[Ffmpeg-devel-irc] ffmpeg-devel.log.20150522

burek burek021 at gmail.com
Sat May 23 02:05:03 CEST 2015


[01:02:26 CEST] <cone-925> ffmpeg 03Urvang Joshi 07master:6b56fcbd10a5: WebP muxer: support a packet containing animated WebP.
[01:48:29 CEST] <pentanol> ,join #linux
[01:48:41 CEST] <pentanol> sorry
[02:37:25 CEST] <cone-925> ffmpeg 03Martin Storsjö 07master:d15eec4d6bdf: tls: Use custom IO to read from the URLContext
[02:37:26 CEST] <cone-925> ffmpeg 03Michael Niedermayer 07master:162644c8331d: Merge commit 'd15eec4d6bdfa3bd4c4b5b7dd2dbd699ba253d02'
[02:49:33 CEST] <cone-925> ffmpeg 03Martin Storsjö 07master:d13b124eaf45: tls: Remove the nonblocking code
[02:49:34 CEST] <cone-925> ffmpeg 03Michael Niedermayer 07master:9221e362fd3b: Merge commit 'd13b124eaf452b267480074b2e6946538ed03a6e'
[03:16:22 CEST] <cone-925> ffmpeg 03Martin Storsjö 07master:94599a6de382: tls: Remove all the local polling loops
[03:16:23 CEST] <cone-925> ffmpeg 03Michael Niedermayer 07master:a0124b89e393: Merge commit '94599a6de3822b13c94096d764868128f388ba28'
[03:42:01 CEST] <rcombs> wm4: ^ oh boy, conflicts abound
[03:46:25 CEST] <cone-925> ffmpeg 03Urvang Joshi 07master:f99fed733d65: WebP encoder: extract out some methods into a separate helper library.
[04:03:23 CEST] <Daemon404> wm4, iirc the patches were his idea.
[05:34:19 CEST] <cone-925> ffmpeg 03James Almer 07master:e91fbfd9cf17: tls: fix compilation when both gnutls and openssl are enabled
[05:44:00 CEST] <Daemon404> i must ask, why would you enable both
[05:44:53 CEST] <jamrial> ask ubitux :p
[05:45:15 CEST] <jamrial> i noticed it on his fate machine that enables all the libs
[05:47:29 CEST] <jamrial> http://fate.ffmpeg.org/report.cgi?time=20150522025204&slot=x86_64-archlinux-gcc-enableshared quite useful to find compilation errors no other fate client would trigger
[05:48:03 CEST] <jamrial> it's still missing a couple libs, though, like webp and x265
[05:51:02 CEST] <rcombs> Daemon404: it might actually make sense to enable both Secure Transport and <either of the others>, if I rearranged the code in tls.c a bit
[05:52:01 CEST] <rcombs> Daemon404: because then you could have <other lib> provide crypto for ffrtmp (which I'm not sure would actually be possible at all with ST), but ST provide TLS (which is important because it can use the system trust store)
[07:34:49 CEST] <Timothy_Gu> jamrial: next time take a look at fatebeta.ffmpeg.org :)
[09:20:22 CEST] <wm4> if both gnutls and openssl are enabled, how do you select either?
[09:20:28 CEST] <wm4> makes not much sense...
[09:20:46 CEST] <wm4> helpful for testing though
[09:20:55 CEST] <wm4> maybe
[09:34:23 CEST] <nevcairiel> it probably uses whichever appears first in code
[11:23:58 CEST] <ubitux> alright, gcc5 packaged
[11:24:09 CEST] <ubitux> updating fate instances..
[12:04:19 CEST] <cone-682> ffmpeg 03James Almer 07master:29216d7fd14d: tls: fix compilation when both gnutls and openssl are enabled
[12:04:20 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:1617a89cf000: Merge commit '29216d7fd14d1cec16821867d17c90b5c49e8c07'
[12:22:36 CEST] <cone-682> ffmpeg 03Tim Nicholson 07master:ebb08f3fe416: avutil/pixfmt: Clarify DV example comments for AVChromaLocation
[14:18:43 CEST] <wm4> rcombs: sent the patch for splitting tls.c to libav-devel
[14:20:54 CEST] <rcombs> \o/
[14:35:54 CEST] <Compn> any chance for embeddeding ssl stuff into ffmpeg
[14:36:04 CEST] <Compn> so https works natively ?
