[Ffmpeg-devel-irc] ffmpeg-devel.log.20120802

burek burek021 at gmail.com
Fri Aug 3 02:05:02 CEST 2012


[02:07] <durandal_1707> michaelni: what poly i should use for crc24?
[02:22] <michaelni> durandal_1707, it should be written in the RFC
[02:23] <durandal_1707> michaelni: but it is bigger than max 24bit number and i cant get it work with crc stuff in lavu
[02:23] <michaelni> what is its value ?
[02:24] <durandal_1707> http://svn.noreply.org/cgi-bin/viewvc.cgi/Mixmaster/branches/Mix-stats/Src/pgp.c?r1=262&r2=274&pathrev=821&logsort=cvs&view=patch
[02:24] <durandal_1707> #define POLY 0X1864CFB
[02:26] <michaelni> try to drop the leading 1
[02:26] <Daemon404> hey michaelni, can you take a look at the latest canopus patch (v5). libav OK'd me on it, but it'd rather get everyone OK'd before I push.
[02:28] <durandal_1707> Daemon404: i cant ok redundant initializations
[02:28] <Daemon404> im going to keep it explicit.
[02:28] <durandal_1707> i'm not going to OK it
[02:28] <Daemon404> if you wanan block it, then tough, as it'll end up in libav anyway.
[02:29] <durandal_1707> i dont wanna block it.....
[02:29] <Daemon404> sure sounds like it.
[02:29] <Daemon404> "your code is doign something non-harmful, and beign explicit. i cannto accept this."
[02:29] <Daemon404> "because it doesnt match my style"
[02:31] <durandal_1707> no it is just redundant, that does not mean code is crap
[02:31] <Daemon404> i prefer it be explicit.
[02:31] <Daemon404> even if it i know priv_Data gets zero'd at alloc time
[02:31] <Daemon404> it's easier to read.
[02:51] <Daemon404> michaelni, thank you
[02:51] <michaelni> np
[02:54] <Compn> lol
[02:54] <Compn> Daemon404 : just means durandal_1707 wont 'LGTM' on it :P
[02:54] <Daemon404> LBTM
[02:54] <Compn> feels bad man
[03:01] <Daemon404> michaelni, im not sure how to fix 'info_offset > avpkt->size - 8' gracefully 
[03:01] <Daemon404> i could add a second check, but thats ugly
[03:03] <durandal_1707> michaelni: removing leading 1 does not help...
[03:04] <michaelni> Daemon404, info_offset + 8 > avpkt->size
[03:04] <Daemon404> ...
[03:04] <Daemon404> note to self: afterworks beers = bad
[03:04] <Daemon404> -s
[03:06] <Daemon404> or maybe im high off teh raid i sprayed in here earlier at some ants
[03:06] Action: Daemon404 shrug
[03:09] <michaelni> durandal_1707, did you try setting the initial value to 0xB704CE
[03:11] <durandal_1707> michaelni: i do that for each av_crc
[03:12] <llogan> Daemon404: i read "rats" at first...
[03:16] <Daemon404> lol
[03:16] <Daemon404> i was having some ants in my apartment
[03:17] <michaelni> durandal_1707, hmm try bswap it
[03:19] <Compn> man
[03:19] <Compn> ants are like the best insects to have
[03:19] <Compn> honestly
[03:19] <Compn> they attack the cockroaches :P
[03:27] <Daemon404> Compn, they also crawl EVERYWHERE
[03:28] <Daemon404> also i saw a hosue centipede 
[03:28] <Daemon404> fucking scary.
[03:28] <Daemon404> michaelni, new patch for canopus sent. should be the last.
[03:29] <llogan> Daemon404: one of those fast bastards? they're weird.
[03:29] <Daemon404> yes
[03:29] <Daemon404> 30 leggers
[03:29] <Daemon404> michaelni, replied to wrong thread... but you can sitll OK it there.
