[Ffmpeg-devel-irc] ffmpeg-devel.log.20131108

burek burek021 at gmail.com
Sat Nov 9 02:05:02 CET 2013


[00:00] <burek> llogan, it detects rss document change
[00:00] <burek> probably something was changed at that moment, although it shows only last 5 items
[00:01] <burek> so it appears as if it is repeating the same thing
[00:01] <burek> i'll investigate in logs
[00:01] <burek> and improve the behavior if possible
[00:02] <llogan> burek: thanks. now i owe you the beer.
[00:02] <burek> +1 :)
[00:02] <cone-764> ffmpeg.git 03Diego Biurrun 07master:86f910806b2c: Remove #undefs for formerly forbidden system functions
[00:02] <cone-764> ffmpeg.git 03Michael Niedermayer 07master:175e689cc5de: avfilter/graphparser: zero filter_ctx in case of deallocation in create_filter()
[00:02] <cone-764> ffmpeg.git 03Michael Niedermayer 07master:48f37664a8b5: Merge remote-tracking branch 'qatar/master'
[00:44] <cone-764> ffmpeg.git 03Michael Niedermayer 07master:b7fea3daabc7: libavcodec/dvdsubdec: remove exit() call in debug code
[05:57] <anshul_> i was trying to run fate and got an following error
[05:57] <anshul_> tests/fate.sh doc/fate_config.sh
[05:58] <anshul_> fatal: premature end of pack file, 300 bytes missing
[05:58] <anshul_> fatal: index-pack failed
[05:58] <anshul_> Error fetching source
[06:23] <anshul_> I deleted the fate_work directory and tried to run again, and got again an error
[06:24] <anshul_> [anshul at daku_daddy ffmpeg]$ tests/fate.sh doc/fate_config.sh
[06:24] <anshul_> fatal: pack has 9 unresolved deltas
[06:24] <anshul_> fatal: index-pack failed
[06:24] <anshul_> Error fetching source
[06:24] <anshul_> rm: cannot remove fate_work/fate.lock: No such file or directory
[06:25] <anshul_> done git pull for  fatal: pack has 9 unresolved deltas
[06:33] <anshul_> I am unable to run fate, can anybody help me
[06:33] <anshul_> [anshul at daku_daddy ffmpeg]$ tests/fate.sh doc/fate_config.sh
[06:33] <anshul_> tests/fate.sh: line 107: fate_work/src/version.sh: No such file or directory
[06:33] <anshul_> rm: cannot remove fate_work/fate.lock: No such file or directory
[06:34] <anshul_> version.sh is present at ./fate_work/fate_work/src/version.sh
[06:36] <anshul_> I think relative paths confuse the fate, trying with absolute path
[07:17] <anshul_> Still unable to run fate
[07:19] <anshul_> cat: /home/anshul/Software/ffmpeg/fate_work/build/tests/data/fate/*.rep: No such file or directory
[07:22] <anshul_> actually /tests/data/fate is at  /home/anshul/Software/ffmpeg/
[07:22] <anshul_> no idea what to try next
[07:22] <anshul_> can any one share there fate_config file to cross verify
[07:24] <anshul_> if I remove all of the work directory , then it try to makedir with empty name
[07:27] <anshul_> I dont think cpoying tests/data/fate/ in fate_work is an good idea
[07:28] <ubitux> did you look at https://ffmpeg.org/fate.html ?
[07:28] <ubitux> did you check the config.log of your fate run?
