[Ffmpeg-devel-irc] ffmpeg-devel.log.20130331

[burek]

[00:14] <saste> michaelni, ping on "lavu/samplefmt: add av_samples_alloc_array_and_pointers()"
[00:52] <cone-108> ffmpeg.git 03Michael Niedermayer 07master:34ea5f418ef8: h264_refs: check for multiple mismatching long term ref assignments
[01:23] <cone-108> ffmpeg.git 03Stephen Hutchinson 07master:8cd7aab1bb29: Use PIX_FMT_BGR24 for the AviSynth demuxer, instead of RGB24.
[10:02] <cone-3> ffmpeg.git 03Clément BSsch 07master:29667a2c2014: decimate: check for avctx ptr before closing.
[10:30] <neXyon> can avcodec_decode_audio3 still be used for planar audio sample formats? (for backward compatible code)
[10:38] <nevcairiel> i dont think it works for planar audio
[11:18] <saste> why are we still using av_buffersrc_add_ref?
[11:39] <cone-3> ffmpeg.git 03Matt Wolenetz 07master:1d6e618939c1: vp8: Fix pthread_cond and pthread_mutex leaks
[11:39] <cone-3> ffmpeg.git 03Derek Buitenhuis 07master:f2f57d16795e: doc/platform: Remove part about dllimport
[11:39] <cone-3> ffmpeg.git 03Michael Niedermayer 07master:f8201cc18b30: Merge remote-tracking branch 'qatar/master'
[12:03] <cone-3> ffmpeg.git 03Clément BSsch 07master:eb054a969362: fate/filter: use aperms to test volume filter.
[12:03] <cone-3> ffmpeg.git 03Clément BSsch 07master:4dae804d3c6d: lavfi/volume: use copy props helper instead of incomplete manual code.
[12:03] <cone-3> ffmpeg.git 03Clément BSsch 07master:01b2d1288ae3: lavfi/volume: use designated initializers for the sample_fmts declaration.
[12:20] <cone-3> ffmpeg.git 03Timothy Gu 07master:874173002e9e: doc/filters: fix width/height typo in crop filter.
[13:26] <ubitux> saste: any idea why devices options don't show up in a ffmpeg -h full?
[13:38] <saste> ubitux, they do
[13:38] <saste> or do you mean generic options?
[13:38] <saste> because they are the same as the format options
[13:41] <ubitux> well i don't see the v4l2 options here
[13:42] <ubitux> or alsa or whatever
[14:46] <saste> ubitux, ALSA, V4L2
[14:46] <saste> I agree it doesn't make much sense
[14:49] <ubitux> wtf
[14:49] <ubitux> thanks..
[15:08] <ubitux> saste: did you commit your option dump prettifying?
[15:08] <saste> ubitux, I don't think so
[15:09] <saste> but it was just a few reindent
[16:37] <BBB> Compn: no response yet, so no samples yet
[16:37] <BBB> michaelni: yeah, let's
[17:53] <saste> michaelni, av_calloc or av_realloc_f?
[17:53] <saste> since I don't need to clear the memory
[17:53] <saste> (even if from the performance POV they should be almost identical)
[17:54] <saste> no, i need to clear the memory after all...
[18:06] <cone-205> ffmpeg.git 03Stefano Sabatini 07master:39aad4346562: lavfi/overlay: fix debug PTS output in try_filter_frame()
[18:06] <cone-205> ffmpeg.git 03Stefano Sabatini 07master:5c73645d91e6: lavu/samplefmt: add av_samples_alloc_array_and_pointers()
[18:06] <cone-205> ffmpeg.git 03Stefano Sabatini 07master:66c92a157575: examples/resampling_audio: make use of av_samples_alloc_array_and_samples()
[18:07] <cone-205> ffmpeg.git 03Michael Niedermayer 07master:cb16d433a982: cmdutils_common_opts: fix loglevel help text as its inaccurate
[18:07] <cone-205> ffmpeg.git 03Michael Niedermayer 07master:5c13ae280fd5: cmdutils: allow forcing log line repeation
[18:16] <ubitux> mmh
[18:16] <ubitux> thx for this :)
[18:16] <ubitux> 'could be documented though
[20:18] <ubitux> i wonder if i shouldn't force the "pthread" mode for atomic op in helgrind/drd
[20:18] <ubitux> maybe it could make more sense
[20:29] <cone-205> ffmpeg.git 03Clément BSsch 07master:7d8ad6c1fa11: cmdutils: avtool -> fftool
[20:29] <cone-205> ffmpeg.git 03Clément BSsch 07master:999a46b12c8a: cmdutils: support filters in the help topic system.
[20:29] <cone-205> ffmpeg.git 03Clément BSsch 07master:d00dcb8c9f5f: opt: attempt to improve options dump output.
