[Ffmpeg-devel-irc] ffmpeg-devel.log.20120326

Tue Mar 27 02:05:04 CEST 2012

[00:17] <durandal_1707> nice you can make openjpeg encoding/decoding faster at expense of inrease in file size
[00:59] <durandal_1707> michaelni: gray16 -> anything is broken in swscale
[01:02] <durandal_1707> or not, looks like strange issue in openjpeg decoder
[01:04] <durandal_1707> yup, it is definetly bug in libopenjpeg decoder
[01:06] <CIA-17> ffmpeg: 03Aaron Colwell 07master * raa9c2e41b6 10ffmpeg/libavcodec/pthread.c: 
[01:06] <CIA-17> ffmpeg: pthread : Remove lock/unlock pairs in worker loop to avoid unexpected state changes.
[01:06] <CIA-17> ffmpeg: Reviewed-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[01:06] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:06] <CIA-17> ffmpeg: 03Paul B Mahol 07master * r25b67dc765 10ffmpeg/libavcodec/pngdec.c: 
[01:06] <CIA-17> ffmpeg: png: fix warning about discarding qualifiers [-Wincompatible-pointer-types]
[01:06] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[01:06] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:06] <CIA-17> ffmpeg: 03Paul B Mahol 07master * r40821980ae 10ffmpeg/ (libavfilter/vf_hflip.c tests/ref/lavfi/pixfmts_hflip): 
[01:06] <CIA-17> ffmpeg: vf_hflip: accept PIX_FMT_RGB444 and PIX_FMT_BGR444 pixel formats
[01:06] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[01:06] <CIA-17> ffmpeg: added fate checksum update
[01:06] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:26] <durandal_1707> i get segv for this sample: http://samples.ffmpeg.org/jpeg2000/jpeg2000_mxf_first_10mb.mxf
[02:34] <durandal_1707> michaelni: av_pix_fmt_descriptors have holes and they are not filled with 0
[02:42] <durandal_1707> michaelni: shouldn't hflip work for any pix_fmt like vflip?
[02:45] <michaelni> YUYV shoudl fail
[02:50] <durandal_1707> so jpeg2000 stream in mxf is not supported?
[02:57] <michaelni> hflip should fail with YUYV, i dunno about j2k-mxf
[02:57] <durandal_1707> got hflip part already ;)
[02:58] <durandal_1707> j2k in mxf is not supported, and sample i mentioned is causing segv
[02:59] <michaelni> i can confirm that
[02:59] <durandal_1707> libav does not segv
[03:01] <michaelni> ill fix the segv
[03:05] <michaelni> segv fixed, will be in my next push
[03:07] <michaelni> and libav does crash on the file
[03:07] <durandal_1707> not for me
[03:07] <michaelni> ==21862== ERROR: AddressSanitizer crashed on unknown address 0x000000000010 (pc 0x000000927be8 sp 0x7ff7fc2a3160 bp 0x7ff7fc2a32d0 ax 0x100000000000 T3)
[03:08] <durandal_1707> ahh, it is lucky
[03:08] <michaelni> does it actually display anything for you with any tool ?
[03:10] <durandal_1707> tools here are at minimum
[03:38] <durandal_1707> libopenjpeg decoder is broken and it likes to segfault
[04:05] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r33f39c02aa 10ffmpeg/libavformat/mxfdec.c: 
[04:05] <CIA-17> ffmpeg: mxfdec: pass correct context to av_timecode_init()
[04:05] <CIA-17> ffmpeg: fixes null ptr dereference
[04:05] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:05] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rc855ece101 10ffmpeg/libavcodec/ (indeo5.c ivi_common.c): 
[04:05] <CIA-17> ffmpeg: indeo5: check motion vectors.
[04:05] <CIA-17> ffmpeg: fixes out of frame reading
[04:05] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[04:05] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:05] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r140c54ea0b 10ffmpeg/libavcodec/indeo4.c: 
[04:05] <CIA-17> ffmpeg: indeo4: better mv check
[04:06] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:31] <durandal_1707> michaelni: actually for that sample which causes segv libopenjpeg managed to show something
[05:45] <CIA-17> ffmpeg: 03Paul B Mahol 07master * r8e45005a84 10ffmpeg/libavcodec/libopenjpegenc.c: 
[05:45] <CIA-17> ffmpeg: libopenjpegenc: use pix_fmt descriptor to get bpp
[05:45] <CIA-17> ffmpeg: This will also simplify adding support for RGB565 and similar.
