[Ffmpeg-devel-irc] ffmpeg-devel.log.20130206

[burek]

[00:21] <cone-290> ffmpeg.git 03Michael Niedermayer 07release/1.0:0c2a350762c3: movtextenc: fix pointer messup and out of array accesses
[00:21] <cone-290> ffmpeg.git 03Michael Niedermayer 07release/1.0:21185028b4fb: dvenc: dont fail hard if the timecode is invalid
[00:21] <cone-290> ffmpeg.git 03Michael Niedermayer 07release/1.1:a55c274f519f: movtextenc: fix pointer messup and out of array accesses
[00:21] <cone-290> ffmpeg.git 03Michael Niedermayer 07release/1.1:f6b50924a5fd: dvenc: dont fail hard if the timecode is invalid
[00:21] <cone-290> ffmpeg.git 03Michael Niedermayer 07release/1.1:5a97a5291a83: riff: fix infinite loop
[00:21] <cone-290> ffmpeg.git 03Michael Niedermayer 07release/1.1:d2c1a8dc2de8: ljpegenc: allocate needed scratch-buffer
[00:24] <cone-290> ffmpeg.git 03Stefano Sabatini 07master:58c4b0f99082: ffplay: extend feedback in case of video changes
[00:24] <cone-290> ffmpeg.git 03Stefano Sabatini 07master:212bb6490fa6: ffplay: report video frame changes only in debug mode
[00:37] <ubitux> llogan: awesome :)
[00:40] <cone-290> ffmpeg.git 03Stefano Sabatini 07master:a12787960e47: lavf/segment: extend PTS debugging log with -reset_timestamps
[01:31] <cone-290> ffmpeg.git 03Michael Niedermayer 07master:ccf9dd00da05: ffmpeg: dont allow -flags to override -pass
[01:47] <cone-290> ffmpeg.git 03Michael Niedermayer 07master:68dfe530e0fb: vdpau: Rename union, avoid name conflict with vdpau
[02:27] <cone-290> ffmpeg.git 03Michael Niedermayer 07release/1.1:0e1bb99f26c2: update for 1.1.2
[02:27] <cone-290> ffmpeg.git 03Michael Niedermayer 07release/1.1:5f3fa5f93087: ffmpeg: dont allow -flags to override -pass
[03:14] <llogan> michaelni: ping on rmlist
[03:15] <michaelni> llogan, iam busy with 1.1.2 release, is it ok i do it tomorrow or later ?
[03:16] <michaelni> i dont want to quickly enter a random command as root ;)
[03:17] <llogan> fine with me. i also personally archived the archives incase you do blow it up by accident
[03:23] <michaelni> llogan, done
[03:28] <llogan> michaelni: thanks
[04:15] <cone-290> ffmpeg.git 03Michael Niedermayer 07master:127ff88639e7: ffmpeg: do not call exit from exit_program()
[04:22] <cone-290> ffmpeg.git 03Michael Niedermayer 07release/1.1:b48cf5412bb4: ffmpeg: do not call exit from exit_program()
[04:46] <cone-290> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n1.1.2': unknown revision or path not in the working tree.
[04:46] <cone-290> Use '--' to separate paths from revisions
[04:46] <cone-290> refs/tags/n1.1.2:HEAD: ffmpeg: do not call exit from exit_program()
[10:31] <cone-533> ffmpeg.git 03Michael Niedermayer 07release/0.10:4d9bde86d024: ffmpeg: dont allow -flags to override -pass
[10:31] <cone-533> ffmpeg.git 03Michael Niedermayer 07release/0.11:3cecef5b0389: ffmpeg: dont allow -flags to override -pass
[10:31] <cone-533> ffmpeg.git 03Michael Niedermayer 07release/0.9:9f0bf48b5ce4: ffmpeg: dont allow -flags to override -pass
[10:31] <cone-533> ffmpeg.git 03Michael Niedermayer 07release/1.0:86f072ceb3e9: ffmpeg: dont allow -flags to override -pass
[11:08] <durandal_1707> michaelni: there is some talk on HA how ffmpeg native resample is slow and ugly quality
[11:18] <LoopHoldYoaBrown> Hi
[11:19] <LoopHoldYoaBrown> How can i make sure libavcodec can't call get_format over and over again during hw decoding?
