[Ffmpeg-devel-irc] ffmpeg-devel.log.20160703

[burek]

[01:24:18 CEST] <kierank> wow that text animation guy spammed me on facebook as well
[04:39:11 CEST] <cone-422> ffmpeg 03Carl Eugen Hoyos 07master:d5edb6c0483b: lavf/aiffdec: Support QDMC demuxing.
[13:50:42 CEST] <Chaz6> Just thought I'd mention if you haven't seen it, https://aomedia.googlesource.com/aom - AOMedia aka Alliance for Open Media
[15:32:41 CEST] <cone-629> ffmpeg 03Clément BSsch 07master:64c619369b3c: lavc/h264_slice: use sps directly when checking for invalid 8x8 inference
[17:36:54 CEST] <fritsch> michaelni: ping - sorry for bothering you: http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=ae215e2b42dd172f6bfd9f2dad44c9d5c11884bf in the else if branch shouldn't this be > buf_size, now it's clamped if it's smaller, but the mp_decode_frame would write buf_size into buf which is a problem if its larger than the buffer. Do I see that right?
[17:44:01 CEST] <nevcairiel> the second part was not changed
[17:44:05 CEST] <nevcairiel> the merge just shows it oddly
[17:45:23 CEST] <fritsch> still i wonder what happens if the following decode call happens and buf_size is larger as our buffer
[17:45:34 CEST] <fritsch> s/as/than/
[17:46:17 CEST] <fritsch> which the old code seemed to have caught
[17:54:08 CEST] <nevcairiel> you are supposed to use the parser to split frames in that case, the decoder would not work either way
[17:56:39 CEST] <nevcairiel> and you seem to read the merge wrong, the only check that was removed is if the buffer was too small, not too big
[18:02:21 CEST] <smitb> hi...I am new to open source...can anybody help me get started with ffmpeg developement?
[18:23:46 CEST] <fritsch> nevcairiel: yeah, exactly, that's my point - with that check removed. the decode call beyond will write too much data into a buffer which is too small.
[18:24:32 CEST] <nevcairiel> you seem to misinterpret what the buffer size represents
[18:24:53 CEST] <fritsch> hopefully I do, that's why I asked
[18:25:00 CEST] <nevcairiel> its an input buffer
[18:25:26 CEST] <fritsch> ah well - then all is fine
[18:26:31 CEST] <nevcairiel> you dont even get to allocate output buffers yourself anymore these days
[18:26:42 CEST] <nevcairiel> or if you do, it tells you how big its supposed to be first
[18:32:42 CEST] <Compn> smitb : sure, what os are you using? have you installed a developer environment / build system yet ? 
[18:33:06 CEST] <Compn> smitb : have you compiled source code before? what exp level would you say you were? that will help us figure out what kind of help you need :)
[18:33:09 CEST] <smitb> hi Compn I'm currently running Arch linux
[18:33:29 CEST] <smitb> I have used cmake before
[18:33:43 CEST] <smitb> and also I've used git in some of my projects
[18:33:52 CEST] <smitb> this is my first time with any organisation
[18:34:41 CEST] <Compn> oh ok 
[18:35:24 CEST] <Compn> well its pretty simple to start. you can review code and look for problems, or review bug reports and look for problems, or start helping on the ffmpeg-user mailing list and learn the code
[18:36:03 CEST] <smitb> can I get a link to some bugs that beginers can track down?
[18:37:49 CEST] <Compn> well
[18:38:18 CEST] <Compn> we need help with that, we dont have a list of easy to fix bugs i dont think
[18:38:21 CEST] <Compn> http://trac.ffmpeg.org
[18:38:29 CEST] <Compn> is our bugtracker
[18:38:54 CEST] <Compn> one min and i know of a good bug for you smitb
[18:41:45 CEST] <Compn> smitb : i have a sample file that you have to add to ffmpeg. ffmpeg can decode it if you add its tag into the code
[18:42:11 CEST] <smitb> alright
[18:42:30 CEST] <Compn> let me upload it here...
[18:44:22 CEST] <michaelni> rcombs, can you take a look at "[FFmpeg-devel] [PATCH] libavcodec/mmaldec.c: add interlaced_frame and format struct to AVFrame for deinterlacing" especially the data[2] stuff looks like it needs a more clear reply than nevcairiels, and you wrote the mmal code ...
[18:47:50 CEST] <Compn> smitb : https://www.datafilehost.com/d/51cb171a
[18:50:25 CEST] <Compn> smitb : you need to add that JR24 tag to the asf demuxer (or probably riff.c)
[18:50:50 CEST] <smitb> ok...