[05:45] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[05:45] <CIA-17> ffmpeg: Reviewed-by: Michael Bradshaw <mbradshaw at sorensonmedia.com>
[05:45] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:45] <CIA-17> ffmpeg: 03Paul B Mahol 07master * rdec8eb6b29 10ffmpeg/libavcodec/libopenjpegenc.c: 
[05:45] <CIA-17> ffmpeg: libopenjpegenc: merge common code
[05:45] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[05:45] <CIA-17> ffmpeg: Reviewed-by: Michael Bradshaw <mbradshaw at sorensonmedia.com>
[05:45] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:45] <CIA-17> ffmpeg: 03Paul B Mahol 07master * r67d5fcc989 10ffmpeg/libavcodec/libopenjpegenc.c: 
[05:45] <CIA-17> ffmpeg: libopenjpegenc: allow changing encoding parameters
[05:45] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[05:45] <CIA-17> ffmpeg: Reviewed-by: Michael Bradshaw <mbradshaw at sorensonmedia.com>
[05:45] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:45] <CIA-17> ffmpeg: libopenjpegenc: use pix_fmt descriptor to get numcomps
[05:45] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[05:45] <CIA-17> ffmpeg: Reviewed-by: Michael Bradshaw <mbradshaw at sorensonmedia.com>
[05:45] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:45] <CIA-17> ffmpeg: 03Paul B Mahol 07master * r3dc0b9d6da 10ffmpeg/libavcodec/libopenjpegdec.c: 
[05:45] <CIA-17> ffmpeg: libopenjpegdec: refactor some code and fix memory leaks
[05:45] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[05:45] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:45] <CIA-17> ffmpeg: 03Paul B Mahol 07master * r02fb320ada 10ffmpeg/libavcodec/libopenjpegdec.c: 
[05:45] <CIA-17> ffmpeg: libopenjpegdec: always check image because decoding may still fail
[05:45] <CIA-17> ffmpeg: Fixes bunch of segfaults.
[05:46] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[05:46] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[10:18] <CIA-17> ffmpeg: 03Martin Storsjö 07release/0.8 * rceeaf42451 10ffmpeg/libavcodec/avcodec.h: (log message trimmed)
[10:18] <CIA-17> ffmpeg: avcodec: Remove a misplaced and useless attribute_deprecated
[10:18] <CIA-17> ffmpeg: If attribute_deprecated is used in an enum declaration, it
[10:18] <CIA-17> ffmpeg: should follow the 'enum' keyword, otherwise it's ignored
[10:18] <CIA-17> ffmpeg: silently. This is the only case of attribute_deprecated for
[10:18] <CIA-17> ffmpeg: enum declarations currently.
[10:18] <CIA-17> ffmpeg: Currently, this attribute_deprecated doesn't have any effect.
[15:47] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r3b370abf16 10ffmpeg/libavformat/mxfdec.c: 
[15:47] <CIA-17> ffmpeg: mxfdec: fix off by one error.
[15:47] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[15:47] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:47] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r7d74aaf698 10ffmpeg/libavcodec/qdm2.c: 
[15:47] <CIA-17> ffmpeg: qdm2dec: fix out of array read
[15:47] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[15:47] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:47] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rcece491daa 10ffmpeg/libavcodec/qdm2.c: 
[15:47] <CIA-17> ffmpeg: qdm2: Fix out of array writes
[15:47] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:47] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rfe91becc2d 10ffmpeg/libavcodec/qdm2.c: 
[15:47] <CIA-17> ffmpeg: qdm2: fix out of array read
[15:47] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:54] <durandal_1707> michaelni: looks like we need implementing icc profile
[15:55] <j-b> for?
[15:56] <durandal_1707> jp2 use this
[15:59] <j-b> brrr
[15:59] <j-b> in XYZ?
