[Ffmpeg-devel-irc] ffmpeg-devel.log.20111124

Fri Nov 25 02:05:03 CET 2011

[00:16] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r656911d84d 10ffmpeg/libavfilter/vsrc_mandelbrot.c: 
[00:16] <CIA-36> ffmpeg: vsrc_mandelbrot: Fix vsrc_mandelbrot.c:358:85: warning: epsilon may be used uninitialized in this function [-Wuninitialized]
[00:16] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:36] <CIA-36> ffmpeg: 03Alexander Strasser 07master * re77b4b4ac6 10ffmpeg/tests/fate_config.sh.template: 
[00:36] <CIA-36> ffmpeg: tests: add a FATE configuration template
[00:36] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:37] <CIA-36> ffmpeg: 03Alexander Strasser 07master * r7ffa9ea05a 10ffmpeg/doc/fate.txt: 
[00:37] <CIA-36> ffmpeg: doc: rewrite FATE documentation to be self-contained
[00:37] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:38] <CIA-36> ffmpeg: 03Mans Rullgard 07master * r3a0b72dee0 10ffmpeg/libavcodec/arm/ (9 files): 
[03:38] <CIA-36> ffmpeg: ARM: remove needless .text/.align directives
[03:38] <CIA-36> ffmpeg: The 'function' macro already includes the appropriate
[03:38] <CIA-36> ffmpeg: directives.
[03:38] <CIA-36> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[03:38] <CIA-36> ffmpeg: 03Anton Khirnov 07master * r50efd3cad8 10ffmpeg/ (libavformat/md5enc.c tests/fate/vp8.mak): 
[03:38] <CIA-36> ffmpeg: md5enc: add flags
[03:38] <CIA-36> ffmpeg: AVFMT_NOTIMESTAMPS for md5, as it ignores the timestamps.
[03:38] <CIA-36> ffmpeg: AVFMT_VARIABLE_FPS for framemd5, as it prints dts.
[03:38] <CIA-36> ffmpeg: -vsync 0 for the vp8 test is needed because with vsync 2 the timestamp
[03:38] <CIA-36> ffmpeg: guessing code gets confused by an altref frame that is never displayed
[03:38] <CIA-36> ffmpeg: and drops a frame later.
[03:38] <CIA-36> ffmpeg: 03Anton Khirnov 07master * ra94c769dad 10ffmpeg/libavcodec/tiffenc.c: tiffenc: add a private option for selecting compression algorithm
[03:38] <CIA-36> ffmpeg: 03Anton Khirnov 07master * r0844b57c11 10ffmpeg/ (12 files in 3 dirs): (log message trimmed)
[03:38] <CIA-36> ffmpeg: crcenc: add flags
[03:38] <CIA-36> ffmpeg: AVFMT_NOTIMESTAMPS for crc, as it ignores the timestamps.
[03:38] <CIA-36> ffmpeg: AVFMT_VARIABLE_FPS for framecrc, as it prints dts.
[03:38] <CIA-36> ffmpeg: Many FATE changes, because avconv is no longer duplicating frames in
[03:38] <CIA-36> ffmpeg: those tests.
[03:38] <CIA-36> ffmpeg: Also added -vsync 0 for some tests to prevent avconv from dropping
[03:38] <CIA-36> ffmpeg: 03Anton Khirnov 07master * raf8ad892bf 10ffmpeg/avconv.c: 
[03:38] <CIA-36> ffmpeg: avconv: use the same behavior when decoding audio/video/subs
[03:38] <CIA-36> ffmpeg: I.e. if the packet was only partially consumed, pass the rest of it into
[03:39] <CIA-36> ffmpeg: 03Justin Ruggles 07master * rbbb46f3ec7 10ffmpeg/ (4 files in 2 dirs): 
[03:39] <CIA-36> ffmpeg: libavutil: add utility functions to simplify allocation of audio buffers.
[03:39] <CIA-36> ffmpeg: Based on code by Stefano Sabatini.
[03:39] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r8e576d5830 10ffmpeg/: (log message trimmed)
[03:39] <CIA-36> ffmpeg: Merge remote-tracking branch 'qatar/master'
[03:39] <CIA-36> ffmpeg: * qatar/master:
[03:39] <CIA-36> ffmpeg:  libavutil: add utility functions to simplify allocation of audio buffers.
