[Ffmpeg-devel-irc] ffmpeg-devel.log.20120728

[burek]

[00:10] <CIA-41> ffmpeg: 03Samuel Pitoiset 07master * r2357f60687 10ffmpeg/libavformat/rtmpproto.c: 
[00:10] <CIA-41> ffmpeg: rtmp: Prevent reading outside of an allocate buffer when receiving server bandwidth packets
[00:10] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[00:10] <CIA-41> ffmpeg: 03Samuel Pitoiset 07master * re49e6b6451 10ffmpeg/libavformat/rtmpproto.c: 
[00:10] <CIA-41> ffmpeg: rtmp: Allow having more unknown data at the end of a chunk size packet without failing
[00:10] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[00:10] <CIA-41> ffmpeg: 03Samuel Pitoiset 07master * r8ea1459bc3 10ffmpeg/libavformat/rtmpproto.c: 
[00:10] <CIA-41> ffmpeg: rtmp: Check the buffer length of ping packets
[00:10] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[00:10] <CIA-41> ffmpeg: 03Ronald S. Bultje 07master * r30b45d9c38 10ffmpeg/ (6 files in 3 dirs): x86inc: automatically insert vzeroupper for YMM functions.
[00:10] <CIA-41> ffmpeg: 03Diego Biurrun 07master * r8784959234 10ffmpeg/ (configure libavcodec/Makefile): build: Rename aandct component to more descriptive aandcttables
[00:10] <CIA-41> ffmpeg: 03Ronald S. Bultje 07master * r02ac28229a 10ffmpeg/libavutil/eval.c: 
[00:10] <CIA-41> ffmpeg: eval: fix printing of NaN in eval fate test.
[00:10] <CIA-41> ffmpeg: This fixes "make fate-eval" on MSVC builds. Without this, the test outputs
[00:10] <CIA-41> ffmpeg: "-1.#NaN" instead of "nan" on MSVS 2010.
[00:10] <CIA-41> ffmpeg: 03Mans Rullgard 07master * rb9d3c37848 10ffmpeg/ (Makefile common.mak doc/Makefile tests/Makefile): 
[00:10] <CIA-41> ffmpeg: build: use COMPILE template for HOSTOBJS
[00:10] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[00:11] <CIA-41> ffmpeg: lavfi: better channel layout negotiation
[00:11] <CIA-41> ffmpeg: Allow substitution of channel pairs in the input for nearby channel pairs in
[00:11] <CIA-41> ffmpeg: the output in order to get a closer match. Also weigh LFE channel mismatch
[00:11] <CIA-41> ffmpeg: differently to favor matching the same layout without LFE over one less
[00:11] <CIA-41> ffmpeg: channel with LFE.
[00:11] <CIA-41> ffmpeg: 03Justin Ruggles 07master * r81f548de57 10ffmpeg/libavcodec/alac.c: 
[00:11] <CIA-41> ffmpeg: alac: check for truncated packets
[00:11] <CIA-41> ffmpeg: This will give a clearer error message when the error is caused by a
[00:11] <CIA-41> ffmpeg: truncated packet.
[00:11] <CIA-41> ffmpeg: 03Ronald S. Bultje 07master * r5361e10a5e 10ffmpeg/libavcodec/x86/proresdsp.asm: proresdsp: port x86 assembly to cpuflags.
[00:11] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rc6963a220d 10ffmpeg/: (log message trimmed)
[00:12] <CIA-41> ffmpeg: Merge remote-tracking branch 'qatar/master'
[00:12] <CIA-41> (10 lines omitted)
[01:47] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r0fa352623d 10ffmpeg/libavcodec/dvdsubenc.c: 
[01:47] <CIA-41> ffmpeg: dvdsubenc: switch to av_assert
[01:47] <CIA-41> ffmpeg: we use av_assert0() because the previous code unconditionally
[01:47] <CIA-41> ffmpeg: enabled the assert.