[15:59] <durandal_1707> jp2 can store anything
[15:59] <durandal_1707> and icc describe what colorpsace is stored
[15:59] <durandal_1707> like 12bit gray
[16:01] <durandal_1707> does anyone know link to some mj2 samples?
[16:01] <j-b> I do
[16:01] <j-b> and I will feed you soon with DCP files
[16:01] <j-b> from actual theatres
[16:06] <durandal_1707> there is hardware which can display XYZ?
[16:07] <j-b> theatre projectors
[16:08] <Tjoppen> probably only a subset of XYZ, but makes more sense than feeding them sRGB
[16:14] <cbsrobot> I don't think the projectors can show xyz colors
[16:15] <j-b> I don't think so either. but they take that as input,IIRC
[16:15] <cbsrobot> the dci specification use xyz colorspace in order to not use a limited colorspace - for futrue usage
[16:16] <cbsrobot> I think the hardware converts it to sRGB for the projection - but yes, the files need to be in XYZ
[16:17] <j-b> I am just wondering about that, since we got so many requests about DCP playback since a few weeks
[16:18] <cbsrobot> well DCP is the new 35mm - so ....
[16:18] <durandal_1707> j-b: perhaps they need XYZ -> rgb/yuv conversion
[16:19] <j-b> durandal_1707: well, so far, the latest VLC displays a nice green screen on 2.0.1 on Mac
[16:19] <durandal_1707> how many bits XYZ use?
[16:19] <j-b> 12
[16:19] <durandal_1707> not 16?
[16:19] <j-b> using lavf MXF demuxer
[16:19] <cbsrobot> ffplay shows the DCP nicely
[16:19] <j-b> and lavc openjpeg wrapper decoder
[16:20] <cbsrobot> no only 12bit
[16:20] <j-b> cbsrobot: it could very well be a VLC integration issue (that is what I assumed, first)
[16:20] <j-b> but it could be the machine is too slow.
[16:20] <durandal_1707> hmm, libopenjpeg wrapper doesn't do colorspace conversion
[16:20] <cbsrobot> j-b: I think I sent you the screenshot ... http://bayimg.com/dakIkAADK
[16:21] <cbsrobot> although ist mplayer - but ffplay shows it aswell
[16:21] <j-b> I do not remember this screenshot
[16:21] <cbsrobot> it's from a few month ago
[16:21] <j-b> and trying to compile against openjpeg 1.5 did not work.
[16:22] <cbsrobot> j-b I have a patch
[16:22] <j-b> configure rejected it, or something
[16:22] <cbsrobot>  http://pastebin.com/sgffmm84
[16:22] <j-b> openjpeg 1.5 was supposed to be a bit less slow
[16:22] <cbsrobot> for openjpeg 1.5
[16:22] <j-b> ohhh
[16:22] <durandal_1707> headers for both versions are same
[16:23] <durandal_1707> and my ffmpeg build use 1.5
[16:23] <cbsrobot> j-b yeah - one or two milliseconds maybe
[16:23] <cbsrobot> or even half
[16:23] <j-b> If we have a way to display that with powerful machines, I might find the money to sponsor more ASM/development around that.
[16:24] <cbsrobot> j-b: that would be great
[16:25] <j-b> well, depends on what is needed :_
[16:25] <durandal_1707> it looks to me conversion would need poverfull machine anyway
[16:25] <j-b> conversion?
[16:26] <durandal_1707> xyz-> ...
[16:26] <j-b> should be doable in GLSL
[16:27] <durandal_1707> you mean outside of sws?
[16:27] <j-b> yes
[16:28] <j-b> I need decoding to YUV 12bits or something that I can convert to RGB in GLSL shaders 
[16:28] <durandal_1707> the only remaining problem i see is XYZ detection for which ICC is needed
[16:28] <durandal_1707> if j2k stream actually have ICC
[16:28] <cbsrobot> durandal_1707: could be a command line switch - no ?
[16:29] <durandal_1707> or where in the file is stored informationn about used colorspace?
