[Ffmpeg-devel-irc] ffmpeg-devel.log.20150110

burek burek021 at gmail.com
Sun Jan 11 02:05:02 CET 2015


[00:03] <cone-671> ffmpeg.git 03James Almer 07master:da02ee127a21: x86/vf_pp7: port dctB_mmx to yasm
[00:25] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:11d8fa5e9d69: ffserver: Use av_malloc_array()
[00:36] <cone-671> ffmpeg.git 03wm4 07master:aab74a38b8ba: rm: fix memory leak on init failure
[02:23] <cone-671> ffmpeg.git 03Arwa Arif 07master:4df01d5d44cf: avfilter: Factorize the use of norm_qscale in pp7
[04:03] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:bdb136d30218: avfilter/internal: add fixme to ff_norm_qscale() to document obvious scaling issue
[04:03] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:7824dc5150c0: avformat/movenc: workaround bug in "PathScale EKOPath(tm) Compiler Suite Version 4.0.12.1"
[05:00] <cone-671> ffmpeg.git 03Andreas Cadhalpun 07master:6108485cf70f: configure: enable vsx together with altivec for ppc64el
[05:00] <cone-671> ffmpeg.git 03Andreas Cadhalpun 07master:f97f2a3527ea: doc/examples: fix lib math dep for decoding_encoding
[05:00] <cone-671> ffmpeg.git 03Michael Niedermayer 07master:6f838dee3cf8: avformat/movenc: move variable to a more local block
[05:20] <cone-671> ffmpeg.git 03Michael Niedermayer 07release/2.5:b089b582500f: avformat/movenc: workaround bug in "PathScale EKOPath(tm) Compiler Suite Version 4.0.12.1"
[05:20] <cone-671> ffmpeg.git 03Andreas Cadhalpun 07release/2.5:07d508e4f55f: doc/examples: fix lib math dep for decoding_encoding
[05:30] <cone-671> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n2.5.3': unknown revision or path not in the working tree.
[05:30] <cone-671> Use '--' to separate paths from revisions
[05:30] <cone-671> refs/tags/n2.5.3:HEAD: avformat/movenc: move variable to a more local block
[14:13] <cone-577> ffmpeg.git 03Michael Niedermayer 07master:35e559ae32f2: avcodec/cavs: Check for av_malloc* failure in ff_cavs_init_top_lines()
[14:22] <ubitux> wtf @ #4240
[14:32] <cbsrobot_> ubitux: I guess spammers trying to blend in
[14:34] <cone-577> ffmpeg.git 03Paul B Mahol 07master:241f858cf4c3: LICENSE.md: mention mpdecimate and not decimate to be GPL
[15:12] <anshul_mahe> Do we have some linklist in ffmpeg utils for using in libavformat
[15:24] <relaxed> @4245, "I bonk investigated and instrument prophesy you on the pros of body wraps." haha
[15:27] <anshul_mahe> we should have something for these spams they will make searching bugs harder
[15:31] <relaxed> maybe instead of searching for bugs you should try these amazing cleanse efx body wraps
[15:50] <anshul_m> relaxed: haha
[15:56] <aetasx> found this while digging around: http://siliconandlithium.blogspot.com/2014/01/
[16:19] <ubitux> cbsrobot_: but i see no link whatsoever
[16:22] <aetasx> its in all the msvc folders within each of his git repos apparently
[16:32] <nevcairiel> aetasx: the first line of that post is already wrong, ffmpeg doesnt use autotools :)
[16:33] <aetasx> yeah I know I saw it but theres updated posts if you look under the newer ones
[16:34] <nevcairiel> seems like an excercise in futility though, what would a generator for a msvc project offer that using make doesnt  offer me
[16:34] Action: nevcairiel shrugs
[16:34] <aetasx> its for building against msvc/icl
[16:34] <nevcairiel> so?
[16:34] <nevcairiel> i can do that without any external tools
[16:35] <aetasx> I doubt thats the case for all the libs hes using
[16:35] <nevcairiel> we went through great lengths to actually make that possible
[16:36] <aetasx> the little project builder thing builds ffmpeg as well as all its dependencies
[16:39] <nevcairiel> i guess it may add some covenience for some use-cases, but its all possible without special tools either way
[16:40] <aetasx> the project I work on does it within msys2 which gets to be a somewhat pain in the ass a lot of times
[16:40] <aetasx> not to mention most of them are all build for gcc
[16:40] <nevcairiel> i only use msvc builds for development/debugging anyway, since for production the performance impact is too big
[16:40] <aetasx> yeah thats why Im doing it against icl.  MSVC doesn't really perform that great otherwise
[16:40] <nevcairiel> its unfortunate, but msvc builds are quite a bit slower
[16:40] <nevcairiel> i dont trust icl to generate working code tbh :p
[16:41] <aetasx> Im sure if it wasnt working for this, he wouldnt bother doing it :p
[16:41] <compn> i think you guys may have misunderstood what the blog post does
[16:41] <anshul__> does any one know about hls->start_pts, it looks like we have forgotten to update it, or may be I dont undestand its purpose
[16:41] <aetasx> it lets me be lazy
