bytewidth)
[00:04] <saste> but not that I like it much
[00:07] <michaelni> that would fail with negative linesize also it would fail with src_linesize<dst_linesize<bytewidth
[00:11] <saste> michaelni: last question: which level of assert?
[00:16] <michaelni> saste, 0 has the advantage that its always checked and i dont think this could have a speed effect there
[00:16] <michaelni> unless you call it on 16x16 blocks of a large frame
[00:16] <saste> uhm ok...
[00:18] <michaelni> but if you prefer a higher level thats fine too
[00:21] <saste> uh asserted in life
[00:25] <saste> ^ no I was wrong
[01:09] <saste> I wonder if it would be useful a function av_samples_get_nb_planes(enum AVSamplesFormat sample_fmt)
[01:09] <saste> or would that bloated?
[01:11] <saste> bloat
[02:31] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r097c64ffcb 10ffmpeg/libavcodec/flashsv2enc.c: 
[02:31] <CIA-56> ffmpeg: flashsv2enc: remove experimental flag, the encoder seems working now
[02:31] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:31] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r9a27fd12f3 10ffmpeg/libavformat/avio.c: 
[02:31] <CIA-56> ffmpeg: avio: replace ETIMEDOUT by EIO
[02:31] <CIA-56> ffmpeg: ETIMEDOUT is not available on all platforms
[02:31] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:31] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r5710dbf88c 10ffmpeg/libavformat/mpegts.c: 
[05:31] <CIA-56> ffmpeg: mpegts: Fix potential memory leak in mpegts_read_packet()
[05:31] <CIA-56> ffmpeg: This leak can happen in case of some error paths.
[05:31] <CIA-56> ffmpeg: Found-by: Tanami, Ohad
[05:31] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:41] <CIA-56> ffmpeg: 03Daniel Verkamp 07master * r124f0b2f46 10ffmpeg/ (3 files in 2 dirs): 
[05:41] <CIA-56> ffmpeg: flashsv2enc: only encode diff blocks when needed
[05:41] <CIA-56> ffmpeg: A flashsv2 block may have a "diff block" to indicate which scan lines of
[05:41] <CIA-56> ffmpeg: the block are actually encoded. However, this diff block need not be
[05:41] <CIA-56> ffmpeg: used when the entire block is coded.
[05:41] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:08] <RT|Chatzilla> "<nevcairiel> RT|AO: i tried your dash, and all it did was crash :(" I'm not sure what you're doing, I wonder if I can reproduce it (base of my MSYS dash is git head version, http://git.kernel.org/?p=utils/dash/dash.git;a=summary )
[06:12] <RT|Chatzilla> and my msys patch is here: http://article.gmane.org/gmane.comp.shells.dash/775
[06:19] <nevcairiel> all i did was grab it, and run my build script with it, like "dash build.sh"
[06:20] <nevcairiel> and boom
[06:33] <nevcairiel> the dash that came with my msys had the same problem, though
[08:31] <RT|Chatzilla> nevcairiel: then it might be version mismatch
[08:32] <RT|Chatzilla> my msys-1.0.dll is 1000.16.0.0
[08:32] <RT|Chatzilla> and dated in 18 December, 2010
[08:33] <nevcairiel> 1000.17.0.0 here, from 25.04.2012
[08:33] <nevcairiel> er
[08:33] <nevcairiel> 2011
[11:43] <RT|Chatzilla> nevcairiel: I tried updating my msys-1.0.dll to your version, but it works as still
[11:44] <nevcairiel> it seems simply running configure works, must be something in the fate script that causes the problem
[11:53] <RT|Chatzilla> you may sent it to dash maillist if you can bisect it. (it seems to be unrelated to my changes)
[11:53] <RT|Chatzilla> s/sent/send/
[11:54] <nevcairiel> bisecting assumes it worked at some point
[11:56] <RT|Chatzilla> or you may try the MSYS port of mksh instead(it is still faster than bash) http://roy.orz.hm/soft/mksh.exe
[11:58] <RT|Chatzilla> functional change is posted here: http://permalink.gmane.org/gmane.os.miros.mksh/50
[12:45] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r8a0118b4b4 10ffmpeg/libavformat/utils.c: 
[12:45] <CIA-56> ffmpeg: ffio_read: dont spam the user at EOF with a truncation error
[12:45] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[13:01] <CIA-56> ffmpeg: 03Mans Rullgard 07master * r5fe64d88f6 10ffmpeg/libavcodec/x86/dsputil_mmx.c: 
[13:01] <CIA-56> ffmpeg: x86: allow using add_hfyu_median_prediction_cmov on any cpu with cmov
[13:01] <CIA-56> ffmpeg: For some reason add_hfyu_median_prediction_cmov is only selected
[13:01] <CIA-56> ffmpeg: on 3Dnow-capable CPUs, even though it uses no 3Dnow instructions.
