[Ffmpeg-devel-irc] ffmpeg-devel.log.20121103

[burek]

[00:09] <cone-698> ffmpeg.git 03Michael Niedermayer 07d8aaec38fc89: amrnbdec: allow sample rates other than 8khz
[00:09] <cone-698> ffmpeg.git 03Michael Niedermayer 072c1e9383a275: amrwbdec: allow sample rates other than 16khz
[00:09] <cone-698> ffmpeg.git 03Michael Niedermayer 070aaf0a07d490: libopencore-amrdec: allow sampling rates other than 8khz
[00:09] <cone-698> ffmpeg.git 03Michael Niedermayer 07b30f4510d8c7: libopencore-amr: allow sampling rates other than 8khz
[00:09] <cone-698> ffmpeg.git 03Michael Niedermayer 074ad6ef00a274: libvo-amrwbenc: allow sampling rates other than 16khz
[00:10] <durandal_1707> michaelni: what i should set for description/explanation that dr1 does not work for mimic?
[00:11] <saste> durandal_1707, fixed
[00:11] <durandal_1707> saste: ???
[00:12] <saste> durandal_1707, was a joke, write "fixed" in the commit log
[00:14] <durandal_1707> it clearly manipulates with AVFrames in awkward way
[00:17] <durandal_1707> and it does not set buffer_hints
[00:18] <durandal_1707> and if i set the artifacts pops up on screen
[00:24] <michaelni> when i looked yesterday, the buffer manipulation looked invalid but looking again now, i think ive misread the code
[00:25] <durandal_1707> it changes linesizes
[00:30] <michaelni> only for its internal buffer, the one it allocates matches what it returns if i didnt miss anything
[00:31] <durandal_1707> well it still does not set buffer_hints - i set all flags
[00:31] <durandal_1707> and blocks of various size are corrupted
[00:42] <durandal_1707> michaelni: about indeo3: it makese more sense to release buffer before calling mmx stuff
[00:49] <cone-698> ffmpeg.git 03Paul B Mahol 07b02116cc4a3a: Revert "Partly revert f174fba: indeo3 decoder does not support direct rendering."
[00:49] <cone-698> ffmpeg.git 03Paul B Mahol 0720ac916e6d94: indeo3: release buffer before calling decode_plane
[00:51] <cone-698> ffmpeg.git 03Michael Niedermayer 076e6033b793a3: gsmdec: fix decoding of non 8khz gsm files
[00:51] <cone-698> ffmpeg.git 03Michael Niedermayer 07d3dfe99bc529: libgsmdec: fix decoding of non 8khz gsm files
[01:15] <durandal_1707> someone wants dvb_teletext captions
[01:18] <durandal_1707> ubitux: ^
[01:19] <Daemon404> i think thats a long-awaited feature
[01:19] <Daemon404> that was non trivial to implement
[01:38] <durandal_1707> michaelni: why to inverted SAR for asf?
[02:04] <michaelni> durandal_1707, that was a bugfix
[02:05] <juanmabc> get your hands into .amv format :D
[02:06] <juanmabc> i have a nice 8gb gadget for that
[02:14] <durandal_1707> michaelni: but there is report after that days that it is wrong....
[02:15] <durandal_1707> juanmabc: what is that?
[02:16] <michaelni> which report ?