[07:36] <anshul_> I was following the same link
[07:37] <anshul_> I dont know what to check in config.log 
[07:39] <ubitux> if it fails or no
[07:40] <ubitux> http://lucy.pkh.me/fate-configs/
[07:40] <ubitux> my configs are here
[07:40] <ubitux> http://lucy.pkh.me/run.sh
[07:40] <ubitux> and this is my fate run script
[07:41] <ubitux> (which is located in ~/fate/run.sh, configs in ~/fate/configs, source in ~/fate/ffmpeg)
[08:08] <anshul_> hey ubitex in you configuration you have kept repo as file:/home/ux/fate/ffmpeg but in mine cfg i have kept git://source.ffmpeg.org/ffmpeg.git; does it make sense to make it git
[08:08] <anshul_> or i should also change the path to local git repository at my pc
[08:12] <anshul_> yeh command worked tests/fate.sh doc/fate_config.sh with echo$? = 0
[08:13] <anshul_> but when i check the  https://ffmpeg.org/fate.html how to confirm that this fate is run y me
[08:14] <ubitux> it's the local path where the sources are pulled
[08:15] <ubitux> the script pulls the sources at regular interval in that directory
[08:16] <ubitux> or well, before each instance run actually
[08:17] <ubitux> on fate.ffmpeg.org, to check if it's yours, you click on the tests, you'll get all kind of info
[09:00] <anshul_>   1 slot=x86-openSuse-gcc-4.7-dakudaddy      # some unique identifier
[09:00] <anshul_>   2 repo=/home/anshul/Software/ffmpeg/  # the source repository
[09:00] <anshul_>   3 samples=fate-suite/                      # path to samples directory
[09:00] <anshul_>   4 workdir=/home/anshul/Software/ffmpeg/fate_work/   # directory in which to do all the work
[09:00] <anshul_>   5 #fate_recv="ssh -T fate at fate.ffmpeg.org" # command to submit report
[09:00] <anshul_>   6 comment=First_time_fate                    # optional description
[09:00] <anshul_>   7 build_only=     # set to "yes" for a compile-only instance that skips tests
[09:00] <anshul_>   8 
[09:00] <anshul_>   9 # the following are optional and map to configure options
[09:00] <anshul_>  10 arch=
[09:00] <anshul_>  11 cpu=
[09:00] <anshul_>  12 cross_prefix=
[09:00] <anshul_>  13 as=
[09:00] <anshul_>  14 cc=
[09:00] <anshul_>  15 ld=
[09:00] <anshul_>  16 target_os=
[09:00] <anshul_>  17 sysroot=
[09:00] <anshul_>  18 target_exec=
[09:00] <anshul_>  19 target_path=
[09:00] <anshul_>  20 target_samples=
[09:00] <anshul_>  21 extra_cflags=
[09:00] <anshul_>  22 extra_ldflags=
[09:00] <anshul_>  23 extra_libs=
[09:00] <anshul_>  24 extra_conf=     # extra configure options not covered above
[09:00] <anshul_>  25 
[09:00] <anshul_>  26 #make=          # name of GNU make if not 'make'
[09:00] <anshul_>  27 makeopts=       # extra options passed to 'make'
[09:00] <anshul_>  28 #tar=           # command to create a tar archive from its arguments on stdout,
[09:00] <anshul_>  29                 # defaults to 'tar c'
[09:00] <anshul_> This my config file
[09:00] <anshul_> but i dont get anything on http://fate.ffmpeg.org/?dsort=npass
[09:13] <ubitux> anshul_: please use pastebin/pastie/... when it goes up to 3 lines&
[09:14] <anshul_> ok
[09:17] <anshul_> in configure.log it shows that
[09:17] <anshul_> Out of tree builds are impossible with config.h in source dir.
[09:23] <ubitux> cleanup your source dir
[09:25] <anshul_> you mean make distclean
[09:26] <michaelni> " fatal: pack has 9 unresolved deltas" /  "fatal: index-pack failed" <-- sounds like some problem with git
[09:26] <anshul_> I am trying to run your scripts now, with making enviorment similar to your
[09:28] <anshul_> @michaelni i pulled the git again, and that error was not seen again
[09:42] <ubitux> anshul_: yes, or git clean -fdx
[10:02] <plepere> hello. coding assembly for HEVC decoding (or at least documenting myself to). Should I do general x86 functions and add options for x64 support or can I just go with x64 from the start ?
[10:03] <ubitux> plepere: as i said, you can go for x64 from the start, and later add 32
[10:03] <plepere> ok thanks. I had to go last time so I didn't see the answer when I asked last time. :)
[10:04] <nevcairiel> It would be great to have 32-bit eventually
[10:04] <nevcairiel> its still in wide use
[10:04] <plepere> of course
[10:05] <plepere> same for SSE instruction use ? start with SSE4 first and support old MMX only processors after ?