[14:36:25 CEST] <Compn> "https protocol not found, recompile with openssl or gnutls enabled."
[14:36:26 CEST] <Compn> :(
[14:37:23 CEST] <iive> right after we merge emacs.
[14:38:47 CEST] <av500> isnt ssl hacked?
[15:46:43 CEST] <TimNich> Really, allow setting a value thats been deprcated for at least 6 years?
[15:46:50 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:d09321b68e83: avcodec/mpeg12dec: Fix chroma location
[15:46:51 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:a45cf639e6fb: libavformat/mux: Add ff_choose_chroma_location()
[15:46:52 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:3b176af001dd: avformat/mxfenc: Add color siting element
[15:59:08 CEST] <michaelni> TimNich, its not only allowed but even hardcoded in bmx/libmxf for some cases if i read the code correctly
[15:59:44 CEST] <michaelni> i dont mind doing it differently, i just dont have any means to test if the resuting files will work with "pro" software
[16:01:01 CEST] <TimNich> I would have thought anything old enough to only like the deprecated value is likely to have other compatibility issues too.
[16:02:24 CEST] <michaelni> quite likely yes
[16:02:55 CEST] <TimNich> Im not aware of anything that didnt actually work when this *optional* value was unset. 
[16:04:28 CEST] <TimNich> Ill have to look again at bmxlib and try to find the rationale for it& could take a few days as its coming up to a bank holiday weekend.
[16:05:02 CEST] <michaelni> ive seen it under if(something_avid || "someoldspec")
[16:05:51 CEST] <michaelni> this one: if ((mFlavour & MXFDESC_AVID_FLAVOUR) || (mFlavour & MXFDESC_SMPTE_377_2004_FLAVOUR))
[16:06:16 CEST] <michaelni> so maybe some avid software has issues with it, but iam just guessing
[16:12:21 CEST] <TimNich> Could well be historical and no longer relevant, rather like the Avid DNX UL that is actually JP2K.
[16:13:04 CEST] <michaelni> possible, yes
[16:14:13 CEST] <TimNich> sometime between 2004 and 2009 it was deprecated, and I doubt anyone is using an Avid that old anymore, but I will try and narrow it down
[16:15:17 CEST] <TimNich> has the option to override been specifically requested?
[16:17:26 CEST] <michaelni> yes, Color siting = 4 (Rec 601)
[16:19:57 CEST] <TimNich> Someone has a use case where they need the value to be 4 not 0?
[16:20:52 CEST] <michaelni> i dont know that, ill ask, ill CC you
[16:20:55 CEST] <TimNich> If so Id like to understand that case better to ensure its not a red herring.
[16:22:31 CEST] <TimNich> OK, we have a site power shutdown coming soon, so I will be going dark in a bit.
[16:38:33 CEST] <TimNich> michaelni If this is the NOA use case then it is a red herring, S386m replicates for convenience data from S377, which contains the updated deprectation.
[16:41:23 CEST] <michaelni> ok, then lets consider the siting=4 patch to be droped for now
[16:48:10 CEST] <TimNich> OK, Ive just added chapter and verse to your email.
[16:58:55 CEST] <rcombs> anyone else think it'd be useful for avutil/attributes.h to have something `warn_unused_result`?
[16:59:25 CEST] <wm4> sounds potentially useful
[16:59:45 CEST] <rcombs> there are a bunch of functions in lav* that should probably have it
[17:00:03 CEST] <wm4> like which?
