[Ffmpeg-devel-irc] ffmpeg-devel.log.20121023

[burek]

[00:00] <ubitux> "movie=f='c:\foo\bar'" doesn't actually receive "f='c:\foo\bar'" as arg, but "f=c:\foo\bar"
[00:00] <saste> but it won't work in the general case, e.g. if the filename contains "'"
[00:00] <saste> ubitux, yes because of the double escaping
[00:00] <ubitux> ...so when running a av_get_token in the src movie filter...
[00:00] <ubitux> it will no honor the non-existing quotes
[00:00] <ubitux> and so it fails :(
[00:00] <ubitux> not*
[00:00] <saste> keep in mind that we run av_get_token two times
[00:01] <saste> first when reading the filter description in the filtergraph
[00:01] <saste> so we read until the next "," or ";"
[00:01] <ubitux> yeah right
[00:01] <saste> then we use again av_get_token to read the various arguments, separated by ":"
[00:01] <ubitux> also there is a second issue, which we can't do much about in the mingw/msvc env
[00:02] <saste> that is?
[00:02] <ubitux> basically msys seems to do some sed to translate the paths
[00:02] <ubitux> but it doesn't detect the path when mixed in the filtergraph in the case of the sws flags
[00:02] <ubitux> http://fate.ffmpeg.org/report.cgi?time=20121022212821&slot=x86_32-msvc10-windows-native
[00:02] <saste> ouch
[00:02] <ubitux> look how the two tests fail differently
[00:03] <ubitux> Error initializing filter 'movie' with args '/d/Dev/ffmpeg/fate/samples/svq3/Vertical400kbit.sorenson3.mov'
[00:03] <ubitux> this one is because the path was not transformed
[00:03] <ubitux> (not detected with the "sws_flags=+accurate_rnd+bitexact;", adding a space after may or not help)
[00:03] <ubitux> and the silencedetect failure is because of the issue we were just discussing
[00:04] <saste> we can workaround that
[00:04] <ubitux> ah?
[00:05] <saste> we just need to read the filtergraph from a file
[00:05] <ubitux> :)
[00:05] <ubitux> the silencedetect failure, right?
[00:05] <ubitux> (that's what you meant by "that")
[00:06] <saste> why it is not finding "/d/Dev/ffmpeg/fate/samples/svq3/Vertical400kbit.sorenson3.mov"?
[00:06] <ubitux> because that path doesn't exist
[00:06] <ubitux> msys is supposed to "sed"
[00:06] <saste> what's the real filename?
[00:06] <ubitux> but it doesn't recognize it when mangled with the sws_flags=...
[00:06] <ubitux> /Dev/ffmpeg/fate/samples/amrwb/seed-12k65.awb i guess
[00:07] <ubitux> oups no
[00:07] <ubitux> d:/Dev/ffmpeg/fate/samples/amrwb/seed-12k65.awb
[00:07] <ubitux> so..
[00:07] <ubitux> d:/Dev/ffmpeg/fate/samples/svq3/Vertical400kbit.sorenson3.mov
[00:07] <cone-571> ffmpeg.git 03Michael Niedermayer 0745ae9a8fc903: j2k: reorder multiplications to avoid interger overflows related to code blocks
[00:07] <cone-571> ffmpeg.git 03Michael Niedermayer 07da317efd9256: j2k: reorder multiplications to avoid interger overflows related to precincts
[00:08] <saste> if you read the arguments from a file, it shouldn't mangle the filenames
[00:08] <saste> something like:
[00:09] <saste> ffprobe -f lavfi -lavfi_graph_file=ffprobe_scenedetect_lavfi_graph ...
[00:10] <saste> for the silencedetect problem: "amovie='$(SRC)',silencedetect=d=.1"
[00:10] <saste> for the silencedetect problem: "amovie=\\'$(SRC)\\',silencedetect=d=.1"
[00:10] <ubitux> how do you put the SAMPLES path into the lavfi graph?
[00:10] <ubitux> (file)
[00:10] <ubitux> generated at runtime?