[11:30] <durandal_1707> is there way to change audio pitch? changing sample rate without changing audio?
[13:22] <cone-533> ffmpeg.git 03Diego Biurrun 07master:25841dfe806a: Use ptrdiff_t instead of int for {avg, put}_pixels line_size parameter.
[13:22] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:ede45c4e1da8: Merge commit '25841dfe806a13de526ae09c11149ab1f83555a8'
[13:22] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:0ddca7d416cc: dsputil: fixup half a dozen bugs with ptrdiff vs int linesize
[13:29] <iive> these commits sound interesting.
[13:33] <cone-533> ffmpeg.git 03Diego Biurrun 07master:c9f933b5b62d: Add av_cold attributes to arch-specific init functions
[13:33] <cone-533> ffmpeg.git 03Diego Biurrun 07master:70b348ec5b6a: configure: Group all hwaccels together in a separate variable
[13:33] <cone-533> ffmpeg.git 03Diego Biurrun 07master:620289a20e02: sh4: Fix silly type vs. variable name search and replace typo
[13:33] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:6c3888487686: Merge commit '620289a20e022b9c16c10d546ef86cc0bb77cc84'
[13:33] <iive> I actually have 2 questions. Is there any improvement for using ptrdiff over int? is ptrdiff_t signed?
[13:34] <durandal_1707> in reality it is just cosmetic change with minor if any gain
[13:35] <DEATH> ptrdiff_t is signed size_t pretty much?
[13:38] <cone-533> ffmpeg.git 03James Almer 07master:8ab2173ed141: lavc/bink: Chech for malloc failure
[13:44] <ubitux> hey there
[13:44] <Compn> hey you
[13:44] <michaelni> hey 
[13:44] <ubitux> seems there are some laught in the cvs-log
[13:44] <Compn> ours ?
[13:45] <ubitux> 25841dfe ’ 0ddca7d4
[13:45] <ubitux> iive: isn't the ptrdiff thing useful for the asm?
[13:45] <ubitux> avoid overread etc
[13:46] <iive> i don't see how.
[13:53] <Compn> devs using unfamiliar calls tend to forget signedness and sizes with the new types. well , afaik anyway
[13:58] <iive> i do see now how. some 64bit asm function did sign extend, so it was removed.
[13:59] <iive> or rather moved in the C code. So in reality it may even affect execution speed.
[14:00] <durandal_1707> at what magnitude?
[14:01] <iive> let's hope it makes it faster.
[14:01] <cone-533> ffmpeg.git 03Tim Walker 07master:7af876a93fa8: flac: add channel layout masks for streams with 7 or 8 channels
[14:09] <iive> but imho it just puts 64bit variable in the stack instead of 32bit. the sign extend should be done somewhere.
[14:51] <cone-533> ffmpeg.git 03Luca Barbato 07master:b2a722a87e3b: sh4: drop unused functions
[14:51] <cone-533> ffmpeg.git 03Vladimir Pantelic 07master:77bcb89600e7: vc1dec: use codec_id instead of codec_tag for VC1IMAGE
[14:51] <cone-533> ffmpeg.git 03Diego Biurrun 07master:48a4ffa722c0: asf: K&R formatting cosmetics
[14:51] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:fadf845973ae: Merge commit '48a4ffa722c0874b251de9d201babed52cef0bcb'
[14:59] <cone-533> ffmpeg.git 03Anton Khirnov 07master:99162f8d46db: vf_yadif: silence a warning.
[15:00] <cone-533> ffmpeg.git 03Anton Khirnov 07master:6ed9fc44badb: svq1: replace struct svq1_frame_size with an array.
[15:00] <cone-533> ffmpeg.git 03Anton Khirnov 07master:238614de679a: cdgraphics: do not rely on get_buffer() initializing the frame.
[15:00] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:6e9f3f3b6522: Merge commit '238614de679a71970c20d7c3fee08a322967ec40'
[15:13] <michaelni> durandal_1707, about ha & ffmpeg resampler, beastd and his friends did some double blind tests to adjust the resampler parameters, so best you ask him
[15:14] <michaelni> ill also ping him about it when i see him
[15:17] <cone-533> ffmpeg.git 03Anton Khirnov 07master:6a399854517f: cmdutils: remove ansi from the list of broken codecs.
