[Ffmpeg-devel-irc] ffmpeg-devel.log.20150324

burek burek021 at gmail.com
Wed Mar 25 02:05:03 CET 2015


[01:24:42 CET] <cone-136> ffmpeg 03Vittorio Giovara 07master:e4fe535d12f4: mov: Write the display matrix in order
[01:24:43 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:dfea1acb21f3: Merge commit 'e4fe535d12f4f30df2dd672e30304af112a5a827'
[01:55:26 CET] <cone-136> ffmpeg 03Vittorio Giovara 07master:7419b34825c2: fate: Add test for tscc2 in mov
[01:55:27 CET] <cone-136> ffmpeg 03Vittorio Giovara 07master:592a04054e64: pixdesc: Replace a few leftover instances of non AV-prefixed flags
[01:55:28 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:cceb737936e6: Merge commit '7419b34825c20d52f667da2fbe4dc710ec8398f8'
[01:55:29 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:d280d43c2666: Merge commit '592a04054e6423be5050efd2bceece48b10b9c1d'
[02:07:20 CET] <cone-136> ffmpeg 03Vittorio Giovara 07master:312a9ef04bf6: pixfmt: Update documentation with prefixed pixel formats
[02:07:21 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:3e3995619807: Merge commit '312a9ef04bf6b55e68e4a61694b169cd3cc69bb8'
[02:31:32 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:b21af32aed7e: lavc: Prefer x264 over openh264 if both libraries are available
[02:31:33 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:2850131f2c7a: Merge commit 'b21af32aed7e343f962753943e9f16baedfb1750'
[02:48:36 CET] <cone-136> ffmpeg 03Hendrik Leppkes 07master:ae4f5d944cce: dxva2_h264: fix slice offset in long slice structs
[02:48:37 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:70cf40f9eb03: Merge commit 'ae4f5d944cce431b1f4715c863122355e1d78007'
[03:09:03 CET] <cone-136> ffmpeg 03Hendrik Leppkes 07master:1ddd3e23a49b: vaapi_h264: fix slice data offset
[03:09:04 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:a2dd2d7998a7: Merge commit '1ddd3e23a49b9eec11c62dca018238cf7349c90f'
[09:36:55 CET] <durandal_1707> I stopped counting students that picked APNG
[09:40:18 CET] Action: av500 picks APNG
[09:54:31 CET] <cone-136> ffmpeg 03Carl Eugen Hoyos 07master:c5d0148c7e6d: doc: Remove non-existing decklink options.
[09:58:54 CET] <cone-136> ffmpeg 03Carl Eugen Hoyos 07release/2.6:a9e683bb7ab9: doc: Fix alphabetic ordering for decklink input device. (cherry picked from commit a47c2a1d5f9f25b43f302bbf86dee3dc7b2e78e4)
[09:58:55 CET] <cone-136> ffmpeg 03Carl Eugen Hoyos 07release/2.6:40607290c904: doc: Remove non-existing decklink options.
[11:50:28 CET] <cone-136> ffmpeg 03Stephen Hutchinson 07master:a8c99205ca87: avisynth: Fix compilation against current 2.6 header(s).
[11:50:29 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:153073201cdc: Merge commit 'a8c99205ca8703bd849efae13fcf844315c7147d'
[12:04:33 CET] <cone-136> ffmpeg 03Stephen Hutchinson 07master:1188119624e4: avisynth: update documentation about the avisynth_c.h header
[12:04:34 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:d66ca43fc1f6: Merge commit '1188119624e42c1b2ce5be6137d2e6c6f4f03328'
[12:16:10 CET] <BBB> so whats the status of the AVFrameSideData patches?
[12:16:16 CET] <BBB> Im not sure what Im supposed to do now
[12:54:53 CET] <cone-136> ffmpeg 03Carl Eugen Hoyos 07master:b76df6efb681: lavf/matroskaenc: List subtitle codecs with fake codec_tags to allow remuxing.
[12:54:54 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:4a6096e48134: Merge remote-tracking branch 'cehoyos/master'
[14:14:15 CET] <cone-136> ffmpeg 03Jochen Strunk 07master:0276b9524294: avformat/mov: check color parameter type explicitly.
[16:09:37 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:f7e1367f5826: avcodec/msrledec: restructure msrle_decode_pal4() based on the line number instead of the pixel pointer
[16:09:38 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:c20eab5c850f: avcodec/msrledec: More verbose error message
[17:15:33 CET] <nevcairiel> fascinating how the avs people manage to break stuff again and again =p
[17:34:34 CET] <cone-136> ffmpeg 03Timo Rothenpieler 07master:48f7c30bf793: avcodec/nvenc: Drop support for old nvenc api
[17:44:59 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:e5de6de8f091: avutil/pixfmt: Add missing AV_ prefix to PIX_FMT_YUV411P
[18:36:07 CET] <Daemon404> these gsoc emails making me frown...