[16:29] <cbsrobot> I think it's not stored
[16:30] <cbsrobot> but all files using cinema_*k profile are in xyz
[16:31] <j-b> cbsrobot: no idea what kind of money is needed
[16:35] <CIA-17> ffmpeg: 03Thilo Borgmann 07master * r04dc7e7e5e 10ffmpeg/libavcodec/alsdec.c: 
[16:35] <CIA-17> ffmpeg: ALS error message pretty print
[16:35] <CIA-17> ffmpeg: One prettier message.
[16:35] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:35] <CIA-17> ffmpeg: 03Paul B Mahol 07master * r3f26a87488 10ffmpeg/libavcodec/libopenjpegenc.c: 
[16:35] <CIA-17> ffmpeg: libopenjpegenc: remove leftover because older version of patch being applied
[16:35] <CIA-17> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[16:35] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:35] <CIA-17> ffmpeg: 03Thilo Borgmann 07master * r5afd9a80db 10ffmpeg/libavcodec/alsdec.c: 
[16:35] <CIA-17> ffmpeg: als: consistent messages
[16:35] <CIA-17> ffmpeg: "." for warnings, "!" for errors.
[16:35] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:37] <cbsrobot> converting a 5 min 1920x2080 prores file to dcp takes about 40 min ... lol
[16:37] <cbsrobot> *1920x1080
[16:37] <durandal_1707> you mean using libopenjpeg encoder?
[16:39] <cbsrobot> yeah its more like prores --(ffmpeg)-->16bit tiff --(opendcp using openjpeg)--> dcp
[16:41] <durandal_1707> hmm what needs doing to make something like this: prores --(ffmpeg)--> dcp
[16:41] <cbsrobot> yes of course
[16:42] <cbsrobot> dcp is more like a package - there is mxf inside - so
[16:56] Action: durandal_1707 gonna fix j2kdec color_space detection for jp2
[17:02] Action: Compn guesses all of this jp2k stuff will break something in mplayer codecs.conf
[17:03] <Compn> er nm, i'm thinking ffj2k :)
[17:03] <gnafu> s/mplayer codecs.conf/something that's already broken/
[17:03] Action: gnafu ducks.
[17:21] <durandal_1707> sucks, it sucks that you need to completly rewrite j2kdec before you start fixing
[17:21] <durandal_1707> it implements own get_bits() braindead
[17:31] <durandal_1707> michaelni: when get_bits stuff where introduced in FFmpeg?
[17:38] <michaelni> durandal_1707, the code in j2kdec probably with the first j2kdec version, the code in libavcodec get bist probably with the first ffmpeg checkin
[20:04] <CIA-17> ffmpeg: 03Reimar Döffinger 07master * rfca62f0500 10ffmpeg/ (libavcodec/flicvideo.c libavformat/isom.c): 
[20:04] <CIA-17> ffmpeg: Support FLIC in MOV.
[20:04] <CIA-17> ffmpeg: Fixes trac issue #626.
[20:04] <CIA-17> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[20:04] <CIA-17> ffmpeg: 03Reimar Döffinger 07master * r699c61d2f0 10ffmpeg/libavformat/utils.c: (log message trimmed)
[20:04] <CIA-17> ffmpeg: Handle 0-size sync packets when only parsing headers.
[20:04] <CIA-17> ffmpeg: Currently, the duration of those packets is just discarded
[20:04] <CIA-17> ffmpeg: when enabling parsing, thus the output of the Metal Gear Solid
[20:04] <CIA-17> ffmpeg: demuxer breaks completely when just setting AVSTREAM_PARSE_HEADERS.
[20:04] <CIA-17> ffmpeg: The result will not be correct if a parser creates a delay even
[20:04] <CIA-17> ffmpeg: with PARSER_FLAG_COMPLETE_FRAMES and there might be other cases
[20:04] <CIA-17> ffmpeg: 03Reimar Döffinger 07master * r60497cb984 10ffmpeg/libavformat/mgsts.c: 
[20:04] <CIA-17> ffmpeg: Enable generic index/seeking for mgsts demuxer.