[10:05] <iive> sse2 at least
[10:05] <nevcairiel> personally i wouldnt bother much with mmx
[10:05] <plepere> ok
[10:06] <nevcairiel> If you want to build SSSE3/SSE4 version first, and add a SSE2 compat later, thats also ok, if you even want to deal with mmx compat ... your choice :)
[10:07] <plepere> I think I'd rather try my hand on ARM optimization if I get the choice. :) so much to learn
[10:08] <plepere> it's what I love about this job : I'm always learning. Forever a student !
[11:15] <anshul_> 62622 Test filter-metadata-ebur128 failed. Look at tests/data/fate/filter-metadata-ebur128.err for details.
[11:15] <anshul_> 62623 make: *** [fate-filter-metadata-ebur128] Error 1
[11:15] <anshul_> 62624 make: Target `fate' not remade because of errors.
[11:15] <anshul_> Still unable to run fate
[11:15] <anshul_> these are last three line in test.log
[11:18] <ubitux> "Look at tests/data/fate/filter-metadata-ebur128.err for details."
[11:19] <anshul_> there is no file named filter-metadata-ebur128.err
[11:21] <anshul_> I there is no file named *.err even
[11:22] <anshul_> Is there any clue i can get from the report generated in work folder
[11:24] <anshul_> from first line of report fate:0:20131108100824:x86-openSuse-gcc-4.7-dakudaddy:N-57965-gb7fea3d:2:error testing:First_time_fate
[11:24] <nevcairiel> simply run that failing test manually to find out more
[11:24] <nevcairiel> make fate-filter-metadata-ebur128
[11:55] <anshul_> [anshul at daku_daddy ffmpeg]$ make fate-filter-metadata-ebur128
[11:55] <anshul_> Makefile:2: config.mak: No such file or directory
[11:55] <anshul_> Makefile:47: /common.mak: No such file or directory
[11:55] <anshul_> Makefile:90: /libavutil/Makefile: No such file or directory
[11:55] <anshul_> Makefile:90: /library.mak: No such file or directory
[11:55] <anshul_> Makefile:167: /doc/Makefile: No such file or directory
[11:55] <anshul_> Makefile:168: /tests/Makefile: No such file or directory
[11:55] <anshul_> make: *** No rule to make target `/tests/Makefile'.  Stop.
[11:55] <anshul_> I dont why its kooking in /doc tarher it should look fordoc
[11:55] <ubitux> this is because the configure wasn't run from this directory
[12:11] <sspiff> can I add CCFLAGS to ffmpeg easily?
[12:11] <sspiff> I'm building for a platform that needs a bit of tweaking :)
[12:11] <sspiff> I can't find anything in ./configure --help, so I'm currently using CCFLAGS=foo make
[12:12] <anshul_> @ubitux from where should i run the make 
[12:12] <JEEB> sspiff, --extra-cflags="-I/ikaruga/kou/include" as an example
[12:13] <anshul_> I am still not getting, from where fate.sh had run the configure.
[12:19] <anshul_> @nevcairiel i tried by configuring in ffmpeg directory and tried the same command but now its say make: *** No rule to make target `fate-filter-metadata-ebur128'.  Stop.
[12:21] <cone-424> ffmpeg.git 03Hendrik Leppkes 07master:019247bdc326: avformat/mov: only force parsing for video tracks if stss is empty
[12:21] <ubitux> you need to run the tests from where you ran the configure and make.
[12:21] <iive> anshul_: debug with  `bash -x fate.sh` , it would output every command as it is executed. 
[12:29] <anshul_> @ubitux i didn't run any configure or make, they are done by fate.sh; 
[12:30] <anshul_> @live that was good hack for scripts :)
[12:55] <BBB> durandal_1707: yeah, you want to do vp9-mt?