[20:31] <saste> ubitux, libavutil/opt output is not necessarily meant for ff* tools, that's why it shouldn't be coupled to the ff* syntax
[20:32] <ubitux> yes, in theory, the '-' should not appear at all
[20:33] <ubitux> but i'm pretty sure a lot of users will have no clue at all on how to set the options
[20:45] <cone-205> ffmpeg.git 03Michael Niedermayer 07master:ac126a98e56c: dvdec: fix lowres=3 with yuv 4:1:1
[20:55] <jezier> hi.. does ffmpeg have its own implementation of aes ctr (like for aes cbc)?
[21:06] <Daemon404> [14:32] <@ubitux> but i'm pretty sure a lot of users will have no clue at all on how to set the options
[21:06] <Daemon404> +1
[21:09] <Compn> jezier : check in avutils ?
[21:11] <jezier> Compn: i've done it already.. just want to be sure
[22:03] <saste> Daemon404, why don't you stick to the log.h doc?
[22:03] <saste> something which crashes the process
[22:07] <Daemon404> ..
[22:07] <Daemon404> http://ffmpeg.org/doxygen/trunk/log_8h_source.html
[22:07] <Daemon404> why is the site's doxy NEVER update
[22:07] <Daemon404> d
[22:07] <Daemon404> it fucks me over time and time again :<
[22:09] <Daemon404> looks like its just broken as hell.
[22:10] <Daemon404> anyway, a PANIC still makes no sense
[22:10] <michaelni> its the doxgen configuration most likely 
[22:10] <Daemon404> if we crash, it's unexpected
[22:10] <Daemon404> and we cant print a log msg
[22:10] <Daemon404> so wtf is the point?
[22:11] <michaelni> "Generated on Sun Mar 31 2013" 
[22:12] <saste> Daemon404, suppose you can print an assert failure check message in the log
[22:12] <Daemon404> except we don't
[22:13] <saste> right now i think it is printed on stderr, but i might be wrong
[22:13] <saste> having that message in the log may be useful, especially if you need to do post-mortem analysis of a process
[22:13] <Daemon404> anyway, still Do Not Appreciate the multitude of different explanations of PANIC i've been given
[22:13] <Daemon404> it's retarded.
[22:13] <saste> Daemon404, there is just one silly definition, the one in log.h
[22:14] <saste> as bad as it is, it is the reference
[22:14] <Daemon404> on the ML i got 3 diffrent ones
[22:14] <Daemon404> dont tell me its my own fault for READING THEN
[22:14] <Daemon404> THEM*
[22:14] <Compn> you dont build your own docs?
[22:14] Action: Compn trolls
[22:14] <Compn> :P
[22:15] <Daemon404> http://25.media.tumblr.com/tumblr_lwnom1wVZI1r7tx6oo1_500.jpg
[22:15] <Daemon404> ^ all i have to say
[22:15] Action: michaelni tries to figure out why doxy isnt looking as it should be
[22:15] <Compn> paper jam dipper dissaproves
[22:16] <saste> Daemon404, i referenced the log.h definition in my first mail, and tried an english translation suited for the manual
[22:17] <Daemon404> and that differered from michael's, and from a later one
[22:17] <saste> the interpretation of michael is a more liberal interpretation of the same definition
[22:17] <Daemon404> and i went and looked for log.h's docs
[22:17] <Daemon404> on the doxy
[22:17] <Daemon404> since well
[22:17] <Daemon404> it's doxygen documentation.
[22:18] <Daemon404> [16:17] <@saste> the interpretation of michael is a more liberal interpretation of the same definition <-- no.
[22:18] <Daemon404> it differs in more than just liberalism
[22:18] <Daemon404> tl;dr http://25.media.tumblr.com/tumblr_lwnom1wVZI1r7tx6oo1_500.jpg
[22:18] <saste> also that was not a "definition", was meant like a rationale for the idea behind it
[22:18] <Daemon404> and now i sea some 'repeat' support as added
[22:18] <Daemon404> with abslute no documentation
[22:19] <Daemon404> to loglevel nonetheless
[22:19] <Daemon404> shit like that just makes me think "why bother"
[22:19] <saste> 1. Only show critical errors leading to the crash of the process.
[22:19] <Daemon404> theyre just gonna shit all over any docs anyway
[22:20] <saste> 2.  show fatal errors which would lead to the process crash.
[22:20] <saste> to me they are the same thing
[22:20] <saste> which is far from:
[22:20] <Daemon404> thats not what michael wroye
[22:20] <Daemon404> if you read
[22:20] <Daemon404> like
[22:20] <Daemon404> actually read it.
[22:20] <saste>  errors which could affect the state of a user's system
[22:20] <saste> which is pretty vague
[22:20] <Daemon404> THAT IS WHAT MICHEL WROTE
[22:20] <Daemon404> sec ima go coypaste that
[22:21] <saste> SHERAZAM!
[22:21] <Daemon404> "while panic meant effects beyond that like failure to restore your
[22:21] <Daemon404> graphics cards registers after some direct access."