[20:04] <CIA-17> ffmpeg: This depends on the proposed parser change for 0-size packets
[20:39] <juanmabc> Hi!, i got a flv vbr video going slowly issue (audio ok), so i got pts and i do: if(pts * av_q2d(media.video.codec->time_base) * 1000000.0 <= av_gettime() - media->masterclock) update_gltexture
[20:43] <juanmabc> if pts is in time_base units and this is in seconds this should be ok, but doesn't, it's like time_base is wrong (?)
[20:57] <michaelni> juanmabc, there are multiple timebases :)
[20:58] <juanmabc> could you explain yourself?
[20:58] <michaelni> AVFormatContext.streams[].time_base and AVCodecContext.time_base
[20:58] <michaelni> one is from the demuxer layer and one codec layer
[20:59] <juanmabc> so i use codec layer should i use demuxer layer?
[20:59] <michaelni> if your pts are from the demuxer layer (AVPackets) yes
[20:59] <juanmabc> nice, i'll give it a try
[20:59] <juanmabc> (hope it solves seeking issue with delta frames ;P)
[20:59] <juanmabc> thanks
[21:00] <michaelni> np
[21:02] <juanmabc> yep, seems to do it, yay :D
[21:06] <juanmabc> mmm, nop, despite this solving sync, my seeking unsync issue persists
[21:08] <juanmabc> i use avformat_seek_file, but the pts of the frame is weird, seems a lot later of audio, and seems to come from the first non delta video frame
[21:10] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * rb2af057a36 10ffmpeg/libavcodec/smacker.c: smacker: convert palette and header reading to bytestream2.
[21:10] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * rc3bbd0b53b 10ffmpeg/libavcodec/alac.c: alac: convert extradata reading to bytestream2.
[21:10] <juanmabc> mmm, i use avcodec_flush_buffers
[21:10] <CIA-17> ffmpeg: 03Jindrich Makovicka 07master * r3fadb29baf 10ffmpeg/libavformat/mpegtsenc.c: 
[21:10] <CIA-17> ffmpeg: mpegtsenc: allow user triggered PES packet flushing
[21:10] <CIA-17> ffmpeg: Signed-off-by: Jindrich Makovicka <jindrich.makovicka at nangu.tv>
[21:10] <CIA-17> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[21:10] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r9c47f2b294 10ffmpeg/doc/APIchanges: APIchanges: mark the place where 0.7 was cut.
[21:10] <CIA-17> ffmpeg: 03Anton Khirnov 07master * rdca9c81d82 10ffmpeg/libavformat/avformat.h: lavf doxy: clarify that an AVPacket contains encoded data.
[21:10] <CIA-17> ffmpeg: 03Anton Khirnov 07master * rf58b8cc3e3 10ffmpeg/libavformat/avformat.h: lavf doxy: add some basic documentation about reading from the demuxer.
[21:10] <CIA-17> ffmpeg: 03Anton Khirnov 07master * re44ada129c 10ffmpeg/libavformat/avformat.h: 
[21:10] <CIA-17> ffmpeg: lavf doxy: improve AVStream.time_base doxy.
[21:10] <CIA-17> ffmpeg: Remove confusing sentence that implied the user should set the timebase.
[21:10] <CIA-17> ffmpeg: Elaborate on how the timebase is set for muxing.
[21:10] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r10fa4ff7bc 10ffmpeg/libavformat/avformat.h: lavf doxy: document passing options to demuxers.
[21:10] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r967923abd1 10ffmpeg/libavformat/avformat.h: lavf doxy: expand AVStream.codec doxy.
[21:10] <CIA-17> ffmpeg: 03Diego Biurrun 07master * r6a7c5312d8 10ffmpeg/configure: build: drop some unnecessary dependencies from the H.264 parser
[21:10] <CIA-17> ffmpeg: 03Diego Biurrun 07master * r193d7eea95 10ffmpeg/ (configure libavutil/Makefile libavutil/lzo.c): libavutil: Remove broken and pointless lzo test program.
[21:10] <CIA-17> ffmpeg: arch subdirectory Makefiles, just putting them in place is enough.
[21:10] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r5d20e7b7ea 10ffmpeg/doc/APIchanges: APIchanges: fill in missing dates and hashes.
[21:10] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r75bdd55e77 10ffmpeg/doc/APIchanges: APIchanges: mark the place where 0.8 was cut.