[13:03] <durandal_1707> BBB: it looks very complicated
[13:14] <BBB> it isn't
[13:14] <BBB> you've done mt before
[13:14] <BBB> it's exactly the same, with some cross-frame progress tracking, which is in whatever unit you like
[13:14] <BBB> it's just a few lines more
[13:16] <BBB> I can hep if you're interested, that is, you'll code and I can provide pointers of stuff to watch out for etc.
[13:16] <BBB> basically it involves one piece of code refactoring, and then the regular framt-mt code insertions of init_done, report_progress, wait_progress
[13:17] <anshul_> I cleaned up every thing and end up with command make filter-metadata-ebur128 -f ../ffmpeg/Makefile
[13:18] <ubitux> anshul_: what the hell are you trying to do?
[13:19] <ubitux> anshul_: create a build directory, run the configure and make from that directory
[13:19] <ubitux> then run the make fate-filter-metadata-ebur128
[13:19] <ubitux> (and make sure the SAMPLES variable is set)
[13:19] <anshul_> ok i will try again;
[13:20] <durandal_1707> BBB: refactoring?
[13:40] <BBB> durandal_1707: so vp9 currently decodes the bitstream in-line with the reconstruction
[13:40] <BBB> durandal_1707: i.e. do { decode block in bitstream - reconstruct it } while (!done)
[13:40] <durandal_1707> so basically i need to rewrite decoder?
[13:40] <BBB> no
[13:41] <BBB> they are separate functions
[13:41] <BBB> they're just called after each other
[13:41] <BBB> what you need to do is to extract that so it does 2 cycles
[13:41] <BBB> do { decode block - store it in a temp mem location } while (!done) thread_init_done; do { reconstruct } while (more blocks in temp mem location)
[13:42] <BBB> I can actually do that if you feel that's troublesome
[13:52] <anshul_> failed again with fate; tried below command
[13:52] <anshul_> ../ffmpeg/configure --samples=../fate-suite/
[13:52] <anshul_> make SAMPLES=../fate-suite/
[13:52] <anshul_> make filter-metadata-ebur128 SAMPLES=../fate-suite/
[13:52] <anshul_> make: *** No rule to make target `filter-metadata-ebur128'.  Stop
[13:53] <nevcairiel> make fate-filter-metadata-ebur128
[13:53] <nevcairiel> missing a fate-
[13:54] <anshul_> same result :(
[13:54] <anshul_> [anshul at daku_daddy fate]$ make fate-filter-metadata-ebur128 SAMPLES=../fate-suite/
[13:54] <anshul_> make: *** No rule to make target `fate-filter-metadata-ebur128'.  Stop.
[13:55] <nevcairiel> just give up on the out of tree build then :p
[13:55] <nevcairiel> you cna get a list of fate targets with make fate-list, i think
[13:57] <nevcairiel> oh, and ebur128 is gpl, so you need to configure with --enable-gpl, or it wont be available
[13:59] <anshul_> I am running fate second time; few months before also i gave up with some error, this time i am determined to make this work.
[14:00] <ubitux> anshul_: if that tests fail it's not really a problem
[14:00] <anshul_> at least it is compiling.
[14:00] <ubitux> it will be reported on fate
[14:00] <ubitux> and we will be able to see the problem
[14:00] <ubitux> what configuration are you setuping?
[14:01] <ubitux> if it's an msvc instance, this test is known to be broken because of a msvc bug
[14:01] <nevcairiel> which is fixed in msvc 2013, hooray
[14:02] <anshul_> i have not posted the ssh public key yet, so it would not be reported 
[14:03] <anshul_> my configuration were just plane ../ffmpeg/configure --samples=../fate-suite/ --enable-gpl
[14:03] <ubitux> i mean what plateform?
[14:04] <anshul_>  version-x86-openSuse-gcc-4.7
[14:10] <ubitux> special cpu?
