[Ffmpeg-devel-irc] ffmpeg-devel.log.20130422

[burek]

[00:47] <ubitux> saste: i'm going to test your filter :)
[00:49] <saste> ubitux, hue is still borken
[00:49] <saste> (see patch)
[00:51] <ubitux> ok, just a moment
[00:51] <ubitux> i'm writing something cool :)
[00:53] <saste> ubitux, what?
[00:53] <ubitux> saste: http://pastie.org/7693879
[00:53] <ubitux> can you try this?
[00:54] <ubitux> (now you should have various shell features, including history etc, and you can easily add some completion routines)
[00:57] <ubitux> i'm looking at the hue patch now
[00:59] <saste> uhm i wonder it could mess with the 3.x/2.x thing
[00:59] <ubitux> :*
[00:59] <ubitux> :(
[00:59] <saste> i have python binding for 2.x, but locally installed the 3.x API
[01:00] <saste> i suppose i'll downgrade the API
[01:00] <ubitux> i have zmq3 api, and python2 and 3 bindings here
[01:00] <ubitux> i can adjust the code if necessary
[01:00] <ubitux> (to python3)
[01:00] <saste> but should we prefer 2.x over 3.x?
[01:01] <ubitux> you mean for zmq api?
[01:01] <saste> 3.x is relatively new, and thus not included in a lot of distros
[01:01] <saste> yes
[01:01] <ubitux> i have only zeromq3 packaged here
[01:01] <saste> ubitux, what distro?
[01:01] <ubitux> archlinux
[01:01] <ubitux> http://pastie.org/7693937
[01:02] <saste> well i don't know
[01:02] <ubitux> it's new code, you can request bleeding edge technos since you don't have to care about retro compat
[01:02] <ubitux> and if necessary, you can add some additionnal compatible code
[01:03] <ubitux> but who cares about debian anyway ;)
[01:03] <saste> ubitux, no i don't like that
[01:03] <saste> i mean both versions, it's a mess to test
[01:03] <ubitux> the latest stable only, 3.
[01:03] <saste> my tendency is to favor 3.x and be done with that
[01:04] <ubitux> same.
[01:04] <saste> now I don't know what I have to do in order to install the required python modules
[01:04] <saste> that's why i wrote a c client
[01:05] <ubitux> :(
[01:05] <ubitux> but my script is cool :(
[01:06] <saste> :)
[01:06] <saste> well i'll sleep and let you test the shiny zmq code
[01:07] <ubitux> i'm on the hue review now
[01:07] <ubitux> 'night :)
[01:07] <saste> the main problem right now is that it used hardcoded buffer size
[01:07] <saste> the zmq message API should allow to overcome that limitation I guess
[01:08] <saste> ok, thanks for reviewing
[02:34] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:de9fbf3dc41a: indeo4: implement haar 8x1 and 1x8 transforms
[02:43] <Compn> where is maksym and his indeo patches ?
[02:52] <highgod> Hi, I want to ask a question, what the usage of mux and demux?
[02:52] <highgod> what can they do?thanks
[02:53] <Compn> mux creates the container , demux takes apart the container and sends the data to codecs
[02:56] <highgod> Compn:thanks
[02:57] <highgod> another question, how can I do the benchmark, you have told me, but I forgot it, sorry.
[03:09] <BBB-> someone broke msvc again
[03:10] <ubitux> it's a conspiracy
[03:12] <ubitux> i guess it's 76c7277385120741914168d02b57a7a1acf87a9b
[03:27] <Compn> highgod : something with oprofile and start_timer stop_timer in the code. but i dont remember exactly. i think its written down in the docs somewhere
[03:28] <highgod> OK
[03:28] <Compn> its in timer.h file
[10:55] <cone-735> ffmpeg.git 03Luca Barbato 07master:1c88617b9421: riff: K&R formatting cosmetics
[10:55] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:e381cc3e4ea6: Merge commit '1c88617b9421c9af8ae961c05d2381b19c585026'
[11:51] <cone-735> ffmpeg.git 03Luca Barbato 07master:8e329dba378c: riff: check for eof if chunk size and code are 0
[11:51] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:369e5dcfc530: Merge commit '8e329dba378cef0ff6400c7df9c51da167d5a1f0'
[12:03] <dekarl> hi, has this fix ever been commited? I'm not seeing it in master http://ffmpeg.org/trac/ffmpeg/ticket/1240
[12:25] <durandal_1707> michaelni: are you in middle of something?