[01:47] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:47] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r55cdd45446 10ffmpeg/libavformat/avio.h: 
[01:47] <CIA-41> ffmpeg: avio: Document that write_packet() is not allowed to change its input
[01:47] <CIA-41> ffmpeg: Found-by: Reimar
[01:47] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:52] <cptspiff> hmm, ^^ gut reaction, wouldn't const'ing the input tell the same story and have the compiler enforce as a bonus?
[01:55] <michaelni> cptspiff, yes but its public API making it const might (but of course doesnt have to) cause problems
[01:55] <michaelni> the docs change was the easy way
[01:56] <michaelni> if someone wants to change it and double check that it has no sideeffects thats welcome of course
[03:11] <cptspiff> the extent of tests i can do right now is compile. on a 2g link, not grabbing fate over it ;) it compiles, diff old.log new.log is blank so no new warnings emitted or such. obviously being public, i cannot guarantee user code, but handwaving i'd say going from a writable to a non-writable ptr should not cause much havoc there, other than some warnings about incompat ptrs (and possibly errors in c++). don't have a tool to compare binary files here
[03:13] <cptspiff> (by incompat ptrs i refer to the function ptrs)
[03:26] <michaelni> cptspiff, it could error with the right compiler flags, and as said i dont mind at all if its changed i just didnt want to do it as a quick warning fix without thought
[05:39] <CIA-41> ffmpeg: 03jamal 07master * r64c53fa16a 10ffmpeg/libavcodec/ (dxva2_h264.c dxva2_mpeg2.c dxva2_vc1.c): 
[05:39] <CIA-41> ffmpeg: dxva2: Fix some warnings about incompatible pointer type
[05:39] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:39] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * ra2f91e7bcd 10ffmpeg/libavcodec/mpegaudiodec.c: 
[05:39] <CIA-41> ffmpeg: mpegaudiodec: switch to av_assert
[05:39] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[09:40] <ubitux> Daemon404: in your canopus patch, i think michael wanted you to explicit the codec id with a MKBETAG
[10:47] <maker> I was wondering, since the task of porting vlc filters to libavfilter has been in the todolist for contributors for a few years, has anybody ported anything since now?
[10:50] <ubitux> vf_blackframe.c: * Ported from MPlayer libmpcodecs/vf_blackframe.c.
[10:50] <ubitux> vf_boxblur.c: * Ported from MPlayer libmpcodecs/vf_boxblur.c.
[10:50] <ubitux> vf_cropdetect.c: * Ported from MPlayer libmpcodecs/vf_cropdetect.c.
[10:50] <ubitux> vf_delogo.c: * Ported from MPlayer libmpcodecs/vf_delogo.c.
[10:50] <ubitux> and a few other ones
[10:52] <maker> Puzzle seems not to have been ported \o/ 
[10:52] <saste> ubitux: from *vlc*
[10:53] <maker> Well, I did git grep "Ported from" , the pattern is the same
[10:53] <saste> "Turns the video in a jigsaw puzzle game"
[10:54] <saste> we have a tile filter but seems pretty a different thing
[10:54] <ubitux> saste: argh sorry misread.
[10:55] <maker> :D
[10:56] <ubitux> maker: according to Daemon404 they have a good ivtc filter that would be welcome in ffmpeg
[10:56] <ubitux> but that's not trivial
[10:57] <maker> ubitux: I don't think you are going to jugde students for having accomplished something, what if I create a new repo somewhere and develop on ivtc as qualification task? 
[10:58] <ubitux> i'm not going to judge anyone ;)
[10:58] <saste> maker: that's ok of course, but for review it's easier to send a patch anyway
[10:59] <ubitux> maker: you're still looking for a filter to do?
[10:59] <maker> as god" said
[10:59] <ubitux> (maker: are you the epita guy?)
[10:59] <maker> ubitux: yay.