[13:01] <CIA-56> ffmpeg: This patch allows it to be selected on any cpu with cmov with the
[13:01] <CIA-56> ffmpeg: possibility of being overridden by the mmxext version.
[13:01] <CIA-56> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[13:01] <CIA-56> ffmpeg: 03Mans Rullgard 07master * r6efb698883 10ffmpeg/libavcodec/ (5 files in 2 dirs): 
[13:01] <CIA-56> ffmpeg: cavsdsp: set idct permutation independently of dsputil
[13:01] <CIA-56> ffmpeg: CAVS uses its own idct so using dsputil to set the permutation
[13:01] <CIA-56> ffmpeg: is fragile.
[13:01] <CIA-56> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[13:01] <CIA-56> ffmpeg: 03Justin Ruggles 07master * r7327525997 10ffmpeg/libavutil/x86/float_dsp.asm: 
[13:01] <CIA-56> ffmpeg: x86: float_dsp: fix ff_vector_fmac_scalar_avx() on Win64
[13:01] <CIA-56> ffmpeg: The SWAP macro does not work for explicit xmm/ymm usage, so instead just move
[13:01] <CIA-56> ffmpeg: the scalar value from xmm2 to xmm0.
[13:01] <CIA-56> ffmpeg: 03Martin Storsjö 07master * r7b699d8136 10ffmpeg/libavformat/mov_chan.c: 
[13:01] <CIA-56> ffmpeg: mov_chan: Only set the channel_layout if setting it to a nonzero value
[13:01] <CIA-56> ffmpeg: If regularly parsing new chan atoms (as in rtpdec_qt), but the
[13:01] <CIA-56> ffmpeg: chan atoms don't actually contain any channel layout, don't reset
[13:01] <CIA-56> ffmpeg: the value that the caller has filled in (by guessing).
[13:02] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[13:02] <CIA-56> ffmpeg: 03Diego Biurrun 07master * r1169f0d0af 10ffmpeg/ (4 files in 2 dirs): x86: more specific checks for availability of required assembly capabilities
[13:02] <CIA-56> ffmpeg: 03Martin Storsjö 07master * r5f72bc02f8 10ffmpeg/libavformat/mov_chan.c: 
[13:02] <CIA-56> ffmpeg: mov_chan: Reindent an incorrectly indented line
[13:02] <CIA-56> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[13:02] <CIA-56> ffmpeg: 03Diego Biurrun 07master * r8cb7ed5562 10ffmpeg/libavcodec/x86/ (5 files): x86: avcodec: Drop silly "_mmx" suffix from dsputil template names
[13:02] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r7beadfe1f7 10ffmpeg/: (log message trimmed)
[13:02] <CIA-56> ffmpeg: Merge remote-tracking branch 'qatar/master'
[13:02] <CIA-56> ffmpeg: * qatar/master:
[13:03] <CIA-56> ffmpeg:  mov_chan: Only set the channel_layout if setting it to a nonzero value
[13:03] <CIA-56> (7 lines omitted)
[13:45] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rb86e18fe46 10ffmpeg/libavformat/yuv4mpeg.c: 
[13:45] <CIA-56> ffmpeg: yuv4mpeg: fix potential alignment bug
[13:45] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:11] <saste> michaelni: sorry to ask always you
[14:11] <saste> would be possible in principle to embed a whatever audio stream (with stream copy) from X to NUT?
[14:12] <saste> right now only a limited subset of stream formats has a defined tag, so for that it is not possible ATM
[14:14] <michaelni> it should be possible
[14:57] <Daemon404> michaelni, im not exactly sure how to fix mplayer
[14:57] <Daemon404> it uses its own config.h
[15:01] <nevcairiel> is that define even in a public header?
[15:02] <Daemon404> it ws in attributes.h before
[15:03] <Daemon404> but its in config.h now, as av_restrict varies between compilers
[15:03] <nevcairiel> attributes.h is public i take it
[15:03] <nevcairiel> and with public, i mean installed
[15:04] <Daemon404> yes
[15:07] <Daemon404> fixing mplayer doesnt seem trivial
[15:07] <nevcairiel> removing it is somewhat of an API break then, maybe just leave it be for the time being? No harm done, etc
[15:07] <Daemon404> yeah
[15:07] <nevcairiel> it doesnt block anything, just cleanup, right?
[15:08] <Daemon404> yes
[16:11] <RT|Chatzilla> nevcairiel: the dash-fate issue might be caused by this: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=642922
[16:14] <nevcairiel> isnt that supposedly fixed
[16:16] <nevcairiel> or only fixed with the debian patches
[16:18] <RT|Chatzilla> it exists as debian patch only, not upstream committed.
[16:26] <RT|Chatzilla> I recompiled with this patch applied, you may redownload it from same URL
[16:39] <nevcairiel> i'm doing some maintenance on my fate system right now, but i'll give it a try when i'm done, thanks
[16:41] <Daemon404> [10:18] < RT|Chatzilla> it exists as debian patch only, not upstream committed.