[21:11] <juanmabc> michaelni: do you know what could mess the pts on seek ?
[21:18] <michaelni> juanmabc, check if it works with ffplay and if it does check what it does different
[21:31] <funman> nefrir: did you work on mpeg4 with dxva2 already?
[21:31] <juanmabc> ok, i'm allways with it as help code, still this issue persist, gonna recheck
[21:33] <juanmabc> it's like there is some trash on the first read packets for video
[21:41] <durandal_1707> -win1
[21:47] <durandal_1707> is there way to ban bug reporters?
[22:27] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * ra84851bef8 10ffmpeg/libavcodec/indeo3.c: 
[22:27] <CIA-17> ffmpeg: indeo3dec: check mv bitstream pointer
[22:27] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[22:27] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:27] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r8a521d57b3 10ffmpeg/libavcodec/indeo3.c: 
[22:27] <CIA-17> ffmpeg: indeo3dec: Fix end pointer.
[22:27] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[22:27] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:45] <ubitux> michaelni: ok to update RELEASE to 0.10.2.git?
[22:46] <michaelni> sure
[22:48] <CIA-17> ffmpeg: 03Clément BSsch 07master * r21d650cb69 10ffmpeg/ffmpeg.c: 
[22:48] <CIA-17> ffmpeg: ffmpeg: more expressive sample rate automatic selection.
[22:48] <CIA-17> ffmpeg: Output now lists the available sample rates with commands like
[22:48] <CIA-17> ffmpeg: ffmpeg -f lavfi -i aevalsrc=0 -ar 20000 -y /tmp/out.mp3
[22:48] <CIA-17> ffmpeg: 03Clément BSsch 07master * ra67d9cfa58 10ffmpeg/ (4 files in 2 dirs): lavfi: add avfilter_fill_frame_from_{audio_,}buffer_ref().
[22:48] <CIA-17> ffmpeg: 03Clément BSsch 07master * r8f0d9b43be 10ffmpeg/RELEASE: RELEASE: bump to current version (0.10.2).
[22:48] <CIA-17> ffmpeg: 03Clément BSsch 07master * r29679f4a37 10ffmpeg/libavcodec/ (aacenc.c libvo-aacenc.c): aac: reference supported sample rates in encoders.
[22:48] <CIA-17> ffmpeg: 03Clément BSsch 07master * rfb8469dd7e 10ffmpeg/doc/APIchanges: APIchanges: fill in missing dates and hashes.
[22:51] <ubitux> i wonder if there is no "auto hash fill"
[23:15] <teratorn> ffs AVDictionary is just the worst thing ever
[23:15] <durandal_1707> why?
[23:15] <teratorn> is it possible to use the new tree.c/h code to set codec options
[23:16] <teratorn> durandal_1707: it's the most brain-dead key/value datastructure imaginable :)
[23:16] <ubitux> why?
[23:16] <ubitux> you mean it's not optimized?
[23:16] <teratorn> it's an array, yeah
[23:16] <teratorn> the documentation says its deprecated and slow
[23:17] <teratorn> *shrug*
[23:17] <ubitux> do you have performance issues at option parsing?
[23:17] <teratorn> I just felt like ranting
[23:17] <teratorn> nope, it's fine
[23:17] <teratorn> but I think a hash-table would even be easier to implement.. but the tree is probably faster to iterate
[23:18] <ubitux> iirc michaelni tried to convince the author about this
[23:18] <ubitux> not sure about the outcome
[23:18] <ubitux> certainly some kind of blinkers issue :)
[23:18] <teratorn> with a hastable you have to iterate all buckets to iterate the key/values, even though most buckets are empty
[23:19] <teratorn> so the binary tree might be nice for a basic key/value store
[23:19] <teratorn> anyway, just ranting...
[23:19] <ubitux> feel like sending a patch to make AVDictionary use AVTree?
[23:19] <teratorn> and wonder if the tree can actually be used for setting codec options
[23:19] <teratorn> ubitux: hrmph. possibly
[23:20] <teratorn> ubitux: I need to get some code working first, but it might not be too hard to do assuming AVTree works
[00:00] --- Tue Mar 27 2012