[02:16] <juanmabc> D A D  adpcm_ima_amv   ADPCM IMA AMV \nDEV    amv             AMV Video
[02:17] <juanmabc> by now it has wrong colorpalete
[02:17] <durandal_1707> juanmabc: a see no bug reports for that
[02:18] <durandal_1707> michaelni: ffloger is your friend: #1883
[02:20] <juanmabc> http://code.google.com/p/amv-codec-tools/ was the merger i belive
[02:21] <juanmabc> http://code.google.com/p/amv-codec-tools/issues/detail?id=15 seems like issue
[02:24] <durandal_1707> juanmabc: that is not valid ffmpeg issue. where is reference amv file and source file so i can compare and actually reproduce issue
[02:24] <juanmabc> ok, i'll report perhaps tomorrow :D
[02:25] <juanmabc> 2:25 AM here
[02:46] <cone-698> ffmpeg.git 03Michael Niedermayer 0703764f673541: af_earwax: remove unused pointer value
[02:46] <cone-698> ffmpeg.git 03Michael Niedermayer 07ac2a3a7a0519: vf_tinterlace: check av_image_get_linesize() return value
[05:53] <cone-245> ffmpeg.git 03Michael Niedermayer 072d8c76eb1f1c: asfdec: parse payload extension list
[05:53] <cone-245> ffmpeg.git 03Michael Niedermayer 078c93269e4207: asfdec: correctly parse payload extensions
[05:53] <cone-245> ffmpeg.git 03Michael Niedermayer 07343c73e252d3: asfdec: parse aspect ratio payload extension
[05:53] <cone-245> ffmpeg.git 03Michael Niedermayer 07e5927910c2a1: asfdec: flip the interpretation of the header aspect back
[06:07] <cone-245> ffmpeg.git 03Michael Niedermayer 07b7280cff9a76: asfdec: check stream_index for validity
[06:56] <nevcairiel> oh that will make some of my users happy, they have been whining about wrong or missing wmv/asf AR for a long time
[12:46] <divVerent> I sometimes get an assertion failed error in libfdk_aac
[12:46] <divVerent> a failing assertion inside libfdk_aac regarding something about quantization
[12:47] <divVerent> before I repeat the command and get the error after 15 min of encoding... is this already known?
[12:48] <divVerent> it deterministically happens always on the same command, but I am currently trying to simplify the command (get video out, etc.)
[12:49] <Compn> dont even know wat libfdk is, sorry
[12:49] <Compn> i see vivo patch back on the list
[12:49] <divVerent> ffmpeg has an encoding driver for it
[12:49] <Compn> hopefully committed after 3 years sitting...
[12:49] <divVerent> just wondering if ffmpeg using its API may be at fault, or if it is sure an error in the library
[12:50] <divVerent> OF COURSE the lib doesn't come with a standalone encoding program
[12:52] <divVerent> ffmpeg: ./libAACenc/src/quantize.cpp:221: void FDKaacEnc_invQuantizeLines(INT, INT, SHORT*, FIXP_DBL*): Assertion `specExp < 14' failed.
[12:56] <divVerent> it happens at any bitrate
[12:56] <divVerent> or not, it happened at 32k, 96k and 128k, but not at 320k
[12:56] <divVerent> looks like it can be easily "tickled away" then ;)
[12:56] <divVerent> like, by volume change, or audio resample
[13:09] <cone-914> ffmpeg.git 03Stefano Sabatini 07c9eaa98a3e0b: doc/ffmpeg: remove -debug option description
[13:09] <cone-914> ffmpeg.git 03Stefano Sabatini 07a62242678ff9: ffmpeg: complain loudly and exit in case removed -sameq option is specified
[13:09] <cone-914> ffmpeg.git 03Stefano Sabatini 077be09a91c299: lswr/swresample.h: use consistent inclusion guard header name
[13:10] <cone-914> ffmpeg.git 03Stefano Sabatini 07481fdeeecfc2: lavu/opt: add AV_OPT_SAMPLE_FMT option
[13:10] <cone-914> ffmpeg.git 03Stefano Sabatini 07ca11f6b240a1: lavu/opt: prefer if(){}else{} over if()else{} construct
[13:10] <cone-914> ffmpeg.git 03Stefano Sabatini 071c755f76676e: lavc/options_table: make "request_sample_fmt" an AV_OPT_TYPE_SAMPLE_FMT option
[13:44] <cone-914> ffmpeg.git 03Diego Biurrun 07012f73e27163: x86inc: Only define program_name if the macro is unset
[13:44] <cone-914> ffmpeg.git 03Diego Biurrun 0761bc2bc7d428: x86util: Add cpuflags_mmxext alias for cpuflags_mmx2
[13:44] <cone-914> ffmpeg.git 03Diego Biurrun 079ce02e14f01d: x86: ac3dsp: port to cpuflags
[13:44] <cone-914> ffmpeg.git 03Michael Niedermayer 071dad48671460: Merge commit '9ce02e14f01de50fcc6f7f459544b140be66d615'
[13:51] <saste> michaelni, maybe it would be better to make av_opt_set_defaults2() return an error (and check it?)
