[Ffmpeg-devel-irc] ffmpeg-devel.log.20120701

[burek]

[01:57] <ubitux> http://news.ycombinator.com/item?id=4183209
[02:01] <durandal_1707> a ask ml maintainers to add rude posters on ignore list
[02:28] <iive> ubitux: i liked reading that. If you are interested in ancient history. The most common complaint against michael was that he demanded the code to be "perfect" before it is accepted into the repository.
[02:29] <ubitux> :)
[02:30] <ubitux> now i just hope my ISP won't joke with my connection until a few days
[02:30] <ubitux> my box is getting raped ATM :D
[02:30] <iive> hum?
[02:30] <ubitux> (i host the blog @home)
[02:31] <ubitux> HN = people & bots army
[02:31] <ubitux> and since i'm on the frontpage.. :p
[02:31] <iive> oh. cable => slow upload.
[02:31] <ubitux> upload: 100ko/sec
[02:31] <ubitux> at best.
[02:31] <iive> and probably some caps.
[02:49] <CIA-41> ffmpeg: 03Marton Balint 07master * r58f22f7012 10ffmpeg/ffplay.c: 
[02:49] <CIA-41> ffmpeg: ffplay: only configure video filters after we got the first frame
[02:49] <CIA-41> ffmpeg: Otherwise the codec width, height and pixel format values may not be set.
[02:49] <CIA-41> ffmpeg: Signed-off-by: Marton Balint <cus at passwd.hu>
[02:49] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rf7cba736ba 10ffmpeg/: 
[02:49] <CIA-41> ffmpeg: Merge remote-tracking branch 'cus/stable'
[02:49] <CIA-41> ffmpeg: * cus/stable:
[02:49] <CIA-41> ffmpeg:  ffplay: only configure video filters after we got the first frame
[02:49] <CIA-41> ffmpeg: Merged-by: Michael Niedermayer <michaelni at gmx.at>
[04:26] <Daemon404> more flamebait i see
[04:29] <durandal_1707> what?
[04:31] Action: Daemon404 is just waiting for holy wars to break out, as tend to happen on the internet whenever blog posts like that surface
[04:31] <durandal_1707> i dont see any holy wars
[04:32] <Daemon404> give it a while.
[04:37] <durandal_1707> where are torrents when you need them?
[04:37] <Daemon404> down, because tbp is down
[04:37] <Daemon404> the internet is in chaos
[04:38] <durandal_1707> crap, cant work on codec because i have nothing to RE it
[04:39] <Daemon404> is that ubitux's blog post btw?
[04:39] <Daemon404> (i did jus tfinish reading it)
[04:41] <bryno> ubitux: great post. you voiced everything that i've been thinking of for some time
[04:42] <Daemon404> i mostly agree with it, but strongly disagree with some parts.
[04:42] <Daemon404> well. one part.
[04:44] <bryno> i haven't kept up with development of either project, but the debian/gentoo part has always bothered me a lot
[04:49] <durandal_1707> Daemon404: what part?
[04:50] <Daemon404> i think all nontrivial code should be reviewed
[04:50] <Daemon404> and coding standards should be enforced
[04:50] <Daemon404> (one of my biggest pet peeves are 'var= val')
[04:51] <Daemon404> not as bikesheddy as libav, but reviewed nonetheless.
[04:52] <durandal_1707> hmm 'var=val+x-t...'
[04:53] <ohsix> enforced by computer programs :]
[04:53] <Daemon404> ohsix, easy enough to set up a git hook
[04:53] <Daemon404> ;)
[04:53] <ohsix> indeed, and a good way to not let garbage out in a push :]
[04:55] <bryno> ohsix: i've seen that idea mentioned before, but i've never seen it in use
[04:55] <bryno> have an example of that?
[04:58] <Daemon404> libav and ffmpeg (i think) have it right now
[04:58] <Daemon404> rejects trailign whitespace
[04:58] <Daemon404> as an example
[04:59] <bryno> ah
[05:57] <ohsix> bryno: eh you can make the hook reject the push too,  but i meant completely local stuff you can do
[12:10] <ubitux> Daemon404: yes that's my blog post
[12:10] <ubitux> bryno :)
[12:11] <ubitux> Daemon404: the color section is small and i explained it was my opinion
[12:11] <ubitux> (see the first sentence)
[14:51] <ubitux> j-b: if you don't raise regressions we can't guess them :)
[14:51] <j-b> already did raised them
[14:52] <ubitux> we couldn't reproduce them iirc
[14:52] <ubitux> s/them/it/ ?