[03:39] <CIA-36> ffmpeg:  libavutil: add planar sample formats and av_sample_fmt_is_planar()
[03:39] <CIA-36> ffmpeg:  avconv: fix segfault at EOF with delayed pictures
[03:39] <CIA-36> ffmpeg:  pcmdec: remove unneeded resetting of samples pointer
[03:39] <CIA-36> ffmpeg: 03Anton Khirnov 07master * re1edfbcb24 10ffmpeg/avconv.c: 
[03:40] <CIA-36> (7 lines omitted)
[03:46] <ubitux> michaelni: do you have plans for libswresample with the new introduced planar sample formats?
[03:47] <ubitux> i wonder if a packed ’ planar mapping function won't be welcome
[03:50] <michaelni> ubitux, swresample fully supports packed & planar
[03:50] <ubitux> yes i know, but with the +0x100 hack
[03:50] <ubitux> (it is even in the doxy)
[03:51] <ubitux> so i was wondering about using the real samples formats instead
[03:51] <ubitux> since they just got added
[03:52] <michaelni> swr should be made to support the "new" sample formats of course
[03:52] <michaelni> if it should use them internally too, i dont know, that depends on which way it ends cleaner
[03:53] <michaelni> but it doesnt really matter much
[03:53] <michaelni> which way its done
[03:53] <michaelni> as its just internal
[03:54] <michaelni> a optimized packed<->planar function is of course welcome and trivial to write in the unlikely case theres none to borrow laying aorund
[03:56] <ubitux> i'm going to try to update swr to those new planar formats
[03:56] <ubitux> in case you don't want to do it instead
[03:56] <michaelni> no no, you can do the 4 line patch :)
[03:57] <michaelni> if(>4) |= 0x100 :)
[03:57] <ubitux> haha
[03:57] <ubitux> i meant doing it "the right way" ©®"
[03:59] <michaelni> you mean:
[03:59] <michaelni> -s-> in.planar= s-> in_sample_fmt >= 0x100;
[03:59] <michaelni> +s-> in.planar= s-> in_sample_fmt >4;
[03:59] <michaelni> ?
[03:59] <ubitux> -    s-> in.planar= s-> in_sample_fmt >= 0x100;
[03:59] <ubitux> -    s->out.planar= s->out_sample_fmt >= 0x100;
[03:59] <ubitux> +    s-> in.planar= av_sample_fmt_is_planar(s-> in_sample_fmt);
[03:59] <ubitux> +    s->out.planar= av_sample_fmt_is_planar(s->out_sample_fmt);
[04:00] <ubitux> but also update the doxy etc
[04:00] Action: michaelni knew it would be 4 lines :)))
[04:00] <ubitux> haha that's not the only thing :)
[04:00] <ubitux> i have to add the packed<->planar helper you mentionned too
[04:00] <ubitux> not much work obviously, but well
[04:00] <ubitux> just a few stuff to care about
[05:55] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r957867ab13 10ffmpeg/ (ffmpeg.c tests/ref/lavf/gxf): (log message trimmed)
[05:55] <CIA-36> ffmpeg: ffmpeg: rewrite vsync / notimestamps handling
[05:55] <CIA-36> ffmpeg: The qatar implementation makes no sense.
[05:55] <CIA-36> ffmpeg: a muxer without timestamps is constant fps thus needs vsync.
[05:55] <CIA-36> ffmpeg: the crc/mp5 are special cases that have timestamps yet allow any
[05:55] <CIA-36> ffmpeg: nonsensical timestamps.
[05:55] <CIA-36> ffmpeg: raw (yuv/rgb) video is constant fps thus needs vsync too.