[03:29] Action: Daemon404 made a royal mess of his threads
[03:30] <Daemon404> llogan, im from northern ontario... i aint used to city bugs
[03:30] <llogan> i took a scanning electron microscopy class once. looked at some silverfish shit. most interesting class.
[03:30] <Daemon404> lol
[03:32] <llogan> i don't have that stuff here. just occasional derelicts/drunks/bums/mentals and mice.
[03:32] <Daemon404> lol
[03:33] <Daemon404> oh well. i killed a bunch of brain cells with raid today.
[03:34] <llogan> damn. i'll miss your contributions.
[04:03] <Daemon404> michaelni, pushed to my github. it can be merged.
[04:05] <durandal_1707> michaelni: bswap what? i have 5byte data and 3byte crc so it can be tested...
[04:28] <durandal_1707> michaelni: this it how table should look like http://www.koders.com/cpp/fid3331C1D57FDF3031632BEEE6022A1E78E3A1FF02.aspx?s=rsa
[04:29] <durandal_1707> that is what codec use (i see it in assembly code)
[04:30] <Daemon404> i didnt think anything used crc24 irl
[04:30] <Daemon404> all 16 or 32
[04:30] <durandal_1707> Daemon404: there is poor which use 8bit
[04:30] <durandal_1707> or even less
[04:30] <Daemon404> not really a good check, now is it
[04:34] <durandal_1707> and table i generate is something else ...
[04:35] <Daemon404> everything i see use crc ever seems to hardcode tables
[04:35] <durandal_1707> Daemon404: lavu have crc which generate crc at runtime
[04:36] <Daemon404> i figured
[04:36] <durandal_1707> and other codecs like mlp dont have big table but follow same behaviour by generating table
[04:37] <durandal_1707> hah table entries need swapping
[04:38] <durandal_1707> actually not, they are in reverse order, is that normal?
[04:38] Action: Daemon404 has no diea.
[04:38] <Daemon404> idea8
[04:39] <durandal_1707> fb4c86 instead of 864cfb
[04:39] <Daemon404> endianness?
[04:44] <durandal_1707> dont think so, swapping causes table to get wrong results...
[04:48] <durandal_1707> got it!!! i need to swap result and parameter
[04:48] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r98a2f4e87b 10ffmpeg/configure: 
[04:48] <CIA-41> ffmpeg: configure: reduce difference to qatar by 2 lines
[04:48] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:48] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r571572fcdd 10ffmpeg/libavcodec/ (mpegaudio.h mpegaudio_tablegen.h mpegaudiodsp_template.c): 
[04:48] <CIA-41> ffmpeg: mp3dec: adjust IMDCT scaling to avoid overflows
[04:48] <CIA-41> ffmpeg: Fixes ticket268
[04:48] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:54] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r91c86d3cf8 10ffmpeg/ (9 files in 4 dirs): 
[04:54] <CIA-41> ffmpeg: Canopus Lossless decoder
[04:54] <CIA-41> ffmpeg: At the moment it only does BGR24, but I plan to add the rest after.
[04:54] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[04:55] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r96cb143800 10ffmpeg/: 
[04:55] <CIA-41> ffmpeg: Merge remote-tracking branch 'dwbuiten/master'
[04:55] <CIA-41> ffmpeg: * dwbuiten/master:
[04:55] <CIA-41> ffmpeg:  Canopus Lossless decoder
[04:55] <CIA-41> ffmpeg: Merged-by: Michael Niedermayer <michaelni at gmx.at>
[06:16] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * ra6aa7e039a 10ffmpeg/libavcodec/mpegaudio_tablegen.h: 
[06:16] <CIA-41> ffmpeg: mpegaudio_tablegen: try to fix hardcoded tables
[06:16] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[08:56] <ubitux> michaelni: are you the maintainer of the asan box on fate?
[08:57] <ubitux> if so, i think you should --disable-memory-poisoning for this one
[11:57] <michaelni> ubitux, added, thx
[12:32] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r8cac86e091 10ffmpeg/libavcodec/vorbisdec.c: 
[12:32] <CIA-41> ffmpeg: vorbisdec: fix heap buffer overflow.