[15:17] <cone-533> ffmpeg.git 03Anton Khirnov 07master:9bd6375d5f16: msrledec: check bounds before constructing a possibly invalid pointer,
[15:17] <cone-533> ffmpeg.git 03Anton Khirnov 07master:da7baaaae79b: aasc: fix output for msrle compression.
[15:17] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:ddc3ca2caa4a: Merge commit 'da7baaaae79b4d7d715d35ea6bcfbdd149edc177'
[15:26] <cone-533> ffmpeg.git 03Anton Khirnov 07master:2cd4068071b9: fraps: fix off-by one bug for version 1.
[15:26] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:a845ac78c942: Merge commit '2cd4068071b9a8908823a3107f97e938211045ce'
[15:31] <cone-533> ffmpeg.git 03Kostya Shishkov 07master:685e6f2e3939: xxan: properly handle odd heights.
[15:31] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:68c0144d028b: Merge commit '685e6f2e3939f124b41c7801cc541dad8252af3d'
[16:03] <cone-533> ffmpeg.git 03Kostya Shishkov 07master:89f11f498b9c: qtrle: fix the topmost line for 1bit
[16:03] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:769eba75eeaa: Merge commit '89f11f498b9c15bc71494a11a7ec560f4adf630d'
[16:09] <cone-533> ffmpeg.git 03Anton Khirnov 07master:06cf597c3525: yop: check that extradata is large enough.
[16:09] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:3138b158dd4b: Merge commit '06cf597c352519d2b70f293518c61b8c312f5d4f'
[16:28] <cone-533> ffmpeg.git 03Anton Khirnov 07master:8136f2344458: yop: check for input overreads.
[16:29] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:73562f1f834e: Merge commit '8136f234445862c94d1c081606b2d1e3d44fccf3'
[16:40] <aballier> michaelni: you should probably change the comment after the qtrle 1bit merge; row_ptr became a pointer to the current row after the merge
[16:41] <michaelni> aballier, will do, thanks
[16:42] <michaelni> aballier, otherwise the code is ok after the merge ? I tested it with the files from the ticket both ways produce the same output
[16:43] <aballier> wasnt sure if the commit was correct, but it seems so
[16:43] <aballier> lemme check again
[16:45] <aballier> for what i remember, all the 1st lines have the skip & 0x80 bit set, I have never been sure if that's the spec or just luck
[16:45] <cone-533> ffmpeg.git 03Anton Khirnov 07master:39a9fdd00fea: yop: initialize palette to 0
[16:45] <cone-533> ffmpeg.git 03Anton Khirnov 07master:dff6197dfb9b: nuv: do not rely on get_buffer() initializing the frame.
[16:46] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:0b74fd113412: Merge commit 'dff6197dfb9b2d145729885c63f865c5fdd184aa'
[16:47] <aballier> meaning doing "pixel_ptr = row_ptr;" before the loop is useless
[16:50] <aballier> and i'm not sure how to interpret 1st line not having skip & 0x80 bit set; it may mean we should process lines_to_change lines and not lines_to_change+1, but i'm lacking a sample like that
[16:50] <michaelni> yes was thinking the same
[16:51] <michaelni> either its correct/unused or we have to wait for a bugreport to provide a sample to know which way it should be interpreted
[16:52] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:d71f8d74db54: qtrle: fix 1bit row_ptr comment
[16:53] <aballier> yep, anyway, if i'm not mistaken, if skip & 0x80 has to be set for 1st lines, then both versions are correct, if it doesnt have to, then its likely both versions are incorrect for that case
[17:08] <cone-533> ffmpeg.git 03Anton Khirnov 07master:293065bdb56e: mpegvideo: initialize dummy reference frames.
[17:08] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:78dc4cb26d5a: Merge remote-tracking branch 'qatar/master'
[18:20] <Compn> seriously?
[18:21] <Compn> i cant write -t 36:30 but have to write -t 00:36:30 ?
[18:21] <michaelni> 36:30 is 36minutes 30 seconds or 36 hours 30 minutes ?
[18:21] <Compn> 36 minutes 30 seconds
[18:22] <Compn> if i write 1:30 , 1 minute, 30 seconds
[18:22] <Compn> if i write -t 30 , it accepts it as seconds ... how does it know its not hours ? :P
[18:22] <michaelni> is this AA:BB  thing a regression ?