[13:54] <michaelni> yes
[13:54] <cone-914> ffmpeg.git 03Stefano Sabatini 07b91fa5fcca70: lavu/opt: set sample format default value, and accept NULL value
[13:56] <saste> damn I wrote the wrong commit hash...
[14:18] <cone-914> ffmpeg.git 03Diego Biurrun 0726f01bd106f6: x86: PMINUB: port to cpuflags
[14:18] <cone-914> ffmpeg.git 03John Stebbins 0785f67c4865d8: PGS subtitles: Expose forced flag
[14:18] <cone-914> ffmpeg.git 03Diego Biurrun 079a07c1332cfe: parser: Move Doxygen documentation to the header files
[14:18] <cone-914> ffmpeg.git 03Michael Niedermayer 071885ffb03d0a: Merge commit '9a07c1332cfe092b57b5758f22b686ca58806c60'
[14:35] <cone-914> ffmpeg.git 03Diego Biurrun 070a7a94f2e53b: x86: Refactor PSWAPD fallback implementations and port to cpuflags
[14:35] <cone-914> ffmpeg.git 03Anton Khirnov 07b68aac7ea34e: vf_drawtext: do not use deprecated av_tree_node_size
[14:35] <cone-914> ffmpeg.git 03Anton Khirnov 07b4dc6dccbaea: crc: move doxy to the header
[14:35] <cone-914> ffmpeg.git 03Anton Khirnov 07c8977f6d2684: avconv: do not free muxed packet on streamcopy.
[14:35] <cone-914> ffmpeg.git 03Anton Khirnov 07717a45099308: lavf: clarify the lifetime of demuxed packets.
[14:35] <cone-914> ffmpeg.git 03Anton Khirnov 079221efef7968: lavf: fix av_interleaved_write_frame() doxy.
[14:36] <cone-914> ffmpeg.git 03Michael Niedermayer 0737e81996dcf3: Merge commit '9221efef7968463f3e3d9ce79ea72eaca082e73f'
[14:54] <cone-914> ffmpeg.git 03Anton Khirnov 0734871beb5eb6: matroskadec: do not use avpacket internals
[14:54] <cone-914> ffmpeg.git 03Janne Grunau 07b3fab1f2cd22: doc: add apidoc target for doxygen API documentation
[14:54] <cone-914> ffmpeg.git 03Michael Niedermayer 072581d9bcbb2e: Merge remote-tracking branch 'qatar/master'
[15:56] <cone-914> ffmpeg.git 03Paul B Mahol 0794f224516cec: westwood_aud: remove stale comment
[17:07] <saste> michaelni: can you explain the weird ranges (-1 - AV_SAMPLE_FMT_NB-1+256) for sample format options in SwResample?
[17:07] <saste> I wanted to adopt AV_OPT_TYPE_SAMPLE_FMT but those ranges baffle me
[17:11] <cone-914> ffmpeg.git 03Nicolas George 078cf6377ac1cf: doc/faq: explain why -sameq is gone.
[17:13] <michaelni> saste, originally it was intended to use 256 as a flag to indicate planar vs packed
[17:14] <saste> michaelni, so i suppose it is safe to drop it?
[17:14] <michaelni> probably
[17:14] <cone-914> ffmpeg.git 03Nicolas George 0773180ecd0adf: doc/faq: make the -sameq question more explicit.
[17:14] <saste> while for internal_format_sample MAX=AV_SAMPLE_FMT_FLTP
[17:14] <saste> why that?
[17:15] <saste> also what's the internal format supposed to be, anyhow?
[18:33] <durandal_1707> who wrote fate breakage notification tool?
[18:59] <durandal_1707> someone claims av_freep have undefined beahaviour, is this really correct?
[19:20] <durandal_1707> nobody?
[19:22] <ubitux> undefined behaviour, how so?