[14:53] <j-b> no, because this is windows only
[16:33] <PapaSmurf007> informative post ubitux of the hatfield-mccoy feud 
[16:35] <ubitux> :)
[16:37] <ubitux> PapaSmurf007: are you "nothacker"? :p
[16:38] <ubitux> (i'm asking because of http://news.ycombinator.com/item?id=4183533)
[17:53] <PapaSmurf007> no i'm not, but that's where i first read the comparison
[17:53] <PapaSmurf007> ubitux: thought it was fitting
[17:53] <ubitux> ok :)
[18:06] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r9eb6e5cc48 10ffmpeg/libavcodec/wmaenc.c: 
[18:06] <CIA-41> ffmpeg: wmaenc: switch all remaining asserts to av_asserts
[18:06] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:07] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r91ce7c21f0 10ffmpeg/libavcodec/ituh263dec.c: 
[18:07] <CIA-41> ffmpeg: ituh263dec: use av_assert
[18:07] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:13] <ubitux> michaelni: "//FIXME not implemented"  any reason not to use "return AVERROR_PATCHWELCOME" instead of the assert?
[18:17] <michaelni> the assert stops the program and tells one something is missing a return in encode_block() just reruns the encode with a different gain silently
[18:18] <michaelni> btw anyone knows how to compile ffmpeg with android NDK ?
[18:38] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r3bee286ad2 10ffmpeg/libavformat/segment.c: 
[18:38] <CIA-41> ffmpeg: lavf/segment: pass the top level chained muxer format context to segment_end()
[18:38] <CIA-41> ffmpeg: Consistent with segment_start() and less confusing.
[18:38] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * raa1cceb5eb 10ffmpeg/libavformat/segment.c: 
[18:38] <CIA-41> ffmpeg: lavf/segment: rename variable "size" to "list_size"
[18:38] <CIA-41> ffmpeg: The new name is more meaningful/less confusing.
[18:41] Action: ubitux thinks it would be nice to have a sharpen video filter
[18:44] <ubitux> michaelni: wasn't the tools/build_libstagefright specially for that purpose?
[18:44] <saste> anyone interested in reviewing my segment patches?
[18:45] Action: ubitux actually never used the segment thing...
[18:45] <durandal_1707> michaelni: working on fixing h264 extradata overread?
[18:48] <michaelni> durandal_1707, ?
[18:50] <durandal_1707> https://ffmpeg.org/trac/ffmpeg/ticket/1498, i'm not sure but valgrinds complains about extradata...
[18:51] <durandal_1707> but fixing overread could just hide real bug which cause segv (guessing because I do not know that code much)
[19:50] <ubitux> http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=6d4731e1f7d3aef5d3e1c0bbe71974d89529d76f
[19:50] <ubitux> http://lists.libav.org/pipermail/libav-devel/2012-July/030160.html
[19:51] <ubitux> sorry, that was free & mean again :)
[20:42] <durandal_1707> ubitux: better ever than never
[20:42] <ubitux> the problem was actually about the credits
[20:42] <ubitux> but it's most likely a hazard
[20:42] <ubitux> like the previous ones...
[20:44] <durandal_1707> that is hazard for Libav
[20:44] <ubitux> (hazard in the sense of "random" i mean)
[20:50] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r0c9283a03f 10ffmpeg/libavcodec/h264_ps.c: 
[20:50] <CIA-41> ffmpeg: h264: check log2_max_poc_lsb for validity
[20:50] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:50] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r87b9868972 10ffmpeg/libavcodec/h264_refs.c: 
[20:50] <CIA-41> ffmpeg: h264: clear the ref lists when the reference are cleared.
