[Ffmpeg-devel-irc] ffmpeg-devel.log.20150428

[burek]

[00:24:13 CEST] <cone-328> ffmpeg 03Michael Niedermayer 07master:0e7444f6e400: avcodec/hevc: Remove current_sps
[00:55:11 CEST] <BtbN> How do i propperly put a comment on top of a git email, without it beeing part of the commit message? Last time i tried it stripped the message entirely.
[00:56:48 CEST] <wm4> BtbN: format-patch --cover-letter
[00:56:56 CEST] <wm4> or something close to it
[01:16:32 CEST] <BtbN> wm4, looks good, thanks
[02:45:16 CEST] <cone-328> ffmpeg 03Vignesh Venkatasubramanian 07master:8ca170a78ab6: libavformat/webm_chunk: Write first audio packet.
[03:32:52 CEST] <cone-328> ffmpeg 03Andreas Cadhalpun 07master:464c49155ce7: apedec: set s->samples only when init_frame_decoder succeeded
[04:00:09 CEST] <cone-328> ffmpeg 03Vignesh Venkatasubramanian 07master:b410c69fb666: webmdashenc: Fix UTCTiming Element
[08:39:26 CEST] <anshul_mahe> ubitux: in ass format isn't \N is treated as newline, or I have to add newline(\n\r) with \N
[08:41:04 CEST] <ubitux> \N is for presentation, \r\n is for output
[08:48:14 CEST] <ubitux> i wonder how you ended up with such output
[08:48:18 CEST] <ubitux> (does it even work?)
[08:48:25 CEST] <ubitux> can you share the .ts?
[08:48:31 CEST] <anshul_mahe> ubitux:I really dont get what u mean, but is it related to changing code of closed caption, or some more ffmpeg option needed to produce output
[08:48:59 CEST] <ubitux> you should have one "Dialogue" per line (physical/file)
[08:49:16 CEST] <ubitux> \N is part of the markup; it produces a line return at the rendering
[08:49:24 CEST] <ubitux> it's 2 different thing
[08:49:26 CEST] <ubitux> +s
[08:49:58 CEST] <anshul_mahe> I use this file http://gsocdev.ccextractor.org/~anshul/test_video/Closedcaption_rollup.ts
[08:50:18 CEST] <ubitux> currently you have multiple "Dialogue" on the same line (at least on the mailing list)
[08:50:31 CEST] <ubitux> so technically, it just won't work at all
[08:51:00 CEST] <anshul_mahe> yes, I usually use the rendered output by ass
[08:51:34 CEST] <anshul_mahe> how would decoder know that they are going to be rendered or just used as it is
[08:52:22 CEST] <ubitux> ?
[08:52:30 CEST] <ubitux> that ass can't be rendered correctly
[08:52:43 CEST] <ubitux> you produce a broken output
[08:52:50 CEST] <ubitux> it can't work
[08:53:44 CEST] <anshul_mahe> may be, I am not sure. can you try that ts file and produce srt output. it show things fine.
[08:53:46 CEST] <ubitux> http://pastie.org/pastes/10117649/text
[08:53:51 CEST] <ubitux> you need to produce something like this
[08:53:53 CEST] <ubitux> no
[08:53:58 CEST] <ubitux> it's broken, period.
[08:54:08 CEST] <ubitux> this generated ass is invalid
[08:54:53 CEST] <ubitux> but i wonder how you ended up with this
[08:55:00 CEST] <ubitux> the api shouldn't allow you to generate such file
[08:55:07 CEST] <ubitux> unless you are abusing it in a very nasty way
[08:55:39 CEST] <ubitux> anyway, gtg afk
[08:57:09 CEST] <anshul_mahe> ok, I would try, it looks like i have to add \N in between dialogue. at same screen. but at end of screen actual \n\r should be used
[08:59:04 CEST] <anshul_mahe> thanks, for help
[11:36:24 CEST] <saste> are you aware of any mpeg ts analyzer available on linux?
[11:36:46 CEST] <saste> possibly something which can be automated?
[12:26:33 CEST] <cone-437> ffmpeg 03Maneesh Gupta 07master:cf234552b83a: OpenCL: Avoid potential buffer overflow in cmdutils_opencl.c
[12:29:11 CEST] <kierank> saste: dvbsnoop
[12:29:21 CEST] <kierank> lglinskih_: congratulations
[12:30:18 CEST] <lglinskih_> kierank: thank you ^_^
[12:30:40 CEST] <saste> kierank, thanks
[12:40:51 CEST] <cone-437> ffmpeg 03Maneesh Gupta 07master:91305c60261c: OpenCL: Fix ABI incompatibility issues
[12:41:18 CEST] <drwx> hi, could anyone look at this bug? https://trac.ffmpeg.org/ticket/2975
[12:41:26 CEST] <drwx> also, https://trac.ffmpeg.org/ticket/4487 looks like a duplicate
[12:45:24 CEST] <wm4> drwx: yeah, that looks somewhat serious
[12:47:59 CEST] <drwx> wm4: yeah because of this i had to remux a lot of files that i encoded, never realised why the audio was slightly OOS
[12:52:26 CEST] <cone-437> ffmpeg 03Andreas Cadhalpun 07master:699341d647f7: apedec: prevent out of array writes in decode_array_0000
[13:50:46 CEST] <alexvf> hi
[13:50:49 CEST] <alexvf> how is a player using libavformat supposed to know the teletext page of a teletext subtitle stream? does it need to parse codec extradata or is there a more civilized way?