[14:24] <CIA-36> ffmpeg: 03Sergiy Gur'yev 07master * r32ed7da135 10ffmpeg/libavcodec/libaacplus.c: 
[14:24] <CIA-36> ffmpeg: Fix adts format creation in aac+ encoder modified: libavcodec/libaacplus.c
[14:24] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:54] <CIA-36> ffmpeg: 03Sergiy Gur'yev 07oldabi * r71ef1e494f 10ffmpeg/libavcodec/libaacplus.c: 
[14:54] <CIA-36> ffmpeg: Fix adts format creation in aac+ encoder modified: libavcodec/libaacplus.c
[14:54] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:54] <CIA-36> ffmpeg: (cherry picked from commit 32ed7da1350e551ec005b75e482da74f2e93fbb9)
[14:54] <CIA-36> ffmpeg: 03Sergiy Gur'yev 07release/0.7 * r47b5fabd6a 10ffmpeg/libavcodec/libaacplus.c: 
[14:54] <CIA-36> ffmpeg: Fix adts format creation in aac+ encoder modified: libavcodec/libaacplus.c
[14:54] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:54] <CIA-36> ffmpeg: (cherry picked from commit 32ed7da1350e551ec005b75e482da74f2e93fbb9)
[14:54] <CIA-36> ffmpeg: 03Sergiy Gur'yev 07release/0.8 * r4f58d8ebc1 10ffmpeg/libavcodec/libaacplus.c: 
[14:54] <CIA-36> ffmpeg: Fix adts format creation in aac+ encoder modified: libavcodec/libaacplus.c
[14:54] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:54] <CIA-36> ffmpeg: (cherry picked from commit 32ed7da1350e551ec005b75e482da74f2e93fbb9)
[15:14] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r58030fc6c8 10ffmpeg/libavformat/avio.c: 
[15:14] <CIA-36> ffmpeg: avio: Fix handling of filenames that contain :
[15:14] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:24] <j-b> Hello
[15:47] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r2bf09826c1 10ffmpeg/libavcodec/loco.c: 
[15:47] <CIA-36> ffmpeg: loco: Fix error handling.
[15:47] <CIA-36> ffmpeg: Fixes null pointer dereference / http://www.google-melange.com/gci/task/view/google/gci2011/7120335
[15:47] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:12] <ubitux> mmh i wonder if instead of the altform for planar we could just assume something like minus/plus AV_SAMPLE_FMT_NB/2
[16:33] <michaelni> ubitux, assuming that the matching formats are at equal distance works as long as nothing is added
[16:33] <ubitux> yes
[16:33] <michaelni> my system that used +256 would have ensured that more solidly ...
[16:34] <ubitux> actually if the "planar" flag was find that way (> AV_SAMPLE_FMT_NB/2) i would have do it
[16:34] <ubitux> but since there is an explicit flag, let's keep it explicit 
[16:34] <ubitux> that doesn't matter much imo :p
[16:34] <michaelni> yes
[16:36] <ubitux> fyi i liked the |0x100 hack :p
[16:37] <ubitux> it would be nice if someone could update the audio filters to use swr&
[16:38] <ubitux> and maybe we could also update avconv so we could get rid of the old API
[16:40] <j-b> michaelni: you can subscribe to GCi, with ubitux, you know
[16:40] <j-b> and suggest tasks.
[16:40] <michaelni> j-b, really ? i didnt know
[16:40] <ubitux> GCi? Google contest...?
[16:40] <j-b> Google COde In
[16:42] <ubitux> i'm not sure i want to :)
[16:42] <ubitux> it seems to have started 3 days agp
[16:42] <ubitux> ago*
[16:43] <j-b> I'm not sure either
[16:43] <ubitux> "The contest is open to all students who are at least thirteen (13) years of age and no older than seventeen (17) years of age on November 21, 2011."
[16:43] <ubitux> i'm much older too :p
[16:46] <michaelni> ubitux, if you figure out how to subscribe as a task suggest monkey, ill try to subscribe too later today probably
[16:47] <michaelni> i dont have time ATM to look at it sadly ...
[16:50] <j-b> it takes a lot of time
[16:51] <ubitux> i don't have much neither :p
[16:52] <ubitux> the main site does not seem to even mention the "task suggest monkey"
[16:52] <ubitux> everything is about the students
[16:56] <CIA-36> ffmpeg: 03Clément BSsch 07master * rafb0470a25 10ffmpeg/libswresample/swresample.c: 
[16:56] <CIA-36> ffmpeg: swr: fix test program.
[16:56] <CIA-36> ffmpeg: The number of used channels need to be reset to zero when
[16:56] <CIA-36> ffmpeg: swr_alloc_set_opts() is called successive times.
[16:56] <CIA-36> ffmpeg: This is a regression introduced in c8136ebd.
[16:56] <CIA-36> ffmpeg: 03Clément BSsch 07master * r573ffbb3b5 10ffmpeg/ (4 files in 2 dirs): lavu: add av_get_alt_sample_fmt().