[00:11] <saste> it's complicated, but you can generate the file with a cat command
[00:12] <ubitux> the \\ seems to work
[00:13] <ubitux> amovie=\'c:\foo\bar\':sp=3,silencedetect: Protocol not found
[00:13] <ubitux> haha
[00:14] <ubitux> i guess it will behave differently in a win environment? :)
[00:14] <saste> yes it is the cervellotic heuristic used for protocol recognition
[00:14] <ubitux> (i just made a symlink c:\foo\bar -> test.mp3)
[00:15] <ubitux> do you want to commit the \\ workaround?
[00:15] <saste> me?
[00:15] <ubitux> sure
[00:15] <saste> no i don't want to take the blame if it borks
[00:15] <ubitux> :D
[00:16] <saste> well i'd need to verify that it works as intended in windows
[00:16] <ubitux> i don't have any to test
[00:17] <ubitux> maybe it's not worth the effort if we're going to use the file method
[00:17] <saste> well it shouldn't broke *nix, and windows is already broken so it can't make things worse
[00:17] <saste> possible
[00:17] <ubitux> so yes we can fix the silencedetect with this
[00:18] <ubitux> now we need something for the other one :p
[00:18] <saste> do you want me to implement -lavfi_graph_file?
[00:18] <ubitux> if it solves the issue, could be nice yes
[00:18] <saste> i don't know if i'll do it today or even tomorrow
[00:19] <ubitux> how do you plan to do that btw?
[00:19] <saste> but if you can wait up to two days...
[00:19] <ubitux> -f lavfi -i "graph_file=foobar.lavfi" ?
[00:19] <saste> i'm too tired to do it now
[00:20] <ubitux> i would have done it if i had time, but it's getting late...
[00:20] <saste> no, -i takes the title, if lavfi_graph is not defined it use the title as a filtergraph description
[00:21] <ubitux> mmh
[00:21] <saste> so it is -lavfi_graph_file -f lavfi -i "foobar"
[00:21] <saste> * -lavfi_graph_file FILE -f lavfi -i "foobar"
[00:22] <ubitux> this will be a lavd/lavfi.c AVOption, right? 
[00:22] <saste> yes
[00:22] <ubitux> what kind of dummy thing will you put for -i?
[00:23] <saste> -i "saste rulez"
[00:23] <ubitux> sounds nice
[00:23] <ubitux> will you quote 'saste'?
[00:24] <saste> seriously -i "silencedetect test"
[00:24] <saste> or something like that
[00:24] <ubitux> ah it will act as a filtergraph name, ok :p
[00:29] <cone-571> ffmpeg.git 03Michael Niedermayer 073318d6b94098: ffmpeg: check avpicture_get_size() retuen value
[01:19] <ubitux> saste yay!
[01:20] <ubitux> thanks a lot :)
[01:24] <cone-571> ffmpeg.git 03Paul B Mahol 07a57d560db3a1: LVF demuxer
[02:33] <cone-571> ffmpeg.git 03Michael Niedermayer 077770c0da0731: fate/mapchan: add bitexact flag
[03:04] <ohsix> is the point of the bot to badger people or what, it doesn't seem to get much reaction
[03:12] <cone-571> ffmpeg.git 03Michael Niedermayer 07e9b61fffbe43: g722dec: s/Libav/FFmpeg/ The copyright years predate the existence of Libav
[03:12] <cone-571> ffmpeg.git 03Michael Niedermayer 0734ccb94796b7: g723_1: remove unneeded cliping
[03:16] <michaelni> ohsix, it got you to react, i admit i fixed it already before the next bunch of boxes tested the old pre fix version
[03:40] <creep> h
[05:03] <ohsix> michaelni: well, it didn't make me react in a manner that fixed the failure :]
[09:22] <ubitux> is it possible from the URLContext to acccess the AVFormatContext?
[09:22] <ubitux> i guess no, but well..
[11:00] <ubitux> mpf, headers option in http protocol doesn't add up, it replaces
[11:01] <ubitux> :(
[12:40] <cone-265> ffmpeg.git 03Anton Khirnov 07b691135d0c6a: lavc: remove stats_out from the options table.