[12:32] <CIA-41> ffmpeg: Found-by: Dale Curtis <dalecurtis at chromium.org>
[12:32] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[12:46] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r65520f5933 10ffmpeg/libavfilter/ (5 files): 
[12:46] <CIA-41> ffmpeg: lavfi/video: remove duplicated function ff_null_start_frame_keep_ref()
[12:46] <CIA-41> ffmpeg: Semantic for the function ff_null_start_frame() was changed in
[12:46] <CIA-41> ffmpeg: 07bad27810cd, and it has now the same behavior of
[12:46] <CIA-41> ffmpeg: ff_null_start_frame_keep_ref(), thus it makes no sense to keep both of
[12:46] <CIA-41> ffmpeg: them.
[12:46] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r3bcde3f06c 10ffmpeg/libavfilter/ (drawutils.c drawutils.h): lavfi/drawutils: make ff_draw_color() accept a const rgba map
[12:46] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * raa5b93fbd8 10ffmpeg/libavfilter/vsrc_testsrc.c: 
[12:46] <CIA-41> ffmpeg: lavfi/rgbtestsrc: use ff_fill_rgba_map() rather than a local copy
[12:46] <CIA-41> ffmpeg: Also convert int rgba_map[4] to uint8_t rgba_map[4], as required by
[12:46] <CIA-41> ffmpeg: ff_fill_rgba_map().
[12:46] <CIA-41> ffmpeg: Simplify.
[14:52] <CIA-41> ffmpeg: 03Nicolas George 07master * r93d428953c 10ffmpeg/libavformat/assdec.c: assdec: return EOF instead of EIO.
[17:07] <CIA-41> ffmpeg: 03Kifayat ullah 07master * r32aeba1275 10ffmpeg/libavformat/mpegtsenc.c: 
[17:07] <CIA-41> ffmpeg: mpegtsenc: Fix AC3 Sound issue with Blu-ray Players.
[17:07] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:11] <Daemon404> michaelni, you can either apply my eadmad patch from ffmpeg-devel, or merge it in from libav
[19:11] <Daemon404> i dont think its worth doign a merge from my repo for
[19:11] <Daemon404> for such a small change
[19:47] <llogan> Daemon404: have you sucessfully gotten fdk-aac to perform a quality_based encode?
[19:48] <Daemon404> yes
[19:48] <llogan> what's the command you used?
[19:50] <Daemon404> -flags +qscale -global_quality 1[-5]
[19:50] <Daemon404> er
[19:50] <Daemon404> [1-5]
[19:51] <llogan> that's what i thought. http://pastebin.com/uckpwjqD
[19:51] <llogan> seems to choke on the -flags +qscale
[19:52] <Daemon404> you cant use 1 (lowest bitrate vbr mode) with aac_low profile
[19:52] <Daemon404> only he-aac
[19:52] <Daemon404> and pals
[19:53] <llogan> ah. i had a feeling i had a hidden -flags +ignorance in there.
[19:53] Action: JEEB pats llogan 
[19:53] <JEEB> :)
[19:53] <JEEB> well, now we are a bit more in the know
[19:53] <Daemon404> ~undocumented things~
[19:53] <llogan> taking it for the team
[20:04] <msmithng> are patches required to be sent in via git send-email?
[20:04] <nevcairiel> no
[20:04] <msmithng> sweet
[20:04] <Daemon404> michaelni, some people attach formatted patches. but git-send email is sitll preffered
[20:04] <Daemon404> er... msmithng 
[20:04] <nevcairiel> its easier to review that way, but sending git format-patch is usually fine too
[20:04] <Daemon404> make sure its make with format-patch, at least
[20:04] <Daemon404> made*
[20:04] <nevcairiel> i cba to figure out how to make git send-mail work on windows
[20:05] <msmithng> yes, I have 2 to submit but I'm fighting with git send-email
[20:05] <nevcairiel> the mingw environment that comes with git is lacking some perl things i think
[20:05] <msmithng> the macports version is missing some ... yup
[20:05] <Daemon404> eh?