[16:56] <CIA-36> ffmpeg: 03Clément BSsch 07master * r58205edcc6 10ffmpeg/libswresample/ (swresample.c swresample.h swresample_test.c): swr: use native planar sample formats.
[16:56] <CIA-36> ffmpeg: 03Clément BSsch 07master * r0232f788b6 10ffmpeg/libswresample/swresample.c: 
[16:56] <CIA-36> ffmpeg: swr: fix invalid sample format message.
[16:56] <CIA-36> ffmpeg: If the sample format is invalid, av_get_sample_fmt_name() will always
[16:56] <CIA-36> ffmpeg: return NULL.
[17:43] <ubitux> khanacademy is really missing signal processing and related lessons :(
[17:44] <av500> there's books on that
[18:08] <michaelni> j-b, if theres a way for me to add any task, join or otherwise, ive not found it, the FAQ and everything is just about students
[18:11] <j-b> signup
[18:11] <j-b> ask for "videolan" join
[18:11] <j-b> and use bulk upload
[18:13] <michaelni> iam on:http://www.google-melange.com/gci/org/google/gci2011/videolan
[18:13] <michaelni> and theres a logout so i assume iam loged in
[18:13] <michaelni> j-b, theres no join and no bulk upload i can see
[18:14] <j-b> did you create your profile?
[18:14] <j-b> http://www.google-melange.com/gsoc/profile/google/gci2011
[18:15] <michaelni> that url gives me "There is no program for url 'google/gci2011' "
[18:17] <michaelni> j-b, iam getting "This page is inaccessible because you do not have a profile in the program at this time."
[18:18] <ubitux> av500: yeah i know, but it's always better when you have someone to explain concepts
[18:18] <michaelni> j-b, the faq says i can register once ive completed a task ;)
[18:19] <j-b> michaelni: that is for students
[18:19] <michaelni> j-b, i know
[18:37] <michaelni> j-b, ive created a profile for gsoc2011 (and can now reach to the request to become a mentor for gci but when i click on that i get "This page is inaccessible because you do not have a profile in the program at this time."
[18:39] <michaelni> theres also a "My incoming invitation" link that similarly ends at "This page is inaccessible because you do not have a profile in the program at this time."
[18:47] <michaelni> j-b, i succeeded asking videolan to join, this web interface is a nightmare
[18:47] <j-b> you are poliet.
[18:47] <j-b> http://www.google-melange.com/gci/dashboard/google/gci2011#bulk_create_tasks
[18:49] <michaelni> j-b, seems working, thanks alot, now i need to think about what tasks to create, any special rules i need to follow ?
[18:50] <j-b> those are students, in high-school. What you think as easy code is probably hard.
[18:50] <ubitux> it does not need to be code, right?
[18:50] <j-b> nope
[18:51] <j-b> michaelni: oh, and be careful, the bulk upload is case-sensitive.
[19:34] <michaelni> j-b, repeatable tasks should be submitted, multiple times ?
[19:38] <michaelni> ubitux, Compn is the following description good? :
[19:38] <michaelni> Go to http://ffmpeg.org/trac/ffmpeg/report/1, find a bug, reproduce it, and fix it. This requires some good C coding skills, definitely above the average of students, but if you are good you can earn many points as there are many bugs.
[19:38] <michaelni> kierank, are you available as gci mentor for ffmpeg ?
[19:39] <ubitux> michaelni: is it possible to add filters to the report list (or a custom query)?
[19:39] <ubitux> so we could remove the already assigned ones
[19:39] <michaelni> ubitux, trac is pretty powerfull, but the assignment is random
[19:40] <michaelni> trac assigns some stuff by default ...
[19:40] <michaelni> so we can filter by unassigned but i dont think its a good idea
[19:40] <ubitux> also imho you should point out to urls: "make yourself comfortable with codebase fixing misc bugs (http://...) and then you might be interested in writing new features (http://...)"
[19:40] <kierank> michaelni: not got time to do anything right now
[19:41] <ubitux> s/to urls/two urls/
[19:42] <michaelni> ubitux, can you write the whole description for me so i only have to copy and paste :) ?