[20:50] <CIA-41> ffmpeg: Fixes reading of freed arrays (Ticket1498)
[20:50] <CIA-41> ffmpeg: Found-by: Piotr Bandurski <ami_stuff at o2.pl>
[20:50] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:09] <CIA-41> ffmpeg: 03Matthieu Bouron 07master * r8ac3868ff3 10ffmpeg/libavformat/aiffdec.c: 
[21:09] <CIA-41> ffmpeg: lavf/aiffdec: handle ID3 chunk
[21:09] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:48] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rc23dcd9751 10ffmpeg/RELEASE: 
[21:48] <CIA-41> ffmpeg: RELEASE: update, seem this has been forgotten
[21:48] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:48] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r23eeffcd48 10ffmpeg/libavformat/mov.c: 
[21:48] <CIA-41> ffmpeg: mov: add workaround for incorrect 0 time point.
[21:48] <CIA-41> ffmpeg: Fixes Ticket1471
[21:48] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:48] <CIA-41> ffmpeg: 03Mans Rullgard 07master * r62634158b7 10ffmpeg/ (7 files in 2 dirs): (log message trimmed)
[22:48] <CIA-41> ffmpeg: ARM: generate position independent code to access data symbols
[22:48] <CIA-41> ffmpeg: This creates proper position independent code when accessing
[22:48] <CIA-41> ffmpeg: data symbols if CONFIG_PIC is set.
[22:48] <CIA-41> ffmpeg: References to external symbols should now use the movrelx macro.
[22:49] <CIA-41> ffmpeg: Some additional code changes are required since this macro may
[22:49] <CIA-41> ffmpeg: need a register to hold the GOT pointer.
[22:49] <CIA-41> ffmpeg: 03Mans Rullgard 07master * r889c1ec4cc 10ffmpeg/libavutil/x86/cpu.c: 
[22:49] <CIA-41> ffmpeg: x86: cpu: clean up check for cpuid instruction support
[22:49] <CIA-41> ffmpeg: This adds macros for accessing the EFLAGS register and uses
[22:49] <CIA-41> ffmpeg: these instead of coding the entire check in inline asm.
[22:49] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[22:49] <CIA-41> ffmpeg: 03Ronald S. Bultje 07master * ra1245d5ca1 10ffmpeg/libavutil/mathematics.h: 
[22:49] <CIA-41> ffmpeg: mathematics: Don't use division by zero in NAN/INFINITY macros
[22:49] <CIA-41> ffmpeg: Some compilers, MSVC among them, don't recognize the divisions by
[22:49] <CIA-41> ffmpeg: zero as meaning infinity/nan.
[22:49] <CIA-41> ffmpeg: These macros should, according to the standard, expand to constant
[22:49] <CIA-41> ffmpeg: expressions, but this shouldn't matter for our usage.
[22:49] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[22:49] <CIA-41> ffmpeg: 03Martin Storsjö 07master * r3893feeccd 10ffmpeg/libavutil/ (eval.c opt.c): 
[22:49] <CIA-41> ffmpeg: opt/eval: Include mathematics.h for NAN/INFINITY
[22:49] <CIA-41> ffmpeg: These files use NAN/INFINITY but didn't include mathematics.h to get
[22:49] <CIA-41> ffmpeg: the fallback definitions if the system lacks the macros.
[22:49] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[22:49] <CIA-41> ffmpeg: 03Martin Storsjö 07master * r710bd8a33e 10ffmpeg/libavcodec/wma.c: 
[22:49] <CIA-41> ffmpeg: wma: Lower the maximum number of channels to 2
[22:49] <CIA-41> ffmpeg: ff_wma_init is used only by wmadec and wmaenc, and neither of them
[22:49] <CIA-41> ffmpeg: can handle more than 2 channels.
[22:50] <CIA-41> ffmpeg: This fixes crashes with invalid files.
[22:50] <CIA-41> ffmpeg: Based on patch by Piotr Bandurski and Michael Niedermayer.
[22:50] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[22:50] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r5eb353e0cd 10ffmpeg/: (log message trimmed)
[22:50] <CIA-41> ffmpeg: Merge remote-tracking branch 'qatar/master'
[22:50] <CIA-41> ffmpeg: * qatar/master:
[22:50] <CIA-41> ffmpeg:  opt/eval: Include mathematics.h for NAN/INFINITY
[22:50] <CIA-41> ffmpeg:  mathematics: Don't use division by zero in NAN/INFINITY macros
[22:50] <CIA-41> ffmpeg:  wma: Lower the maximum number of channels to 2
[22:50] <CIA-41> ffmpeg:  x86: cpu: clean up check for cpuid instruction support
[00:00] --- Mon Jul  2 2012