[12:40] <cone-265> ffmpeg.git 03Anton Khirnov 075957aefc7c75: libxvid: remove useless doxy comments.
[12:40] <cone-265> ffmpeg.git 03Mans Rullgard 07abdee95224fe: pcm: define AVCodec instances only for enabled codecs
[12:40] <cone-265> ffmpeg.git 03Justin Ruggles 07e55d53905f34: atrac3: cosmetics: pretty-printing and renaming
[12:40] <cone-265> ffmpeg.git 03Justin Ruggles 07aefdb735c3df: atrac3: simplify some loop indexing
[12:40] <cone-265> ffmpeg.git 03Justin Ruggles 075ac673b5531d: atrac3: use AVCodecContext.channels instead of keeping a private copy
[12:40] <cone-265> ffmpeg.git 03Michael Niedermayer 07dcb0d1193a3d: Merge commit '5ac673b5531d846b79a3d77e3e932e0cb1234c45'
[12:49] <cone-265> ffmpeg.git 03Justin Ruggles 0756a9d2b44e97: atrac3: move the 'version' field from ATRAC3Context to where it is used
[12:49] <cone-265> ffmpeg.git 03Justin Ruggles 0764ebbb8f89d1: atrac3: move the 'delay' field from ATRAC3Context to where it is used
[12:49] <cone-265> ffmpeg.git 03Justin Ruggles 07cdd0e0de816d: atrac3: use AVCodecContext.block_align instead of keeping a private copy
[12:49] <cone-265> ffmpeg.git 03Justin Ruggles 077c1f93afe6b0: atrac3: remove unused ATRAC3Context field, samples_per_channel
[12:49] <cone-265> ffmpeg.git 03Justin Ruggles 07a2664c91fba1: atrac3: move the 'samples_per_frame' field from ATRAC3Context to where it is used
[12:49] <cone-265> ffmpeg.git 03Michael Niedermayer 078748472f3996: Merge commit 'a2664c91fba15a1307f676ffad511f8f86fb3a27'
[12:58] <cone-265> ffmpeg.git 03Justin Ruggles 077e76f270819e: atrac3: remove unused ATRAC3Context field, bit_rate
[12:58] <cone-265> ffmpeg.git 03Justin Ruggles 07c51311b907e0: atrac3: move the 'frame_factor' field from ATRAC3Context to where it is used
[12:58] <cone-265> ffmpeg.git 03Justin Ruggles 0778edce3f1973: atrac3: separate window initialization from IMDCT initialization
[12:58] <cone-265> ffmpeg.git 03Justin Ruggles 075d1007f74dd4: atrac3: initialize static tables in AVCodec.init_static_data()
[12:58] <cone-265> ffmpeg.git 03Michael Niedermayer 077bc6631dd027: Merge commit '5d1007f74dd496d54b932242004382f44e3b22b4'
[13:12] <cone-265> ffmpeg.git 03Justin Ruggles 07327747de15be: atrac3: simplify MDCT window calculation
[13:12] <cone-265> ffmpeg.git 03Justin Ruggles 0789a6c32bc1bc: atrac3: use sizeof(variable) instead of sizeof(type)
[13:12] <cone-265> ffmpeg.git 03Justin Ruggles 07808686fc1ea6: atrac3: remove unused ATRAC3Context field, sample_rate
[13:12] <cone-265> ffmpeg.git 03Justin Ruggles 07a1f4cd371ac7: atrac3: replace a calculation with FFALIGN()
[13:12] <cone-265> ffmpeg.git 03Anton Khirnov 074e61a38aa038: avconv: only apply presets when we have an encoder.