[20:05] <msmithng> probably the same issues... git-core is shy of a small library
[20:05] <Daemon404> sounds like the packager did it wrong
[20:05] <nevcairiel> msys env that is, not mingw
[21:13] <saste> msmithng: patches written in a postcard and sent to a developers are also fine, especially if packaged with goodies, e.g. a chocolate bar, a box of beers, a bag of gold coins
[21:54] <Mista_D> Encoded TS file with x264 and faac; demuxed it; can't mux it back to TS. MP4/AVI are ok.  http://pastebin.ca/2176923
[22:02] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rf92a99dae5 10ffmpeg/libavcodec/flacenc.c: 
[22:02] <CIA-41> ffmpeg: flacenc: switch to av_assert
[22:02] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:03] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r4d38b838f0 10ffmpeg/libavcodec/flacdec.c: 
[22:03] <CIA-41> ffmpeg: flacdec: switch to av_assert
[22:03] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:05] <michaelni> Mista_D, open a bugreport on https://ffmpeg.org/trac/ffmpeg
[22:06] <Mista_D> michaelni: will do.
[22:11] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r19d40c7912 10ffmpeg/libavcodec/eamad.c: 
[22:11] <CIA-41> ffmpeg: eamad: Use dsputils instead of its own bswap16_buf
[22:11] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[22:11] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:39] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r45eaac02cb 10ffmpeg/ (8 files in 4 dirs): 
[23:39] <CIA-41> ffmpeg: Canopus Lossless decoder
[23:39] <CIA-41> ffmpeg: At the moment it only does BGR24, but I plan to add the rest after.
[23:39] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[23:40] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * ra675d73d57 10ffmpeg/libavcodec/eamad.c: 
[23:40] <CIA-41> ffmpeg: eamad: Use dsputils instead of a custom bswap16_buf
[23:40] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[23:40] <CIA-41> ffmpeg: 03Mans Rullgard 07master * rcf5781fad0 10ffmpeg/libavcodec/vp8.h: 
[23:40] <CIA-41> ffmpeg: vp8: pack struct VP8ThreadData more efficiently
[23:40] <CIA-41> ffmpeg: Reordering the members in this struct reduces the holes required
[23:40] <CIA-41> ffmpeg: to maintain alignment. With this order, the only remaining, and
[23:40] <CIA-41> ffmpeg: unavoidable, hole is 3 bytes following left_nnz.
[23:40] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[23:40] <CIA-41> ffmpeg: 03Mans Rullgard 07master * rec7c501ed5 10ffmpeg/ (6 files in 3 dirs): 
[23:40] <CIA-41> ffmpeg: x86: remove libmpeg2 mmx(ext) idct functions
[23:40] <CIA-41> ffmpeg: These functions are not faster than other mmx implementations on
[23:40] <CIA-41> ffmpeg: any hardware I have been able to test on, and they are horribly
[23:40] <CIA-41> ffmpeg: inaccurate. There is thus no reason to ever use them.
[23:40] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[23:40] <CIA-41> ffmpeg: 03Mans Rullgard 07master * raf500c08bb 10ffmpeg/libavcodec/Makefile: 
[23:40] <CIA-41> ffmpeg: dct-test: always link with aandcttab.o
[23:40] <CIA-41> ffmpeg: Signed-off-by: Luca Barbato <lu_zero at gentoo.org>
[23:40] <CIA-41> ffmpeg: 03Mans Rullgard 07master * rcfb1091898 10ffmpeg/libavcodec/vc1dec.c: 
[23:40] <CIA-41> ffmpeg: vc1dec: remove useless #include simple_idct.h
[23:40] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[23:40] <CIA-41> (20 lines omitted)
[00:00] --- Fri Aug  3 2012


More information about the Ffmpeg-devel-irc mailing list