[Ffmpeg-devel-irc] ffmpeg-devel.log.20141129

[burek]

[00:23] <cone-437> ffmpeg.git 03Neil Birkbeck 07master:dd5d61795690: avfilter/vf_idet: Fixing idet for single-frame inputs.
[01:08] <cone-437> ffmpeg.git 03Lukasz Marek 07master:0cb994dfe32e: lavu/opt: add escaping to av_opt_serialize
[04:27] <cone-784> ffmpeg.git 03Vadim Kalinsky 07master:2db29482f1ae: avutil/bprint: C++ compatible AVBPrint definition.
[05:35] <jamrial> http://fate.ffmpeg.org/report.cgi?time=20141128085005&slot=x86.os2.446 something broke mp2 decoding on os/2 a month or so ago
[05:35] <jamrial> encoding, rather
[10:23] <ubitux> pross: yeah so actually, there is this: c70a6a41ddb759a5c6e8e810ebd603e39c36a315
[12:03] <cone-428> ffmpeg.git 03Dave Yeo 07master:090a7801a881: libavutil/thread.h: Support OS/2 threads
[12:03] <cone-428> ffmpeg.git 03Dave Yeo 07master:32eadfe453c3: libavutil/threads.h: correct an include to be local
[13:20] <cone-428> ffmpeg.git 03Michael Niedermayer 07master:c299b6fd0807: avutil/buffer: Move USE_ATOMICS to thread.h to avoid it becoming out of sync with it
[13:49] <cone-428> ffmpeg.git 03Michael Niedermayer 07master:5bf4cd8e5b77: avutil/ppc/util_altivec: add () to VEC_SPLAT16 macro
[14:25] <cehoyos> Daemon404: Hi Derek, what's wrong with pastebin (or Compn)?
[14:25] <cehoyos> Anyway, I will leave the winodw open, thank you!
[15:03] <cone-428> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n0.5.15': unknown revision or path not in the working tree.
[15:03] <cone-428> Use '--' to separate paths from revisions
[15:03] <cone-428> refs/tags/n0.5.15:HEAD: avutil/ppc/util_altivec: add () to VEC_SPLAT16 macro
[15:43] <kurosu> Is there a non-convoluted way to export options for a bitstream filter ?
[15:44] <kurosu> There's AV_OPT_FLAG_FILTERING_PARAM but no class/... way to then export said options except for libavcodec/options_table.h
[15:46] <kurosu> passing a const char *args as a parameter to the bsf seem a dead giveaway it predates that kind of things
[19:04] <BtbN> wow, linking ffmpeg.exe takes longer than the entire compiling, in cygwin
[19:06] <Compn> mingw configure script takes foreeeeeeeever 
[19:06] <BtbN> Configure runs as fast as on my linux box
[19:07] <BtbN> But it takes like 10 minutes to link ffmpeg.exe
[19:07] <Compn> mingw and cygwin are diff , of course :p
[19:07] <Compn> static ?
[19:07] <BtbN> yes
[19:08] <cone-428> ffmpeg.git 03Michael Niedermayer 07master:cae851c78921: avcodec/ac3enc: Use avpriv_float_dsp_alloc()
[19:08] <cone-428> ffmpeg.git 03Michael Niedermayer 07master:14285c3331bc: avcodec/aacenc: Use avpriv_float_dsp_alloc()
[19:08] <cone-428> ffmpeg.git 03Michael Niedermayer 07master:b0464212bd81: avcodec/on2avc: Use avpriv_float_dsp_alloc()
[19:09] <jamrial> linking on windows also trashes my hdd like mad
[19:10] <jamrial> 10 min is excessive, though
[19:14] <Compn> whenever i install mingw and try compiling ffmpeg or mplayer, seems like i quickly pick up a bunch of errors on the disk
[19:14] <Compn> >windows problems
[19:17] <jamrial> in general, running a cross-compiler on a linux VM is much faster than using a cygwin/msys environment
[19:19] <jamrial> i rarely do it because i'm too lazy to boot up one. especially now that i started using msys2 which has git send-email working out of the box without a hassle :p
[19:20] <Compn> ooo
[19:20] <kurosu> does msys2 allow building windows binary without linking to the equivalent of cygwin's DLLs ? (ie a bit like -mno-cygwin)
[19:20] <Compn> but yes cross compile is faster in all cases.
[19:21] <Compn> if msys is using mingw, then yes. because mingw uses native binary not cygwin
[19:21] <kurosu> iirc, msys2 is based off cygwin from I had read from afar
[19:22] <jamrial> mingw is independent from the msys/msys2 environemnt
[19:22] <Compn> iirc msys is just the shell, so 
[19:23] <Compn> i dont know what msys2 is doing yet
[19:23] <kurosu> ok but cygwin usually impacts both the environment and the toolchain, hence my question
[19:23] <Compn> cehoyos should force me to use msys2 , because i dont use git-send-email on windows
[19:23] <Compn> send patches attached to emails ehe
[19:24] <jamrial> yeah. the same happens if you use msys gcc
[19:24] <kurosu> depends on the number of patches
[19:24] <kurosu> if I'm sending only one as an update to a previous post, then I do it manually
[19:24] Action: Compn prefers direct commit anyway
[19:24] <jamrial> but mingw gcc is independent regardless of what environment you use
[19:24] <BtbN> I'm using cygwin for everything on Windows. I basicaly cross compile in it, if i want to build for normal windows.
[19:24] <BtbN> mingw is available as cygwin package, so you use it like you would with a cross compiler on linux.
[19:24] <kurosu> probably I could learn how to use message ID to do just that, but...
[19:25] <jamrial> you mean "git send-email --in-reply-to=<message ID> file.patch"?
[19:29] <kurosu> yeah
[19:38] <BtbN> Is there some example on how to use the dynarray from avutil/dynarray.h?
[21:26] <cone-428> ffmpeg.git 03Michael Niedermayer 07master:2336c76d5a6e: avcodec/ra288: Use avpriv_float_dsp_alloc()
[21:26] <cone-428> ffmpeg.git 03Michael Niedermayer 07master:fc9ced41e48f: avcodec/twinvq: Use avpriv_float_dsp_alloc()
[21:44] <BtbN> https://bpaste.net/show/6755cbec2e4d Is this me, doing some weird memory corruption in my nvenc encoder, or is this an actual bug in the mkv muxer?
[23:01] <michaelni> BtbN, no idea from this, try valgrind or asan
[23:20] <BtbN_> michaelni, already found it. Just forgot to set a variable, so i called av_malloc with a size of 0
[23:21] <BtbN_> It's interesting that this ringbuffer actualy does squeeze out a few more fps over the old linked list
[23:21] <BtbN_> is malloc/free realy that expensive?
[23:22] <nevcairiel> it can be
[23:23] <BtbN_> yeah, but i didn't expect it to actualy take some time, even if the CPU isn't busy
[00:00] --- Sun Nov 30 2014