[16:41] <compn> it looks like he removes the dependency of using mingw/gcc on windows all together if you are just using vs
[16:41] <aetasx> correct
[16:42] <nevcairiel> aetasx: you need to disable a bunch of optimizations in icl which are on by default so it doesnt miscompile ffmpeg (configure does this for you) ... tells you a lot about their compiler :p
[16:42] <nevcairiel> compn: you d ont need mingw/gcc on windows at all if you compile with vs
[16:42] <nevcairiel> all you need is gnu make and a sh shell
[16:42] <aetasx> if this guy couldnt get it to build correctly, he wouldnt continue to be updating it
[16:43] <nevcairiel> the fact remains that for a very expensive compiler, icl is quite crappy :P
[16:43] <nevcairiel> its popular in the "ricing" crowd, since apparently it claims to produce faster code
[16:44] <aetasx> yeah, for intel platforms it does
[16:44] <nevcairiel> .. at the cost of miscompiling occasionally
[16:45] <nevcairiel> it may as well work well, but i wouldnt pay for it
[17:09] <cone-577> ffmpeg.git 03Michael Niedermayer 07master:1673436d85d6: avfilter/vf_pp7: use av_malloc_array()
[17:09] <cone-577> ffmpeg.git 03Michael Niedermayer 07master:0b52bdfb3b47: avfilter/vf_pp7: Add GBRP pixel format
[19:06] <compn> nevcairiel : well unxutils project (sh and make .exe files, amongst others like grep) isnt very well known.
[19:06] <compn> nevcairiel : also, if ffmpeg can be built like you say so, can you write down those instructions in our documentations ?
[19:07] <compn> do we have a vs build guide ?
[19:07] <JEEB> yes, in the documentation
[19:07] <JEEB> and Daemon404 even made a vimeo clip
[19:08] <cone-577> ffmpeg.git 03Michael Niedermayer 07master:a4d3cf10b2ec: avcodec/diracdec: Check slices malloc and propagate error code
[19:15] <BBB> if people still cant build ffmpeg with vs nowadays, they likely wouldnt be able to build it with gcc under linux either
[19:16] <JEEB> it's the usual crowd that expects that they just need to double-click a visual studio solution and ctrl+alt+f7 (rebuild)
[19:21] <wm4> JEEB: I'd expect that too
[19:21] <wm4> it's equivalent to ./configure && make
[19:21] <wm4> _if_ there's a vs project file
[19:22] <wm4> though I never ever had a case where this actually worked with any vs project file with any project
[19:22] <wm4> not even MS examples
[19:22] <wm4> (confusing)
[19:23] <nevcairiel> I wanted to say my project works like that, but i would be lying, since you need to compile ffmpeg first :D
[19:24] <Daemon404> JEEB, the clip is outdated
[19:24] <Daemon404> you dont need c99conv at all for msvc2013
[19:24] <JEEB> yeah
[19:25] <Daemon404> also the only reason i made that clip is that one guy who kept emailing me was *so dumb*
[19:25] <Daemon404> he couldnt follow step-by-step
[19:25] <Daemon404> side note: if i google how to do something and i find videos but no text, i rage
[19:25] <compn> nevcairiel : liar! :p
[19:26] <compn> Daemon404 : you rage because you are basically watching a 5 minute video for the guy to read off 8 sentences of instructions? :P
[19:26] <Daemon404> yea
[19:26] <Daemon404> 10 seconds vs 5 minutes of my lie
[19:26] <Daemon404> life*
[19:31] <wm4> Daemon404: agreed, fuck video "tutorials"
[19:35] <aetasx> how would you guys compare transcoding a movie in x264 vs ffmpeg speed-wise?
[19:35] <aetasx> like to squeeze something that I've already run through ffmpeg
[19:39] <cone-577> ffmpeg.git 03Michael Niedermayer 07master:ebe3a41ea3b9: avcodec/dvbsubdec: Add some av_malloc() failure checks
[19:39] <cone-577> ffmpeg.git 03Michael Niedermayer 07master:1bf747ae84fc: avcodec/dvbsubdec: Return proper error codes from more functions
[19:39] <cone-577> ffmpeg.git 03Michael Niedermayer 07master:0898a6d4e4b6: avcodec/ratecontrol: Check for av_malloc*() failures
[20:25] <cone-577> ffmpeg.git 03wm4 07master:09b4ad15681b: vp9: avoid infinite loop with broken files
[21:20] <cone-577> ffmpeg.git 03Matt Oliver 07master:87deebe11a3a: avcodec/libvpxenc.c: Fix linking with codec interface with msvc.
[22:53] <michaelni> ubitux, is the closed caption patch from a few days ago ok to comit or was there some issue remaining ?
[22:59] <wm4> where is that patch? I want to take a look too
[23:46] <wm4> ubitux: what this cc patch would need it a way to realize the cc grid layout in ASS...
[23:46] <wm4> don't know if there's a good way to enforce this
[23:47] <kierank> does the cc patch match the test vectors?
[23:48] <kierank> if not then it shouldn't be committed
[23:53] <wm4> who knows
[23:53] <wm4> and some advanced stuff like formatting is not supported (there's a "todo" comment in the patch)
[00:00] --- Sun Jan 11 2015


More information about the Ffmpeg-devel-irc mailing list