[19:42] <michaelni> i would add typos otherwise ;)
[19:48] <ubitux> michaelni: then maybe: "A good way to be comfortable with the code of a project is to start fixing various bug: go to http://ffmpeg.org/trac/ffmpeg/query?status=new&status=open&status=reopened&type=defect&col=id&col=summary&col=status&col=type&col=owner&col=priority&col=component&order=priority, find a bug you'd like to fix, reproduce it, and fix it. This requires some good C coding skills, definitely above the average of ...
[19:48] <ubitux> ... students, but if you are good you can earn many points as there are many bugs. Then you might be interested in writing new features, so you might want to check http://ffmpeg.org/trac/ffmpeg/query?status=new&status=open&status=reopened&type=enhancement&col=id&col=summary&col=status&col=type&col=owner&col=priority&col=component&order=priority for a large panel of choice."
[19:48] <ubitux> bugs*
[19:48] <ubitux> feel free to fix the bad english wording, and my typos ;)
[19:50] <michaelni> ubitux, time to complete 24h ok ? ;)
[19:50] <michaelni> this isnt a very realistic task for students anyway i think
[19:50] <cbsrobot> when will gci be finished ?
[19:51] <ubitux> arh it's 24h per task?
[19:51] <michaelni> it can be multiple days
[19:52] <ubitux> i think i just missed the point then, you original proposition should be fine then
[19:52] <cbsrobot> I suggest documenting would be a nice task or doxy
[19:52] <cbsrobot> lets say an easy task
[19:52] <michaelni> ubitux, you think my desription is better ?
[19:52] <ubitux> yes
[19:53] <michaelni> cbsrobot, please register and add tasks yourself i cant do it all the web interface is unfunny
[19:54] <michaelni> atm i mainly try to create 1 task to see how it works ...
[19:58] <michaelni> i created the first but it does not show up in the list :/
[20:17] <michaelni> i created 2 tasks, j-b what was it with that bulk upload you mentioned? i see nothing thats called bulk upload
[20:36] <ubitux> it seems calling avfilter_unref_buffer on both inpicref and outpicref solves the huge boxblur memleak
[20:36] <ubitux> but i have absolutely no idea why we need to do this
[20:58] <michaelni> hey iive do you want to mentor google code in tasks ?
[20:58] <iive> no
[21:01] <michaelni> do you want to help maintain some code in ffmpeg ? ther are alot of areas that needs a maintainer ?
[21:03] <michaelni> iive, also did you see "[RFC] Talk about subtitles", i think you where active in the last subtitle discussion ...
[21:03] <iive> I was, but I can't find time to research the subject again.
[21:04] <Compn> michaelni : your description sounds fine to me
[21:04] Action: Compn afk
[21:05] <michaelni> iive, ok
[21:41] <cbsrobot> michaelni: there are still some styling in 10-bit openjpegdec.c but this should not stop getting it pushed
[21:42] <michaelni> cbsrobot, thx alot for the review! :)
[21:43] <cbsrobot> np - btw please don't add me to maintainers ... j2k is such a mess.
[21:44] <michaelni> i know, ive worked a bit on j2k too ...
[21:45] <CIA-36> ffmpeg: 03Carl Eugen Hoyos 07master * r0a4f02117b 10ffmpeg/ (avconv.c doc/avconv.texi doc/ffmpeg.texi ffmpeg.c): Add option -n to exit if output file exists.
[21:51] <CIA-36> ffmpeg: 03Alex Zhukov 07master * reb511ef636 10ffmpeg/libavcodec/libopenjpegdec.c: 
[21:51] <CIA-36> ffmpeg: yuv pixel formats support in openjpeg decoder + 10bit support
[21:51] <CIA-36> ffmpeg: Signed-off-by: Alex Zhukov <zhukov.alex at gmail.com>
[21:51] <CIA-36> ffmpeg: Reviewed-by: Jean First <jeanfirst at gmail.com>
[21:51] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:48] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * re421b79d01 10ffmpeg/libavcodec/indeo3.c: 
[22:48] <CIA-36> ffmpeg: indeo3: fix null pointer dereference in vulnerable.mov
[22:48] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:00] <Nagy> how far along is avformat_seek_file?
[23:01] <michaelni> hasnt moved lately
[23:01] <Nagy> is it advisable to use it?
[23:03] <nevcairiel> there is no advantage in using it, it internally basically falls back to av_seek_frame
[23:03] <nevcairiel> considering that, there is also no downside
[23:04] <Nagy> ic, thx
[00:00] --- Fri Nov 25 2011