[14:11] <anshul_> no ts normal intel corei3
[14:12] <anshul_> In my company 100 of employees use the same cpu
[14:14] <anshul_> @nevcairie --enable-gpl fixed the issue; i am trying to run fate again with passsing extra_conf=--enable-gpl and waiting for error report
[14:16] <anshul_> I think fate shoul not test those test files whose codecs are not enable in configuration. can this issue reported as bug
[14:21] <ubitux> anshul_: fate has --enable-gpl by default
[14:22] <anshul_> make was successfull at make filter-metadata-ebur128 SAMPLES=../fate-suite/
[14:22] <anshul_> but why it was failing in fate
[14:22] <anshul_> but why it was failing in fate report
[14:30] <anshul_> @ubitux i dont think fate has --enable-gpl by default
[14:31] <ubitux> then look at the file tests/fate.sh
[14:31] <ubitux> configure() function
[14:32] <anshul_> yaaa 8)
[14:49] <durandal_1707> BBB: what about tile threading?
[15:00] <anshul_>   1 ffprobe version N-57966-g019247b Copyright (c) 2007-2013 the FFmpeg developers
[15:00] <anshul_>   2   built on Nov  8 2013 19:17:56 with gcc 4.7 (SUSE Linux)
[15:00] <anshul_>   3   configuration: --prefix=/home/anshul/Software/fate/fate_work/install --samples=fate-suite --enable-gpl --enable-memory-poisoning --enable-avresample --    enable-gpl
[15:00] <anshul_>   4   libavutil      52. 52.100 / 52. 52.100
[15:00] <anshul_>   5   libavcodec     55. 41.100 / 55. 41.100
[15:00] <anshul_>   6   libavformat    55. 21.100 / 55. 21.100
[15:00] <anshul_>   7   libavdevice    55.  5.100 / 55.  5.100
[15:00] <anshul_>   8   libavfilter     3. 90.102 /  3. 90.102
[15:00] <anshul_>   9   libavresample   1.  1.  0 /  1.  1.  0
[15:00] <anshul_>  10   libswscale      2.  5.101 /  2.  5.101
[15:00] <anshul_>  11   libswresample   0. 17.104 /  0. 17.104
[15:00] <anshul_>  12   libpostproc    52.  3.100 / 52.  3.100
[15:00] <anshul_>  13 [Parsed_amovie_0 @ 0xa7b9a20] Failed to avformat_open_input 'fate-suite/filter/seq-3341-7_seq-3342-5-24bit.flac'
[15:00] <anshul_>  14 [lavfi @ 0xa7b8980] Error initializing filter 'amovie' with args 'fate-suite/filter/seq-3341-7_seq-3342-5-24bit.flac'
[15:00] <anshul_>  15 amovie=fate-suite/filter/seq-3341-7_seq-3342-5-24bit.flac,ebur128=metadata=1: No such file or directory
[15:00] <durandal_1707> can you learn to use pastebin or similar?
[15:00] <anshul_> sorry
[15:01] <durandal_1707> iirc, amovie filter needs full path
[15:01] <anshul_> found the error file
[15:03] <anshul_> actually i was not trying to use amovie, i was trying to run fate on my system successfully without any error
[15:04] <anshul_> so for this error do i need to file bug, because i think while passing file name some part has been added ",ebur128=metadata=1"
[15:05] <durandal_1707> nope, everything is explained in documentation
[15:05] <durandal_1707> and from error i do not see that extra stuff have been added to filename
[15:05] <anshul_> so do i need to change the fate test case for amovie
[15:05] <durandal_1707> also you are on wrong channel
[15:06] <durandal_1707> anshul_: what are you actually doing?
[15:07] <anshul_> i am analysing ffmpeg.
[15:09] <anshul_> last time one of my patch was rejected due to fate, so today i was just getting the basic understanding to use fate.
[15:10] <anshul_> I thought fate is ffmpeg_developer tools ffmpeg users might not know how to use it
[15:10] <durandal_1707> so what is patch about?
[15:12] <anshul_> actually it was about an memory leak, long time back, this time i am just solving 2077 ffplay crash
[15:12] <anshul_> This time i have not submitted the patch till it work properly at my place
[15:15] <anshul_> to test that my change work properly, i have to make the refrence system work, that was what i am trying to do.
[15:15] <anshul_> but fate took my all day.