[19:22] <durandal_1707> pointer aliasing violation
[19:48] <durandal_1707> 4.13 entry in FAQ is stale
[20:08] <michaelni> saste, did you see: "libavcodec/xface.c:313:27: warning: assuming signed overflow does not occur when assuming that (X - c) > X is always false [-Wstrict-overflow]"
[20:09] <saste> michaelni, it's the only one in xfaceenc.c, right?
[20:09] <saste> i also tried to mark it as invalid, but failed
[20:12] <michaelni> thats gcc not coverity
[20:16] <michaelni> note, ive no idea why gcc prints this warning, it might even be a gcc bug
[20:25] <michaelni> saste, about coverity, xfaceenc.c CID739470 is still unclassified, did you mean that one ?
[20:25] <saste> yes, i'm a lot of stuff to do and I'm having serious problems with catching up
[20:25] <saste> i still have to spend some time with coverity and understand how it is supposed to work
[20:26] <cone-979> ffmpeg.git 03Michael Niedermayer 07ab796ded757e: avstring: fix compiler warning about freeing const pointers
[20:26] <cone-979> ffmpeg.git 03Michael Niedermayer 07930226cf52ad: crystalhd: fix missing braces warning
[20:27] <michaelni> ok, no hurry, i just wanted to make sure you dont miss that theres a issue left in xface* in coverity
[20:44] <cone-979> ffmpeg.git 03Stefano Sabatini 070372ecfa4825: lavc/options_table: force non-negative bitrate range
[20:44] <cone-979> ffmpeg.git 03Stefano Sabatini 076450c05ba56d: lswr/swresample: adopt globally consistent rules for options help message
[21:39] <cone-979> ffmpeg.git 03Xidorn Quan 070ecef3a0eff8: amr_decode_fix_avctx should not return value
[22:11] <michaelni> durandal_1707, fate-suite/wavpack/lossy/4.0_8-bit.wv -acodec copy test.wv doesnt produce a playable file
[22:43] <durandal_1707> michaelni: you picked that file out of random or what?
[22:47] <michaelni> random
[22:48] <michaelni> it was actually the only file i tested
[22:48] <michaelni> and i tested because it seems fate isnt testing wvenc
[22:48] <durandal_1707> well it looks to be only for multichannel
[22:49] <durandal_1707> because it some other idiotic way to store data
[22:51] <cone-979> ffmpeg.git 03Michael Niedermayer 0733d54af427f7: fate: add filmstrip test
[22:52] <durandal_1707> michaelni: you have coverage report that claims filmstrip is not supported?
[23:02] <michaelni> durandal_1707, yes
[23:03] <michaelni> ubitux, what problem did you had with coverage ?
[23:04] <ubitux> http://lucy.pkh.me/coverage.log
[23:04] <ubitux> the fate instance is still trying to run the coverage at each run
[23:04] <ubitux> and this is the latest log
[23:05] <michaelni> did you try to clean the directory of all the generated files ?
[23:05] <ubitux> it's within fate, so it should be using a clean working dir each time
[23:05] <ubitux> try by yourself, --enable-coverage
[23:05] <ubitux> then make fate
[23:06] <ubitux> then make coverage-html
[23:07] <michaelni> does it fail for you on your local box ?
[23:07] <ubitux> yes
[23:07] <michaelni> or just the fate client ?
[23:07] <ubitux> no locally as well
[23:07] <michaelni> you tried a clean checkout in a new directory ?
[23:07] <ubitux> i used git clean, tried without threading, etc
[23:08] <ubitux> i really wasn't able to make it work again
[23:08] <saste> michaelni, how/where is coverage info updated?
[23:10] <durandal_1707> saste: what is wrong with histeq?
[23:11] <saste> durandal_1707, possibly overflow issues
[23:11] <saste> result in block artifacts
[23:12] <michaelni> saste, add --disable-optimizations --enable-coverage --extra-cflags='-Dav_always_inline=inline' then run make fate and make coverage-html and you will have coverage locally
[23:12] <ubitux> oh, a new version of lcov
[23:12] <michaelni> the  --disable-optimization and --extra-cflags='-Dav_always_inline=inline' are optional
[23:12] <ubitux> gonna try it and see if it helps
[23:13] <michaelni> ubitux, also if it worked in  the past, why did it stop / what did you upgrade if anything ?