[18:59:15 CET] <cone-136> ffmpeg 03Michael Niedermayer 07master:19321a295a64: avformat/utils: Terminate SPS in avci100_1080i_extradata to avoid warnings during parsing
[19:12:30 CET] <kierank> why does avfilter leave threads lying around
[19:12:36 CET] <kierank> this is insane
[19:13:25 CET] <durandal_1707> Fill bug report
[19:13:58 CET] <kierank> I can't
[19:14:00 CET] <kierank> it's an api bug report
[19:38:30 CET] <jamrial> kierank: if it can't be reproduced with ffmpeg.c, can't you then post a sample program that one can compile to reproduce it?
[19:38:40 CET] <kierank> I think I know why it happens
[19:39:05 CET] <kierank> my fault i think
[19:39:23 CET] <kierank> I'm surprised it did slice threading behind my back
[20:00:46 CET] <nevcairiel> look at t hat, libav found another pet project to annoy devs about, structure packing \o/ gotta squeeze out the last 2 bytes from those pesky codec contexts!
[20:03:06 CET] <iive> i thought that structures are packed so the most common members are in the same cache lines
[20:03:39 CET] <Daemon404> thats hardly an issue with cache sizes nowadays
[20:03:41 CET] <iive> but well, why not make decoders even more slower :)
[20:03:43 CET] <Daemon404> certainly not for a 10 byt struct
[20:04:06 CET] <nevcairiel> 12 with padding!
[20:04:07 CET] <nevcairiel> :P
[20:04:13 CET] <Daemon404> ho ho ho.
[20:04:32 CET] <iive> well, their memory size is even more irrelevant :)
[20:04:43 CET] <nevcairiel> in before they split functionally related parameters because they would create a hole
[20:05:05 CET] <kierank> For big structs like x264_t it is useful
[20:05:14 CET] <nevcairiel> imagine something in the size of h264context with everything ordered by size, instead of functional relationship
[20:05:43 CET] <Daemon404> except that never happened for the big stuff anyway
[20:05:52 CET] <Daemon404> so nit pick about tiny new ones instead
[20:06:10 CET] <iive> i have another proposal!
[20:06:18 CET] <iive> SORT THEM ALPHABETICALLY!
[20:06:31 CET] <Daemon404> you must also pack them though
[20:06:35 CET] <Daemon404> what a fun game it could be
[20:07:07 CET] <iive> well, if you sort by type first, it should come naturally
[20:07:08 CET] <nevcairiel> MS just cheats in their big structs, and issues a pragma to disable member alignment :p
[20:07:23 CET] <nevcairiel> ie. in the dxva structs
[20:07:26 CET] <iive> gcc also have such thing.
[20:07:49 CET] <nevcairiel> of course, same pragma even
[20:08:01 CET] <iive> but it can't work on all architectures and is making code slower.
[22:31:06 CET] <kierank> av500: did you see loongsoon patches
[22:31:13 CET] <kierank> loongsoon got jealous of MIPS
[22:33:24 CET] <nevcairiel> wtf is loongsoon
[22:34:22 CET] <kierank> loongson*
[22:34:26 CET] <kierank> chinese processor
[22:35:38 CET] <nevcairiel> more optimizations noone cares about i guess
[22:36:10 CET] <nevcairiel> as long as its not like crazy mips, where they actually have a different C implementation for mips for some reason, and not asm =p
[22:36:54 CET] <nevcairiel> i wonder what happened to the aac patches, he kind of vanished after he had to bother fixing the mips stuff :p
[22:41:56 CET] <jamrial> some of the alpha optimizations are also c for some reason
[22:44:20 CET] <nevcairiel> carefully crafted C that produces desired code on one specific  compiler
[22:44:23 CET] <nevcairiel> what could go wrong
[23:22:31 CET] <iive> nevcairiel: i think the aac code should be merged
[23:32:51 CET] <BBB> didnt we remove most of these optimizations?
[23:42:57 CET] <kierank> we should probably get rid of TOMI
[23:42:57 CET] <kierank> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavutil/tomi/intreadwrite.h;h=7dec4158d3cd1578d27e4b3c20a0ba721d1c898b;hb=HEAD
[23:43:40 CET] <kierank> iirc it was some thing by a crazy guy who claimed to be a special forces operative
[23:44:16 CET] <wm4> yep definitely should be part of ffmpeg
[23:44:20 CET] <wm4> moar is better
[23:44:49 CET] <kierank> wm4: yeah, someone might use it
[23:45:03 CET] <kierank> same with segfaulting filters
[23:45:06 CET] <kierank> someone might use it
[23:45:11 CET] <wm4> even if not, it doesn't hurt leaving it in!
[23:51:49 CET] <cone-086> ffmpeg 03Giorgio Vazzana 07master:62e2fae8e402: ffmpeg: fix option name in documentation
[00:00:00 CET] --- Wed Mar 25 2015


More information about the Ffmpeg-devel-irc mailing list