[18:51:11 CEST] <Compn> then make a patch, like it says in our devel rules, http://ffmpeg.org/developer.html#Submitting-patches
[18:51:15 CEST] <Compn> and send it to the mailing list
[18:51:29 CEST] <smitb> ok
[18:51:34 CEST] <smitb> thanks for the help
[18:52:34 CEST] <Compn> sure no problem
[18:52:53 CEST] <Compn> yeah i think you just need to add the tag to the libavformat/riff.c source code. its an mjpeg tag
[18:53:18 CEST] <Compn> then compile the code and make sure it works. :)
[19:00:30 CEST] <Compn> smitb : does that look like something you want to tackle ?
[19:00:43 CEST] <Compn> you can look at previous commits to riff.c so you can find out what you need to do
[19:00:54 CEST] <Compn> http://git.videolan.org has our web git repo for easy checking
[19:01:58 CEST] <Compn> http://git.videolan.org/?p=ffmpeg.git;a=tree;f=libavformat
[19:02:45 CEST] <smitb> yeah...sure
[19:03:04 CEST] <Compn> should only be a one line patch :)
[19:05:13 CEST] <smitb> yeah...currently I'm cloning the repository...I was having some problem due to unusually slow connections in my place
[19:11:37 CEST] <smitb> ok Compn I guess in riff.c I'll have to add a MKTAG right?
[19:14:17 CEST] <Compn> yep
[19:14:19 CEST] <Compn> in the mjpeg area
[19:15:40 CEST] <smitb> ok so in the MJPEG area making a MKTAG which will have the tag JR24 right?
[19:15:50 CEST] <Compn> you can add a comment too, its "Quadrox Mjpeg"
[19:15:50 CEST] <Compn> yes
[19:16:04 CEST] <smitb> thanks
[19:16:33 CEST] <Compn> by mjpeg area, i mean just with all the other mjpeg tags
[19:16:34 CEST] <Compn> ehhe
[19:16:50 CEST] <smitb> yeah I got that :)
[19:25:23 CEST] <Compn> j-b : do you know of any way to organize easy bugs for new contributors ? maybe its as easy as a tag on the trac ?
[19:26:39 CEST] <Compn> michaelni : do you think we should add a category of easy bugs for new contributors ? what should we call it ?
[19:27:08 CEST] <Compn> smitb : the easy part is coding, the hard part is doing organization within the project ;)
[19:27:35 CEST] <smitb> Compn I'm trying to build the cloned repo...and after running I'm getting yasm/nasm not found or too old
[19:27:42 CEST] <Compn> yeah you should install yasm
[19:28:05 CEST] <Compn> one of the very few things ffmpeg relies on
[19:28:39 CEST] <smitb> ok...I'm using arch linux...so I guess the one available through pacman, the package manager should do right?
[19:43:19 CEST] <haasn> What would it take to get support for PNG's gAMA, cHRM and iCCP chunks into libavcodec?
[19:43:47 CEST] <haasn> Parsing the things is easy enough, but it would require plumbing the colorimetry metadata and icc profile through the APIs somehow
[19:43:56 CEST] <haasn> And I have no idea where to even start
[19:54:07 CEST] <Compn> smitb : yes that arch linux repo for yasm should do
[19:54:11 CEST] <Compn> hopefully
[19:54:27 CEST] <smitb> yes its compiled fine
[19:54:43 CEST] <Compn> haasn : good question. not sure what internal api you should be using for those
[19:54:58 CEST] <Compn> haasn : better wait for ubitux or other person who knows :D
[19:59:25 CEST] <michaelni> Compn, people could add a keyword like "easy" to easy bugs ...
[20:00:47 CEST] <michaelni> we also could add a drop down thingy like for priorty but not sure if its useful enough to do that, depends on the community what is preferred
[20:00:47 CEST] <JEEB> haasn: inb4 another type of side data
[20:01:11 CEST] <haasn> JEEB: yeah probably
[20:01:19 CEST] <haasn> embedded ICC profiles are not exactly PNG-specific
[20:01:44 CEST] <Compn> michaelni : ok i'll write a mail to -devel then asking about it
[20:02:00 CEST] <haasn> embedded gamma/primaries are more of a PNG thing, but only in practice - not in principle. I think a single type of side data struct that exposes all forms of embedded colorimetry could work well
[20:02:21 CEST] <haasn> then the PNG, JPEG decoders etc. could all set those based on the actual metadata available to them
[20:04:11 CEST] <JEEB> ayup
[20:36:47 CEST] <Compn> michaelni : sent mail to list about bug trac difficulty tag, thanks
[20:36:52 CEST] Action: Compn afk
[23:06:06 CEST] <cone-980> ffmpeg 03James Almer 07master:293484fa5e44: avcodec: add missing xmm/neon clobber test wrappers for the new decode API
[23:38:32 CEST] <cone-980> ffmpeg 03Hendrik Leppkes 07master:1ad4471526c7: configure: disable the new optimizer in Visual Studio 2015 Update 3
[00:00:00 CEST] --- Mon Jul  4 2016