[23:14] <ubitux> i don't really remember, it's been a while now
[23:14] <saste> i have to say that i find all this coverage business pretty boring
[23:19] <durandal_1707> michaelni: can you upload coverage results somewhere (my machine is crap and it takes almos half day to get anything...)
[23:23] <ubitux> michaelni: does it work without --disable-optimizations and the extra cflags for you?
[23:24] <ubitux> i'm going to add these flags
[23:24] <ubitux> and see if it helps
[23:25] <michaelni> ubitux, i think it did work without them
[23:25] <ubitux> oh, it seems to work with lcov 1.10 now
[23:26] <ubitux> well, hopefully it will work on the fate box soon then
[23:26] <ubitux> "soon" = the time to finish the insanely slow threading instances
[23:27] <ubitux> mmh maybe i can run it in parallel
[23:31] <michaelni> saste, make distclean doesnt remove tools/fourcc2pixfmt
[23:31] <saste> michaelni, and the other tools?
[23:32] <michaelni> seem they disappear
[23:32] <michaelni> i tried distclean and git clean --dry-run
[23:34] <ubitux> coverage running fate atm, might take a while
[23:34] <durandal_1707> michaelni: i will try to fix wv for >2 channels in following few hours, if i fail feel free to fix it or disable multi channel muxing, wv demuxer is doing some silly things ....
[23:38] <cone-979> ffmpeg.git 03Stefano Sabatini 07aa48446c9a42: lavc/xfaceenc: prefer do{}while() over for(){} construct in xface_encode_frame() loop
[23:38] <michaelni> saste, correction, the other tools dont disappear either, but fourcc2pixfmt isnt in gitignore
[23:39] <saste> michaelni, we lack a rule which removes tools
[23:39] <michaelni> ubitux, btw this also means a make distclean + git clean will not remove everything
[23:40] <ubitux> michaelni: yeah i guess, but isn't fate supposed to create a new working dir every time?
[23:40] <michaelni> yes i think so but i dont trust fate
[23:40] <ubitux> git clean should remove everything, really
[23:40] <michaelni> ive once seen fate appearently execute code from a old revission
[23:41] <ubitux> anyway, it seems to work fine here/locally now
[23:41] <ubitux> so it should be ok on the server pretty soon
[23:41] <durandal_1707> so what fixed it?
[23:41] <ubitux> lcov 1.9 ’ 1.10
[23:42] <ubitux> i guess 1.9 was too old for recent gcc, dunno
[23:42] <cone-979> ffmpeg.git 03Stefano Sabatini 073af7919be40f: add tools/fourcc2pixfmt to .gitignore
[23:42] Action: michaelni has lcov: LCOV version 1.9
[23:42] <ubitux> what gcc?
[23:42] <michaelni> plain gcc seems 4.6.3 which is what i used IIRC
[23:42] <ubitux> ok, dunno then
[23:43] <ubitux> anyway, we'll know soon if it fixes the issue
[23:44] <ubitux> but it should, my local setup and the one on the server is pretty similar
[23:44] <ubitux> coverage looks kind of low btw
[23:44] <ubitux> or at least, lower than previously
[23:44] <ubitux> iirc
[23:46] <ubitux> http://lucy.pkh.me/ffmpeg-coverage/
[23:46] <ubitux> restored :)
[23:46] <ubitux> and i just moved from lcov 1.9 to lcov 1.10, nothing else.
[23:46] <saste> what is classification "Pending"?
[23:46] <ubitux> michaelni: do you think it makes sense to disable optims?
[23:52] <michaelni> ubitux, i think it makes coverage more accurate if optims and inlining is disabled 
[23:52] <ubitux> ok, i'm going to add these flags for next run
[23:54] <ubitux> oups
[23:56] <durandal_1707> what is this new error now?
[23:57] <ubitux> it's ok, i just messed up
[23:57] <ubitux> no worry
[00:00] --- Sun Nov  4 2012