[13:12] <cone-265> ffmpeg.git 03Anton Khirnov 079cac8a519980: APIchanges: update lavr bump date
[13:12] <cone-265> ffmpeg.git 03Anton Khirnov 072b8dd371e4d2: lavu: postpone recent deprecations until the next major bump
[13:12] <cone-265> ffmpeg.git 03Michael Niedermayer 07af3fe43d841c: Merge commit '2b8dd371e4d276ca0d342e82b8b4cc281be0630a'
[13:26] <cone-265> ffmpeg.git 03Anton Khirnov 07add3a6902ea4: lavu: bump major to 52
[13:26] <cone-265> ffmpeg.git 03Michael Niedermayer 07c2a25e141e01: Merge commit 'add3a6902ea4d459f36655656a1f8daa409118a7'
[13:46] <cone-265> ffmpeg.git 03Anton Khirnov 07c6b25d1d5e6a: Revert "avutil: Add a copy of ff_sqrt_tab back into avutil to restore ABI compatibility"
[13:46] <cone-265> ffmpeg.git 03Anton Khirnov 07468ea9d5b14f: Revert "avutil: make some tables visible again"
[13:46] <cone-265> ffmpeg.git 03Michael Niedermayer 07a65bc2712db2: Merge commit '468ea9d5b14f92fe61f47f034e67066f65163f5f'
[14:56] <cone-265> ffmpeg.git 03Anton Khirnov 078207c67a2417: lavu: remove disabled FF_API_GET_BITS_PER_SAMPLE_FMT cruft
[14:56] <cone-265> ffmpeg.git 03Anton Khirnov 07c692957c4edd: lavu: remove disabled avoptions cruft
[14:56] <cone-265> ffmpeg.git 03Michael Niedermayer 07d0ac26a264e5: Merge commit 'c692957c4edd119920cd7f2121628dd183589e6a'
[15:04] <cone-265> ffmpeg.git 03Anton Khirnov 0700ed7f5958d0: lavu: remove disabled FF_API_AV_FIFO_PEEK cruft
[15:04] <cone-265> ffmpeg.git 03Diego Biurrun 078f4c414df6b8: Improve wording and spelling of av_log_missing_feature messages.
[15:04] <cone-265> ffmpeg.git 03Diego Biurrun 07e273a1a54291: fate: Add dependencies for WMA and WavPack tests
[15:04] <cone-265> ffmpeg.git 03Diego Biurrun 078b8899ac3233: fate: Declare avcodec/avformat deps in the respective Makefile snippets
[15:04] <cone-265> ffmpeg.git 03Michael Niedermayer 0735ce42e07029: Merge commit '8b8899ac3233b4f7af83ded0dc032fad8902d714'
[15:12] <cone-265> ffmpeg.git 03Diego Biurrun 077d420a4294dc: fate: More fine-grained dependencies for libavcodec test programs
[15:12] <cone-265> ffmpeg.git 03Diego Biurrun 0771f3eb4536d9: fate: cosmetics: Group idct8x8 test together with all other libavcodec tests
[15:12] <cone-265> ffmpeg.git 03Diego Biurrun 0788bdec03b986: fate: Add dependencies for audio tests
[15:12] <cone-265> ffmpeg.git 03Diego Biurrun 072f17f265ed79: mpegaudio: Refactor mp3on4 flush function
[15:12] <cone-265> ffmpeg.git 03Michael Niedermayer 07ec444c84cfd3: mpegaudiodec: Fix buffer handling on random access
[15:12] <cone-265> ffmpeg.git 03Michael Niedermayer 0779ec524ff52a: Merge commit 'ec444c84cfd30ef34a4e7b4ec7ee77d7e5250d25'
[15:15] <cone-265> ffmpeg.git 03Mans Rullgard 076f9018fc5ecd: fate: fix ENCMUX macro
[15:15] <cone-265> ffmpeg.git 03Mans Rullgard 07a1f6ad69c779: fate: check that dependencies actually exist
[15:16] <cone-265> ffmpeg.git 03Michael Niedermayer 074d54ecf4b8e8: Merge remote-tracking branch 'qatar/master'
[15:16] <ubitux> @_@
[15:18] <durandal_1707> that happens when stuff are not written right from the start
[15:51] <cone-265> ffmpeg.git 03Peter Ross 078469fc17e289: wtvdec: display warning for encrypted wtv files (instead of 'unsupported chunk' messages)
[16:18] <cone-265> ffmpeg.git 03Michael Niedermayer 07d8f27eceaabf: asfdec: ignore stored duration for truncated files
[17:32] <cone-265> ffmpeg.git 03Michael Niedermayer 0731cdf6002ae5: lavf: increase retries to guess duration from pts by 1
[18:19] <cone-265> ffmpeg.git 03Michael Niedermayer 075442c6cce33d: lavf: fix duration estimation for multi program TS
[19:28] <cone-265> ffmpeg.git 03Paul B Mahol 07e2820d99f25d: pnmdec: use more meaningful error codes
[20:54] <kolbyjack> Hi, I'm trying to use libavformat to write an mp4 file, and vlc will only play the first second with its default demuxer.  If I reprocess the file with ffmpeg -i myfile.mp4 -vcodec copy ffmpeg.mp4, it plays fine.  Is there an easy way I can track down what I'm doing differently from ffmpeg?  I'm finding it kind of difficult to follow.