[16:41] <Daemon404> ^ i hate package maintainers that do this
[16:42] <nevcairiel> debian is full of those people, even though sometimes its just driven by them using years old versions and having to patch that into shape somehow
[16:42] <RT|Chatzilla> dash upstream isn't very actively maintained by the way
[16:45] <Daemon404> nevcairiel, many times it is laziness
[16:45] <Daemon404> or stupid foss grudges
[16:45] <nevcairiel> that too
[16:45] <Daemon404> (we dont have any of those around here, right? right.)
[16:45] <burek> or lack of beer
[16:46] <Daemon404> i usually try and rescue patches from oe-core hell
[16:46] <Daemon404> since i can watch that fairly easily
[16:46] <Daemon404> but watching all of debian is lololol
[17:48] <Daemon404> anyone wanna peek at a few of my patches?
[17:48] <Daemon404> some have no review yet :)
[17:52] <saste> Daemon404: I'm not really qualified on that area, but I'll try to have a look
[17:52] <Daemon404> :)
[17:58] <Compn> Daemon404 : you should just fix mp filters :P
[17:59] <Compn> reimar reviews patches quickly :P
[17:59] <Daemon404> working on mplayer ftl
[17:59] <Daemon404> >svn
[17:59] <Daemon404> >custom configure bs
[17:59] <Daemon404> >have to add inline asm disabling support
[17:59] <Daemon404> yeah no
[18:00] <Compn> ffmpeg's configure is custom
[18:00] <Compn> no autotools here :P
[18:00] <Daemon404> autotools > mplayers configure
[18:01] <Daemon404> ffmpeg's special flower configure is kind of annoying too sometimes
[18:01] <Compn> where are your patches foo
[18:02] <Daemon404> id have a better chance of teaching the entirety of the usa to speak cantonese than converting ffmpeg to autotools
[18:02] <Daemon404> without at least 40 death threats
[18:03] Action: Compn cant remember if he likes mandarin or cantonese better
[18:03] <Compn> whichever language the movie was originally filmed in...
[18:04] <Daemon404> 'the movie' ?
[18:04] <CIA-56> ffmpeg: 03Derek Buitenhuis 07master * r9ab538e31f 10ffmpeg/libavutil/random_seed.c: 
[18:04] <CIA-56> ffmpeg: random_seed: Replace a VLA with a normal array
[18:04] <CIA-56> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[18:04] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:06] <Compn> i watch a lot of films from china...
[18:06] <Daemon404> i see
[18:06] <Daemon404> i am made to watch many against my will
[18:06] <Compn> many of them have dual language audio tracks
[18:06] <Compn> you dont like hopping vampires ?
[18:07] <Compn> Daemon404 : http://www.youtube.com/watch?v=kbuGbG3Fv1Y
[18:08] <Compn> not a hopping vampire
[18:08] <Daemon404> >_>
[18:08] <Compn> but funny
[18:08] <Daemon404> im mostly made to watch wuxia
[18:08] <Daemon404> or terrible historical dramas
[18:08] <Compn> oh yeah i avoid those
[18:08] <Daemon404> ofc all rmvb
[18:08] <Daemon404> and 30 mb
[18:09] <Compn> dunno how china picked rmvb after vcd
[18:09] <Compn> or why they call it rmvb instead of rm
[18:23] <Daemon404> oh wee, more msvc-related patches in
[19:36] <Daemon404> michaelni, https://github.com/mstorsjo/c99-to-c89/commit/1a6abcfb0887bb6244fb6a868b5ff99d83ed3e54 <-- it exists, but it isn't in master branch yet. libav needs to work some things out afaik.
[21:35] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rebc854e4e9 10ffmpeg/libswresample/rematrix.c: 
[21:35] <CIA-56> ffmpeg: swr/auto_matrix: move layouts to local variables
[21:35] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:35] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rc8ffab06f4 10ffmpeg/libswresample/rematrix.c: 
[21:35] <CIA-56> ffmpeg: swr/auto_matrix: add clean layout function
[21:35] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:35] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * r0e78e575ed 10ffmpeg/libswresample/rematrix.c: 
[21:35] <CIA-56> ffmpeg: swr/auto_matrix: treat a single channel as mono instead of failing
[21:35] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:35] <CIA-56> ffmpeg: 03Michael Niedermayer 07master * rf8d74cbdc3 10ffmpeg/libswresample/rematrix.c: 
[21:35] <CIA-56> ffmpeg: swr/auto_matrix: treat stereo downmix like stereo.
[21:35] <CIA-56> ffmpeg: Based on code by John Stebbins
[21:35] <CIA-56> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:00] --- Sun Sep  9 2012


More information about the Ffmpeg-devel-irc mailing list