[22:21] <Daemon404> how is that not "the user's system"
[22:22] <saste> yes and this doesn't sound to me like a definition, especially considering that's not what log.h says
[22:22] <Daemon404> this is what michael wrote to explain panic.
[22:22] <Daemon404> literally.
[22:22] <Daemon404> taht is ALL of it.
[22:23] <Daemon404> yes yes
[22:23] <saste> to explain, not to define
[22:23] <Daemon404> your agrument is "you should have no read that and used log.h"
[22:23] <Daemon404> ...
[22:23] <Daemon404> for fuck's sake
[22:23] Action: michaelni meant "while panic meant effects beyond that of fatal"
[22:23] <Daemon404> i swear yu lto are just fuckign with me
[22:23] <Daemon404> either that
[22:23] <Daemon404> or have no concept of effective communicaiton at all
[22:23] <Daemon404> and accomplish the opposite
[22:24] <Daemon404> because that sentence i pasted couldnt /possibly/ be construed as user's system.
[22:24] <Compn> you guys do speak different languages 
[22:24] <saste> we should try with chinese
[22:24] <Compn> saste is from italy :)
[22:25] <Daemon404> anyway
[22:25] <Daemon404> unrelated: seeing the art of the docs i was working on immediately be shit on (via "repeat" being added with 0 docuentation_
[22:25] <Daemon404> now i have 0 motivatio nto help with docs
[22:25] <Daemon404> caue why bother
[22:25] <Daemon404> itll just be wrong in 2 days anywya
[22:25] <Daemon404> nobody wll update them.
[22:25] <Daemon404> and ffmpeg continues as normal
[22:26] <iive> from Daemon404 complaints it seems like ffmpeg is going to cause kernel packing the ffmpeg messes with video card registers.
[22:26] <Daemon404> iive, michael wrote the gfx card stuff. nt me.
[22:26] <Daemon404> http://ffmpeg.org/pipermail/ffmpeg-devel/2013-March/141480.html
[22:26] <Daemon404> ^
[22:26] <Compn> 'panic' like mplayer's 'core dump' message ? 
[22:27] <Compn> as in, repurpose common terms just to be difficult and elicit a troll or two
[22:27] <michaelni> Daemon404, i wanted to fix the repeat docs after your log patch would be commited
[22:27] <michaelni> because it would coinflict otherwise
[22:27] <Daemon404> it's a sentence above
[22:27] <Daemon404> no conflicts
[22:28] <Daemon404> (not to mention that this was mentioned precisely nowhere)
[22:30] <iive> Daemon404: you are the one who quote him. Don't worry I still have no idea what you (all) are talking about.
[22:32] <Compn> i certainly dont
[22:32] <Compn> but i hope Daemon404 does work on the docs
[22:32] <Compn> so it doesnt confuse other devs
[22:32] <Daemon404> tl;dr English comprehension levels around ffmpeg are rock bottom.
[22:32] <Compn> posting patches i 100x faster than discussing an issue, i've learned quickly
[22:33] <Compn> especially with some devs.
[22:33] <Compn> i > is
[22:33] <Daemon404> that's pretty shit.
[22:33] <Compn> its a fact
[22:33] <michaelni> Daemon404, fixed doxy @ ffmpeg.org
[22:33] <Daemon404> it seems to break every other day
[22:33] <Daemon404> it's mysterious
[22:34] <Compn> byproduct of losing some admin support
[22:36] <Daemon404> as an aside, it's a little non-obvious to have specific cpu name as "cpu flags"
[22:36] <michaelni> the bug existed in doxygen since 10 years
[22:36] <michaelni> doxygen config that is
[22:37] <Daemon404> ic
[22:37] <michaelni> 2003-03-06 10:21:57 +0000  712) STRIP_CODE_COMMENTS    = YES
[22:37] <Daemon404> michaelni, but why were the doxygen-style comments not showing up in the actual doxygen docs
[22:38] <Daemon404> not just the code
[22:38] <Daemon404> http://ffmpeg.org/doxygen/trunk/log_8h.html <---
[22:38] <Daemon404> theyre not there
[22:38] <Daemon404> and some stfuf is repeated a ton of times
[22:39] <Daemon404> oh it's there now
[22:39] <Daemon404> but on like the 12th occurrence of PANC
[22:39] <Daemon404> PANIC*
[22:42] <jojva> Daemon404, the last occurence specifically
[22:42] <Daemon404> yes
[22:55] <iive> Daemon404: I'm afraid your blabbering would be incomprehensible on any language. :P
[23:22] <cone-205> ffmpeg.git 03Michael Niedermayer 07master:a9f8182a9dbc: doxygen: dont strip code comments
[23:36] <michaelni> About the duplication of doxygen stuff, i cant reproduce this with my locally installed doxygen (which is older than whats on the server)
[23:36] <michaelni> if someone has an idea what could case it ...
[23:37] <michaelni> caUse
[23:55] <Compn> michaelni : anything strange in the build logs on mphq ?
[00:00] --- Mon Apr  1 2013