[21:04] <funman> kolbyjack: you can try vlc --demux avformat
[21:04] <kolbyjack> Yeah, that plays fine, but I need to play it on an ipad, unfortunately.
[21:05] <kolbyjack> I just need to figure out what I'm doing wrong when I'm muxing, since the video is apparently okay
[21:49] <cone-265> ffmpeg.git 03Isaac Dooley 07469a65b15025: x11: Fix x11grab BadCursor
[21:54] <thegeek> I'm getting a crash in swresample_0!ff_mix_2_1_a_int16_sse2 in my mingw-compiled ffmpeg 1.0 (x64), any tips for things to double-check in the toolchain before I start debugging etc?
[21:55] <thegeek> I'm using mingw-w64 4.7.2 and I can not reproduce it with zeranoe's builds, so it's probably something I'm doing wrong, but it's wierd because a lot of other stuff works fine;P
[21:58] <thegeek> aand if I disable sse/sse2 it happens in the mmx version of the function
[21:58] <michaelni> that excludes SSE alignemnt issues
[21:58] <michaelni> what happens without asm ?
[21:59] <thegeek> hang on
[21:59] <iive> i hit an alignment bug today, in intrinsic code inside mesa3d v9.0. using gcc-4.7.1 
[22:01] <thegeek> I was using mingw-w64 4.7.0 and also some older deps of gcc, updating to the latest did not make a difference
[22:07] <thegeek> michaelni: disabling asm fixed it
[22:09] <thegeek> I'll do some more research on it and support a bug report
[22:10] <thegeek> as far as I can tell I trigger it by using the concat filter two join two video files
[22:11] <iive> thegeek: is there valgrind (like) program for windows?
[22:11] <thegeek> I have used some in the past
[22:11] <thegeek> though I can't remember the names of any of them;P
[22:11] <thegeek> the big problem though is the lack of "windows" debug symbols
[22:11] <thegeek> makes it hard to use any windows tools
[22:12] <thegeek> I'll use gdb etc tomorrow
[22:14] <cone-265> ffmpeg.git 03Stefano Sabatini 07c7065f1f8953: lavfi/sendcmd: release file in case of failed allocation
[22:14] <cone-265> ffmpeg.git 03Stefano Sabatini 07bd6240e7e9f3: lavfi/sendcmd: move buf init() variable to internal scope where it is used
[22:14] <cone-265> ffmpeg.git 03Stefano Sabatini 078b03cd3cd7a3: lavd/lavfi: add graph_file option
[22:15] <thegeek> and hey, it seems you can use valgrind on windows as well through mingw
[22:54] <cone-265> ffmpeg.git 03Carl Eugen Hoyos 072fdc6f355cfc: Do not detect mov with maximum score if the atom size is too small.
[23:25] <saste> michaelni, any: ping on lavc/utils: propagate or return meaningful error codes in avcodec_open2()
[23:54] <llogan> saste: "enabled, add" is a comma splice. I think.
[23:55] <saste> llogan: i like comma splices
[23:56] <llogan> everybody likes comma splices.
[23:57] <saste> everybody likes comma splices, they give spice to boring single sentences.
[23:58] <llogan> He who controls the Splice, controls the universe!
[23:59] <llogan> why did my server turn off by itself?
[23:59] <saste> he's tired
[23:59] <saste> to serve
[00:00] --- Wed Oct 24 2012