[12:26] <av500> otherwise he would be on the edge of the world
[12:33] <cone-735> ffmpeg.git 03Peter Ross 07master:e544782623ca: riff: Move guid structs and helper functions into riff
[12:33] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:52e40155a8e1: Merge commit 'e544782623caf1eb4b2429e9c4f35d6b627accba'
[12:34] <durandal_1707> av500: something like that
[12:41] <durandal_1707> michaelni: what you think about sh4 code removal?
[12:41] <cone-735> ffmpeg.git 03Peter Ross 07master:a3bf80a5aea5: riff: Perform full lookup on WAVEFORMATEXTENSIBLE subformat guid
[12:41] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:05deb05fe9ce: Merge commit 'a3bf80a5aea577663354bd538e345b283bcb73de'
[12:46] <J_Darnley> dekarl: Yes, if got log is to be believed
[12:46] <J_Darnley> 364c71c80e9124a2624e3bfeb8e84c5cddeda222 says it fixes 1240
[12:47] <J_Darnley> ... that was supposed to be "git log"
[12:52] <durandal_1707> dekarl: if bug is still present, reopen bug and upload file with steps how to reproduce bug
[12:52] <michaelni> durandal_1707, do we have someone who wants to maintain sh4 ?
[12:52] <cone-735> ffmpeg.git 03Peter Ross 07master:516089d5d86b: riff: Add ACTRAC3+ guid
[12:52] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:51286aaffcad: Merge commit '516089d5d86b6809a9458f9cf171f2bc9a8e2a25'
[12:55] <ubitux> btw, i'll apply the lavfi timeline tonight
[12:57] <durandal_1707> michaelni: so no maintainer -> rm -rf *
[12:58] <av500> sh4 is dead
[12:58] <av500> renesas told me
[12:58] <av500> they went arm
[13:01] <durandal_1707> av500: citation needed
[13:01] <cone-735> ffmpeg.git 03Luca Barbato 07master:b845f5e97b65: riff: Factor out WAVEFORMATEX parsing
[13:01] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:645d0384e85b: Merge commit 'b845f5e97b655de0a191f736594777fec9754cf5'
[13:02] <michaelni> durandal_1707, no maintainer -> code might break, and when it breaks without anyone willing to fix then removal is the only option
[13:02] <michaelni> otherwise ffmpeg wouldnt build/work anymore on sh4 ...
[13:05] <av500> durandal_1707: renesas guy to me at last embedded world
[13:05] <av500> when I asked him for HW to keep sh4 alive
[13:08] <cone-735> ffmpeg.git 03Anton Khirnov 07master:9bfc6e02bae9: afifo: fix request_samples on the last frame in certain cases
[13:08] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:e4c0f258d38a: Merge commit '9bfc6e02bae9de354fb9ba09a8a140e83eeadf7d'
[13:10] <dekarl> J_Darnley: thanks for finding the commit
[13:10] <dekarl> durandal_1707: looks like I have to rework our transcoder to look for the frame_size/duration in the new place. http://code.mythtv.org/trac/ticket/2077#comment:42
[13:13] <cone-735> ffmpeg.git 03Diego Biurrun 07master:311a592dfcfc: x86: Remove some duplicate function declarations
[13:13] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:2288c7768970: Merge remote-tracking branch 'qatar/master'
[14:28] <cone-735> ffmpeg.git 03Paul B Mahol 07master:0327400ee31f: doc/filters: add examples for stereo3d
[14:28] <cone-735> ffmpeg.git 03Paul B Mahol 07master:2f11aa141a01: lavfi: remove mp=down3dright
[15:07] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:80d4d3a41de6: fate: re-enable af_join test
[16:05] <cone-735> ffmpeg.git 03highgod0401 07master:6d7ec4fc2ef7: lavu/opencl:optimize compile kernel operation
[17:40] <burek> hi guys :)
[17:41] <burek> i don't want to bug you, but does anyone know what to answer to this guy: http://ffmpeg.gusari.org/viewtopic.php?f=11&t=897
[18:34] <cone-735> ffmpeg.git 03Diego Biurrun 07master:78fa0bd0f706: x86: cavs: Put mmx-specific code into its own init function
[18:34] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:f84e373797eb: Merge commit '78fa0bd0f7067868943c0899907e313414492426'