[10:59] <maker> epita? Probably not
[10:59] <ubitux> oh ok, my bad then
[10:59] <maker> saste: yep, just in case there won't be enough time for submitting a patch
[11:00] <ubitux> i'm wondering if porting the vf geq wouldn't be a good exercise
[11:00] <ubitux> i wanted to do it for a while but didn't have the time
[11:00] <ubitux> (this is from mplayer again, not vlc)
[11:00] <maker> geq? uhm
[11:00] <saste> ubitux: i seems to remember having started them, but there were some design issues
[11:01] <saste> the way it dealt with RGB/YUV was borken iirc
[11:01] <ubitux> saste: "them"?
[11:01] <ubitux> vlc ones?
[11:01] <saste> ubitux: them => geq
[11:01] <ubitux> ok
[11:02] <ubitux> iirc the main issue was how to deal with reading the values
[11:03] <ubitux> but a vf geq without the read luma & chroma feature could be a first
[11:04] <ubitux> well, just a suggestion :)
[11:05] <maker> mulitmedia.cx is down?
[11:05] <ubitux> i just saw pretty interesting geq expression a while ago, and thought that was a nice filter
[11:05] <maker> :o
[11:05] <ubitux> looks like so
[11:05] <ubitux> fate as well :(
[11:16] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rfd2b6c1e4b 10ffmpeg/doc/filters.texi: doc/filters: itemize movie examples
[11:17] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r66a5201819 10ffmpeg/ (7 files in 3 dirs): lavfi: add flite audio source
[11:17] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r7cf9aaddcd 10ffmpeg/libavutil/bprint.h: 
[11:17] <CIA-41> ffmpeg: lavu/bprint: extend/clarify documentation for av_bprint_init() size_max value
[11:17] <CIA-41> ffmpeg: Also define AV_BPRINT_SIZE_* macros before av_bprint_init() declaration,
[11:17] <CIA-41> ffmpeg: and mention them in the av_bprint_init() doxy.
[11:17] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r32390d6767 10ffmpeg/ (4 files in 2 dirs): 
[11:17] <CIA-41> ffmpeg: lavfi: move movie and amovie to a "multimedia sources" section
[11:17] <CIA-41> ffmpeg: Since the recent changes, movie and amovie are able to deal with more
[11:17] <CIA-41> ffmpeg: than one type of stream, so they should be categorized as "multimedia
[11:17] <CIA-41> ffmpeg: sources" rather than audio/video sources.
[11:49] <CIA-41> ffmpeg: 03Nicolas George 07master * r9a757cca34 10ffmpeg/libavfilter/avf_concat.c: avf_concat: fix min value for v option.
[12:31] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rfb0688b079 10ffmpeg/doc/general.texi: doc/general: add missing device support information
[13:32] <vitrums> Hi there. I'm looking for some to solve my problem of screencasting via rtmp (justin.tv, own3d), when I always get black screen as an output. However local recordings work just perfect. Any ideas?
[13:33] <vitrums> for some help*
[15:30] <CIA-41> ffmpeg: 03Nicolas George 07master * r4ce87ecf2a 10ffmpeg/ (doc/filters.texi libavfilter/asrc_flite.c): (log message trimmed)
[15:30] <CIA-41> ffmpeg: asrc_flite: do not crash on multiple instances.
[15:30] <CIA-41> ffmpeg: The voice register functions return the same voice structure
[15:30] <CIA-41> ffmpeg: upon multiple registration. It causes us two problems:
[15:30] <CIA-41> ffmpeg: If we delete a voice without deregistering it, it leaves
[15:30] <CIA-41> ffmpeg: a dangling pointer inside the library.
[15:30] <CIA-41> ffmpeg: If we delete or unregister a voice at uninit, it may still
[15:31] <CIA-41> ffmpeg: 03Nicolas George 07master * refcfcf8793 10ffmpeg/libavfilter/avf_concat.c: 
[15:31] <CIA-41> ffmpeg: avf_concat: fix silence duration.
[15:31] <CIA-41> ffmpeg: concat_in.pts starts at 0 for each segment, so the duration
[15:31] <CIA-41> ffmpeg: of the silence was too long, except at the first stitch.