[18:23] <Compn> i'm just used to it working in mencoder :)
[18:23] <Compn> the docs mention hh:mm:ss so , its feature request, not regression
[18:23] <Compn> -ss behaves the same way
[18:23] <Compn> as -t , requiring hh:mm:ss
[18:24] <michaelni> probably best to open a ticket so its not forgotten
[18:24] <Compn> i guess :)
[18:24] <Compn> new question, how to specify container aspect ratio ?
[18:25] <cone-533> ffmpeg.git 03Vignesh Venkatasubramanian 07master:579c4add6334: Fate test to print side_data crc if present
[18:25] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:b2c443194484: framecrcenc: Also print the size of side data elements
[18:25] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:864fdfa0627e: framecrcenc: print all side data elements, not just the first
[18:25] <Compn> maybe i should ask in #ffmpeg :)
[20:20] <cone-533> ffmpeg.git 03rogerdpack 07master:6d33026668c4: dshow: fix log message
[20:20] <cone-533> ffmpeg.git 03rogerdpack 07master:3b9166fc7efb: dshow: better error reporting
[20:20] <cone-533> ffmpeg.git 03rogerdpack 07master:c916389bd286: dshow: allow for more codec ID's
[20:20] <cone-533> ffmpeg.git 03Michael Niedermayer 07master:474519433634: Merge remote-tracking branch 'rdp/combined2'
[20:33] <Zeranoe> Why does FFmpeg have "double versions", such as "libavcodec     54. 91.100 / 54. 91.100" both sides of the / are the same.
[20:33] <wm4> isn't that header version vs. library version
[20:34] <wm4> they can be different if you use shared libraries and are criminally insane enough to use newer library binaries with programs compiled against older versions
[20:59] <michaelni> anyone wants to review the 32bit fixed point FFT patch ?
[20:59] <michaelni> Nedeljko Babic  ( 46K) [FFmpeg-devel] [PATCH 1/2] libavcodec: Implementation of 32 bit fixed point FFT
[21:00] <Compn> thats the mips version ?
[21:02] <michaelni> theres generic C code for it and MIPS code
[21:03] <michaelni> ive been wrong, rechecking, its just C code no asm/mips code
[21:09] <wm4> <Rodeo> FFmpeg applied http://patches.libav.org/patch/33823/ without http://patches.libav.org/patch/33822/
[21:09] <wm4> <Rodeo> I wonder if I should tell them?
[21:09] <wm4> <Rodeo> at the very least <= 6 should be <= 8
[21:09] Action: wm4 wonders if he's a spy now
[21:12] <llogan> I think the question is: why not tell us?
[21:13] <wm4> because dicks?
[21:48] <michaelni> beastd, "<durandal_1707> michaelni: there is some talk on HA how ffmpeg native resample is slow and ugly quality"
[21:49] <beastd> michaelni: hydrogenaudio?
[21:49] <michaelni> yes
[21:55] <wm4> is libswsresample better?
[22:05] <Compn> michaelni : unless HA did a double blind study like they did for vorbis, opinions dont mean squat :)
[22:08] <beastd> the infinitywave graph is done with 0.10.4 from 2012-06 FWIW
[22:11] <llogan> anyone try libsoxr?
[22:48] <cone-533> ffmpeg.git 03Matthieu Bouron 07master:59d40fc7e66e: lavf/mov: skip version and flags attributes in mov_read_chan function
[22:57] <cone-533> ffmpeg.git 03Stefano Sabatini 07master:b3d2c6f8b9d0: lavf/segment: fix DTS inconsistencies with -reset_timestamps
[22:57] <cone-533> ffmpeg.git 03Stefano Sabatini 07master:d11ac64ecf03: doc: replace '×' symbol with 'x'
[23:37] <cone-533> ffmpeg.git 03Matthieu Bouron 07release/1.0:f3144b0cc694: lavf/mov: skip version and flags attributes in mov_read_chan function
[23:37] <cone-533> ffmpeg.git 03Matthieu Bouron 07release/1.1:02d3ad860957: lavf/mov: skip version and flags attributes in mov_read_chan function
[00:00] --- Thu Feb  7 2013