[17:00:34 CEST] <rcombs> anything that allocates anything (and doesn't return it), anything that seeks or reads data&
[17:25:44 CEST] <jermy> I'm doing some work using ffm2 streams to move some content between multiple ffmpeg processes, however it turns out that ffmdec isn't expecting streamed input - it attempts to reset the buffer and do a seek and doesn't handle eof very nicely
[17:26:19 CEST] <jermy> I'm guessing this isn't a particularly normal use-case, but wondering if this is anything worth pushing patches back for, or fairly unimportant
[17:32:39 CEST] <cone-682> ffmpeg 03Niklesh 07master:337aa17e6109: avcodec/movtextdec: Add support for large boxes(>32 bit)
[20:34:35 CEST] <cone-682> ffmpeg 03Vittorio Giovara 07master:848e86f74d3e: mpegvideo: Drop flags and flags2
[20:34:36 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:d9b264bc7372: Merge commit '848e86f74d3e6e87fa592ee8ba8c184cc5fd9a42'
[20:49:00 CEST] <cone-682> ffmpeg 03Vittorio Giovara 07master:9c1db92ad372: mpegvideo: Drop err_recognition
[20:49:01 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:d5227ceff13e: Merge commit '9c1db92ad372d4cd69e0490e691c56e4097cb193'
[20:56:36 CEST] <cone-682> ffmpeg 03Vittorio Giovara 07master:6f54dc43cee6: mpegvideo: Drop stream_codec_tag
[20:56:37 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:bc373595f28e: Merge commit '6f54dc43cee6b2f5d183acf98b32a3cf8be4a4fc'
[21:16:18 CEST] <cone-682> ffmpeg 03Vittorio Giovara 07master:d52804555882: mpegvideo: Have ff_mpeg_unref_picture use AVCodecContext directly
[21:16:19 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:a25ee5f9228d: Merge commit 'd528045558825f01472e9bee873f60c98d661e53'
[21:25:16 CEST] <cone-682> ffmpeg 03Vittorio Giovara 07master:a3f4c930ac3f: mpegvideo: Have ff_mpeg_ref_picture use AVCodecContext directly
[21:25:17 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:179527f34ea0: Merge commit 'a3f4c930ac3f49f47b6e6ffda925d0dcf80320e2'
[22:29:53 CEST] <sh4rm4^bnc> why do i have AVInputFormat ff_v4l2_demuxer only in libavdevice.a but not in libavdevice.so ? linker bug ?
[22:30:43 CEST] <sh4rm4^bnc> and is ff_v4l2_demuxer deprecated ? i couldnt find any references in the code
[22:31:09 CEST] <sh4rm4^bnc> /src/build/ffmpeg/ffmpeg-2.6.3 # find . -name '*.c' | xargs grep ff_v4l2_demuxer
[22:31:10 CEST] <sh4rm4^bnc> ./libavdevice/v4l2.c:AVInputFormat ff_v4l2_demuxer = {
[22:31:10 CEST] <sh4rm4^bnc> /src/build/ffmpeg/ffmpeg-2.6.3 # 
[22:35:24 CEST] <sh4rm4^bnc> i constructed a demo app (mostly c/p from example code) accessing v4l2 device using ff_v4l2_demuxer http://sprunge.us/bgZI
[22:36:08 CEST] <sh4rm4^bnc> if there's another better way to access the v4l2 driver let me know...
[22:36:36 CEST] <sh4rm4^bnc> (also weird that no header declares AVInputFormat ff_v4l2_demuxer)
[22:38:24 CEST] <ubitux> ff_ symbols are private, you need to open a file through avformat with "v4l2" (or "video4linux2") format string
[22:48:12 CEST] <cone-682> ffmpeg 03Vittorio Giovara 07master:4e17946f10d3: mpegvideo: Rework various functions not to use MpegEncContext directly
[22:48:13 CEST] <cone-682> ffmpeg 03Michael Niedermayer 07master:f275f9eaee9c: Merge commit '4e17946f10d39eec6cc03fb249ae8147373141b6'
[00:00:00 CEST] --- Sat May 23 2015


More information about the Ffmpeg-devel-irc mailing list