[15:31] <saste> michaelni: is it OK to extend lavu/integer in order to support bigger numbers, or to make the max value configurable at init?
[15:32] <saste> i need it for xface, so I don't need to duplicate big integer code
[15:33] <michaelni> saste, sure
[15:37] <cptspiff> michaelni: right, then i misunderstood what you meant (and you can see i struggled to figure out what you meant double check it doesnt break anything ;))
[16:34] <durandal_1707> why is fate down?
[17:06] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r05a6c625f4 10ffmpeg/tests/Makefile: 
[17:06] <CIA-41> ffmpeg: fate: switch rsync to fate-suite.ffmpeg.org
[17:06] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:06] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r157f956973 10ffmpeg/ffmpeg.c: 
[17:06] <CIA-41> ffmpeg: ffmpeg: add code to perform ts wrap correction
[17:06] <CIA-41> ffmpeg: Fixes ticket1385
[17:06] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:52] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r3adc9c9996 10ffmpeg/ffmpeg.c: 
[17:53] <CIA-41> ffmpeg: ffmpeg: allow using the r_frame_rate as fundamental timebase for stream copy in avi
[17:53] <CIA-41> ffmpeg: Fixes Ticket1434
[17:53] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:05] <cptspiff> hmm i think i'm starting to see why http://trac.ffmpeg.org/trac/ffmpeg/ticket/604 sticks around
[18:06] <nevcairiel> my guess is mostly because it doesnt affect the ffmpeg apps, only third party apps
[18:06] <cptspiff> do i read this correctly that i's pretty much impossible to solve without a the painful reopen cycle?
[18:06] <cptspiff> well it's an arch issue
[18:06] <cptspiff> and it really hurts in these evil 10-bit animeidiot days ;)
[18:07] <nevcairiel> thinking about it, i guess the decoder really would have a hard time changing its own threads value
[18:08] <cptspiff> yeah, plus the hwaccel isnt really setup before the entire chain is setup (decode_slice_header for h264 afaict) so you cannot detect it up front and hop to 1 thread.
[18:08] <nevcairiel> i just disable threading when dxva is requested, and since my dxva codepath is completely seperate from the software decoding, i do a full codec re-open anyway
[18:08] <cptspiff> yeah we do the same
[18:08] <nevcairiel> but i guess if you want to keep it simpler, you might not have that luxury
[18:08] <cptspiff> but the full codec re-open hurts on streams.
[18:09] <cptspiff> (in particular volatile urls and such crap)
[18:11] <nevcairiel> when i have the h264 sequence headers present i manually scan those for hw compat to try to avoid that, but in streaming situations sadly those arent always available, so there is that
[18:13] <cptspiff> yup. it's the one size fits all solution that is hard.
[18:14] <nevcairiel> although you need the SPS before you can start decoding, and once you have it, you know if its hw compatible
[18:24] <CIA-41> ffmpeg: 03Nicolas George 07master * r0f8f924847 10ffmpeg/doc/faq.texi: faq: suggest the concat filter.
[18:24] <CIA-41> ffmpeg: 03Nicolas George 07master * r2d6f442ca1 10ffmpeg/doc/faq.texi: faq: disambiguate "join".
[20:04] <maker> Uhm, is the documentation present in multimedia.cx concerning libavfilter up to date?
[20:05] <maker> Becouse I see some variation on the constraints, but maybe apis and logis is stil the same
[20:05] <maker> *because
[20:05] <maker> *logics are
[20:05] <maker> Sorry for that. :|
[20:08] <michaelni> maker, best look at the source code in ffmpeg git (various headers and some simple filters) as documentation
[20:09] <michaelni> i dunno how upto date multimedia.cx is
[20:10] <maker> There isn't any doxygen somewhere?
[20:20] <michaelni> maker, http://ffmpeg.org/doxygen/trunk/index.html
[20:22] <maker> ty
[00:00] --- Sun Jul 29 2012