[Ffmpeg-devel-irc] ffmpeg-devel.log.20161120

burek burek021 at gmail.com
Mon Nov 21 03:05:02 EET 2016


[00:14:49 CET] <cone-137> ffmpeg 03James Almer 07master:0ffea3565700: avformat/utils: check for overflow before reallocating side data
[00:28:07 CET] <cone-137> ffmpeg 03James Almer 07master:574929d8b6de: avcodec/avpacket: fix leak on realloc in av_packet_add_side_data()
[00:28:08 CET] <cone-137> ffmpeg 03James Almer 07release/3.0:492bbe58902d: avcodec/avpacket: fix leak on realloc in av_packet_add_side_data()
[00:28:09 CET] <cone-137> ffmpeg 03James Almer 07release/3.1:c3f97bf54496: avcodec/avpacket: fix leak on realloc in av_packet_add_side_data()
[00:28:11 CET] <cone-137> ffmpeg 03James Almer 07release/3.2:3bd7ad58a7e1: avcodec/avpacket: fix leak on realloc in av_packet_add_side_data()
[01:53:27 CET] <philipl> Weird. I've moved the setting of pix_fmt to the callback and everywhere I check the format, it's the right value at the right time, yet my 10bit is mangled.
[01:54:51 CET] <philipl> hah. got it.
[02:12:16 CET] <philipl> BtbN: https://github.com/philipl/FFmpeg/commit/faa7b2256d41bfb1a3ef03403d8fef9b85a38428
[03:52:24 CET] <cone-230> ffmpeg 03James Almer 07release/3.2:ee5677737942: avcodec/rawdec: check for side data before checking its size
[13:00:46 CET] <cone-946> ffmpeg 03Michael Niedermayer 07master:ac206bb1756d: swscale/options: Use AV_OPT_TYPE_PIXEL_FMT
[18:00:00 CET] <kierank> dammit please please please don't put in a psd decoder
[18:04:20 CET] <RiCON> kierank: might want to say so in the ML then
[18:04:28 CET] <RiCON> only atomnuker said anything against
[18:04:31 CET] <kierank> atomnuker has already said what I wanted to say
[18:04:45 CET] <kierank> objections are not a quantity over quality thing
[18:05:58 CET] <JEEB> is it due to the fact that it only supports the "preview" picture?
[18:06:25 CET] <kierank> well it's a nightmarishly complex format of which ffmpeg will only support a tiny subset
[18:06:36 CET] <JEEB> yes
[18:06:54 CET] <kierank> Not to mention that close to 90%
[18:06:54 CET] <kierank> of the features it has are unsupported by the framework, be it vectors, the
[18:06:54 CET] <kierank> colorspaces, layers, etc
[18:10:12 CET] <JEEB> yup
[18:11:42 CET] <durandal_1707> what about fits decoder?
[18:12:26 CET] <durandal_1707> what about eia-608 Iirc scanner?
[18:12:47 CET] <wm4> PSD is a format of which AFAIK 99% can't be supported
[18:12:51 CET] <durandal_1707> as filter of course
[18:12:54 CET] <wm4> so it's very dumb to add "support
[18:12:55 CET] <wm4> "
[18:13:01 CET] <wm4> what's next, .doc?
[18:13:07 CET] <wm4> .html?
[18:13:13 CET] <durandal_1707> gimp format 
[18:13:23 CET] <wm4> well for .html you could at least add a webkit wrapper
[18:13:32 CET] <wm4> which would actually be more sane than this psd "support"
[18:13:47 CET] <wm4> and that's crazy
[18:13:55 CET] <durandal_1707> exr is already supported somehow 
[18:14:30 CET] <durandal_1707> and raw text via awesome demuxer 
[18:40:30 CET] <tab1293> are there any resources on how to implement an ffmpeg protocol?
[18:41:29 CET] <durandal_1707> tab1293: explain? 
[18:42:00 CET] <tab1293> like the http protocol for instance, https://github.com/FFmpeg/FFmpeg/blob/master/libavformat/http.c
[18:42:16 CET] <tab1293> I would like to try to implement a protocol for websockets
[18:42:34 CET] <tab1293> is there any high level documenation on how to do this?
[18:42:46 CET] <BtbN> no, look at the other protocols.
[18:43:37 CET] <tab1293> does http inherit the url protocol?
[18:44:01 CET] <JEEB> it utilizes a URLContext
[18:44:37 CET] <JEEB> it then checks if the url starts with http:// (and a few other things)
[18:45:14 CET] <JEEB> you can look at the functions the protocol registers for the callbacks
[18:45:21 CET] <JEEB> https://github.com/FFmpeg/FFmpeg/blob/master/libavformat/http.c#L1663
[18:45:30 CET] <JEEB> and yes, http is a URLProtocol
[18:45:45 CET] <JEEB> oh wait, that's the "use proxy" one
[18:45:57 CET] <tab1293> yeah I saw the regular one about it
[18:46:00 CET] <JEEB> this one is the base http https://github.com/FFmpeg/FFmpeg/blob/master/libavformat/http.c#L1527
[18:46:22 CET] <tab1293> okay thanks, im going to try to go through a couple of these protocols
[18:46:33 CET] <tab1293> see I can find the general pattern
[18:47:50 CET] <JEEB> the process is somewhat similar to demuxers, although not exactly https://wiki.libav.org/Internals/DemuxerHowTo
[18:48:13 CET] <JEEB> you define your context, and then add your thing into a few files
[18:48:32 CET] <JEEB> the functions you define for the callbacks then get utilized by the framework
[18:51:36 CET] <tab1293> cool, yeah I see whats happening
[18:51:56 CET] <tab1293> where is urlcontext defined? I dont see it in libavformat/url.c
[18:52:08 CET] <JEEB> url.h gets included so I guess it's there
[18:52:24 CET] <JEEB> git grep is one of your best friends
[18:52:24 CET] <tab1293> ah
[20:32:23 CET] <cone-946> ffmpeg 03Andreas Cadhalpun 07master:1c7da19a4b45: ffmdec: validate codec parameters
[22:26:31 CET] <cone-946> ffmpeg 03Carl Eugen Hoyos 07master:322568c079d6: lavc/ffv1: Support YUV4xxP12 and GRAY12.
[22:40:00 CET] <cone-946> ffmpeg 03Michael Niedermayer 07master:748ce9d70286: avutil/opt: Fix setting int64 to its maximum
[23:32:16 CET] <cone-946> ffmpeg 03James Almer 07master:ee888cfbe777: avformat/matroskaenc: postpone writing the Tracks master
[23:32:17 CET] <cone-946> ffmpeg 03James Almer 07master:8c1342e631d6: avformat/matroskaenc: write updated STREAMINFO metadata for FLAC streams if available
[23:59:09 CET] <cone-946> ffmpeg 03Marton Balint 07master:58af48f1c3cc: ffplay: add support for negative RGBA linesize
[23:59:10 CET] <cone-946> ffmpeg 03Marton Balint 07master:28d87899e33f: ffplay: allow fallback to software renderer
[00:00:00 CET] --- Mon Nov 21 2016


More information about the Ffmpeg-devel-irc mailing list