[18:38] <cone-735> ffmpeg.git 03Diego Biurrun 07master:fce99322b0cc: h264: Drop unused variable
[18:38] <cone-735> ffmpeg.git 03Diego Biurrun 07master:a0d5ca4f0a8e: mxfenc: Use correct printf format specifier for int64_t
[18:38] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:74e1aa17df99: Merge commit 'a0d5ca4f0a8e2c34d784d503a12af6303424ac6a'
[18:45] <cone-735> ffmpeg.git 03Diego Biurrun 07master:b4ad7c54c878: x86: cavs: Refactor duplicate dspfunc macro
[18:45] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:430d69c9421f: Merge commit 'b4ad7c54c878dead7dfa4838b912a530c1debe85'
[19:05] <cone-735> ffmpeg.git 03Diego Biurrun 07master:0c15a9aa7e16: sh4: Remove dubious aligned dsputil code
[19:06] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:7a556ebccfd6: Merge commit '0c15a9aa7e1654a19144eb594f9639a57fd47482'
[20:14] <cone-735> ffmpeg.git 03Nicolas Bertrand 07master:c81a70638116: JPEG 2000 decoder for DCinema
[20:14] <cone-735> ffmpeg.git 03Diego Biurrun 07master:f13888afcc5e: avcodec: Bump minor for JPEG 2000 decoder
[20:14] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:c1c2b0b33904: avcodec: rename jpeg2000 decoder to j2k
[20:15] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:e2e9bee2daef: Merge remote-tracking branch 'qatar/master'
[20:36] <mateo`> hello guys :)
[20:37] <mateo`> testing the brand new jpeg2000 decoder, is there a way to convert xyz colorspace to other format ?
[20:37] <ubitux> mateo`: there is a pending hack filter to do the convert
[20:37] <ubitux> it doesn't apply anymore though
[20:37] <ubitux> the other side may be working on it, but they prefer not to answer me when i asked
[20:38] <mateo`> my sudden excitement has just vanished :(
[20:40] <mateo`> anyway the decoder does not like the SD samples i have
[20:40] <nevcairiel> its only optimized for DCinema jpeg2000 profiles
[20:43] <j-b> indeed
[20:44] <j-b> and it's currently slow
[20:44] <michaelni> slower than the old j2k decoder ?
[20:44] <j-b> probably, because it has 0 asm in it
[20:45] <j-b> but, it will be faster
[20:45] <j-b> and it is in the code, so easier to leverage ASM skillz
[20:48] <nevcairiel> its a brand new decoder, clean and working implementation is the first goal, optimizing can start now
[20:48] <nevcairiel> the dwt code can probably be simded
[20:50] <j-b> indeed
[20:50] <j-b> And yes, so far, it is XYZ DCP profiles, because we need it...
[20:51] <ubitux> we don't yet optim for libavcodec/j2k_dwt.c i guess?
[20:51] <michaelni> its based on kamil nowosads decoder like ours, so iam not sure "brand new" is a completely correct description
[20:51] <ubitux> it's the exact same basis, the guy wasn't aware that is was upstream here
[20:52] <ubitux> and of course no one told him when he start working on it :p
[20:53] <j-b> Well, this one works. Sorry to be harsh, but I can't say the same for the other implementations
[20:54] <michaelni> j-b, no harshness here, we want the best code ...
[20:55] <j-b> michaelni: and for this work, we care about DCP.
[21:00] <michaelni> also nicolas code should contain all bugfixes from ffmpeg according to him
[21:01] <michaelni> so the "other" decoder can possibly be droped soon
[21:49] <j-b> michaelni: Yeah, I told him to look at it
[21:54] <cone-735> ffmpeg.git 03Michael Niedermayer 07master:2ae91c86f386: avcodec_get_context_defaults3: set codec_id
[22:11] <cone-735> ffmpeg.git 03Piotr Bandurski 07master:f4596e8bb6f7: mjpegdec: Add support for 4:1:1 YUV
[23:17] <ubitux> http://prog21.dadgum.com/170.html
[23:53] <michaelni> Is someone looking into the MSVC x86-64 aac-sbr noise compile failure ?
[23:54] Action: ubitux isn't
[23:54] <michaelni> iam no windows or msvc expert and dont have the compiler setup ...
[00:00] --- Tue Apr 23 2013