[15:16] <durandal_1707> that is strange - it should not happen
[15:16] <durandal_1707> you sure you have fate-suite directory set?
[15:16] <durandal_1707> and all files fetched?
[15:16] <nevcairiel> you dont need to setup a full fate box just for testing, the important point is to simply run make fate and see if your changes broke anything :)
[15:17] <durandal_1707> iirc this is is also mentioned in documentation
[15:17] <durandal_1707> it looks like there are tests that fails ...
[15:20] <anshul_> yes there are some test failing, talk to you guys tommarow, already late night!
[15:26] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:4c439f6e3999: avformat/idcin: check the decompressed frame size during probing
[15:29] <ubitux> i wonder what the hell that guy is trying to do
[15:31] <durandal_1707> fix ffplay crash
[15:44] <michaelni> ods15, there are 2 patches for libnut on nut-devel
[15:47] <michaelni> ods15, do you want to review/apply them ? should i apply them ? do you still maintain libnut ? should we try to find a new maintainer? ...
[15:48] <michaelni> http://lists.mplayerhq.hu/pipermail/nut-devel/2013-October/date.html
[15:50] <durandal_1707> michaelni: isn't libnut separate project and have nothing to do with wrapper?
[15:55] <michaelni> durandal_1707, yes, the patches are for libnut not the wrapper, my note here is probably strictly speaking off topic for ffmpeg
[15:58] <durandal_1707> iirc i do not think ods15 is maintainer of nut code, just of wrapper, correct me if i'm wrong
[15:59] <durandal_1707> also have fourcc been synced with code in ffmpeg?
[16:09] <michaelni> durandal_1707, ods15 is the author of libnut, last comit to it seems from 2008, also dont mix up the nut spec, ffmpegs nut implementation, the libnut wraper in ffmpeg and libnut itself
[16:10] <durandal_1707> i do not mix stuff
[16:12] <wm4> why is there a libnut wrapper in ffmpeg?
[16:12] <durandal_1707> apparently ods15 is/was maintainer of libnut and libnut wrapper
[16:12] <durandal_1707> wm4: because
[16:13] <durandal_1707> how otherwise would you test it?
[16:14] <wm4> whatever else uses libnut?
[16:26] <av500> nothing
[17:12] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:f8a9cf77040e: avformat/lvfdec: check stream count during probing
[18:01] <cone-424> ffmpeg.git 03Diego Biurrun 07master:0338c396987c: dsputil: Split off H.263 bits into their own H263DSPContext
[18:01] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:a30f7918b52d: Merge commit '0338c396987c82b41d322630ea9712fe5f9561d6'
[18:09] <cone-424> ffmpeg.git 03Diego Biurrun 07master:cb56c3961b18: h263dsp: K&R formatting cosmetics
[18:09] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:e93523a06f27: Merge remote-tracking branch 'qatar/master'
[20:23] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:4f0e85517f5d: avformat/icodec: reduce score returned on probing
[20:24] <cone-424> ffmpeg.git 03Paul B Mahol 07master:2145e8f19353: avcodec/vmdav: use av_freep() and reset unpack_buffer_size
[20:24] <cone-424> ffmpeg.git 03Paul B Mahol 07master:b8f276e64574: avcodec/mimic: use av_freep() and reset swap_buf_size
[22:40] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:32584667ea51: tools/probetest: use named constant for the maximum number of formats
[22:40] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:6b4f58d448f4: tools/probetest: also print the time the probe functions needed
[22:41] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:a26e83da6c96: avformat/mxfdec: optimize probing
[23:42] <cone-424> ffmpeg.git 03Kostya Shishkov 07master:5462dde1947e: metasound: Fix error message argument
[23:43] <cone-424> ffmpeg.git 03Michael Niedermayer 07master:ac4447b6ec4e: Merge commit '5462dde1947e0ecfcb2db99ae29ce9dee8933b45'
[23:53] <ubitux> michaelni: ah sorry, just get it for AS
[23:53] <ubitux> will re-send a patch
[00:00] --- Sat Nov  9 2013


More information about the Ffmpeg-devel-irc mailing list