[14:11:22 CEST] <alexvf> which is the right channel to discuss libavformat usage?
[16:11:23 CEST] <cone-437> ffmpeg 03Michael Niedermayer 07master:e2877bdf3862: ffmpeg: remove incorrect network deinit
[16:59:21 CEST] <HSKW> hello! i need help to watch/record an .ts segment (3seconds durations) with same name every time
[17:44:38 CEST] <Daemon404> https://www.dropbox.com/s/rkznx34diznb0lb/watgcc.png?dl=0 <-- lolgcc
[17:46:34 CEST] <nevcairiel> so it works thist ime? :p
[17:47:01 CEST] <Daemon404> yep!
[17:47:04 CEST] <Daemon404> (i think(
[17:47:54 CEST] <nevcairiel> building with warnings-as-errors is always risky if new compilers show up =p
[17:48:04 CEST] <Daemon404> g 42
[17:48:06 CEST] <iive> hey, wasn't there a gcc x.y.0 version that didn't miscompile ffmpeg?
[17:48:27 CEST] <nevcairiel> 4.8 i think
[17:48:30 CEST] <Daemon404> thats why they skipped .0 this time.
[17:48:32 CEST] <Daemon404> ;)
[17:48:41 CEST] <nevcairiel> their tarball is called 5.1.0
[17:48:53 CEST] <HSKW> hello! i need help to watch/record an .ts segment (3seconds durations) with same name every time
[17:49:10 CEST] <Daemon404> nevcairiel, damn.
[17:49:50 CEST] <nevcairiel> whats with everyone and inventing ways to quickly inflate versino numbers
[17:51:27 CEST] <BtbN> inflating version numbers do help a bit to motivate users to upgrade faster.
[17:51:43 CEST] <Daemon404> users dont upgrade gcc
[17:51:46 CEST] <Daemon404> diestros do
[17:51:48 CEST] <Daemon404> distros*
[17:51:58 CEST] <BtbN> Yeah, for gcc it's quite pointless
[17:52:26 CEST] <nevcairiel> chrome and firefox mostly auto-update, so its equally pointless, well, having any kind of version number kinda is in that case
[17:55:52 CEST] <jamrial> no, it was 4.9.0
[17:56:07 CEST] <nevcairiel> didnt 4.9 have the flac trouble
[17:56:11 CEST] <jamrial> yeah
[17:56:42 CEST] <jamrial> afaik it was golomb
[17:56:53 CEST] <jamrial> flac and other codecs were affected
[17:57:09 CEST] <nevcairiel> then 4.9.0 isnt a version that didnt miscompile ffmpeg =p
[17:57:39 CEST] <jamrial> oh, haha. misread iive's question :p
[17:57:42 CEST] <nevcairiel> i think 4.8.0 fared pretty well, but i dont usually use .0 versions just to be safe
[17:58:02 CEST] <Daemon404> i heard LTO in 5.1 may actually not crash gcc
[17:58:05 CEST] <jamrial> gcc 5.1.0 passes fate on both linux and mingw-w64 here
[17:58:08 CEST] <Daemon404> and, may, stay with me here
[17:58:12 CEST] <Daemon404> make the code *actually* faster
[17:58:13 CEST] <Daemon404> maybe.
[17:58:13 CEST] <nevcairiel> i still didnt bother to setup 5.1
[17:58:34 CEST] <nevcairiel> some early tests with x264 and x265 seemed to indicate gcc 5 produces slower code
[17:58:40 CEST] <Daemon404> rofl
[18:00:07 CEST] <iive> well, it is known that gcc optimizes for newest cpu and older cpu settings kind of produce worse code.
[18:00:39 CEST] <nevcairiel> i think tests were done on a ivy bridge cpu, so hardly the oldest model
[18:01:01 CEST] <jamrial> starting with 4.9, mtune=generic (the default) optimizes for latest intel cpus rather than last century's
[19:41:32 CEST] <cone-437> ffmpeg 03foo86 07master:191dd219d9c9: avformat/dtsdec: Extend DTS core sync word
[19:43:20 CEST] <Daemon404> O_o
[20:03:35 CEST] <rcombs> O_o
[20:04:58 CEST] <wm4> you're O_o'ing at the DTS commit?
[20:05:44 CEST] <rcombs> I was
[20:35:47 CEST] <cone-437> ffmpeg 03Michael Niedermayer 07master:5bebd13d2d86: avcodec/vqavideo: Use ff_tlog() for byte and opcode tracing
[20:55:45 CEST] <cone-437> ffmpeg 03Andreas Cadhalpun 07master:6621105877ce: nutdec: check for negative frame rate in decode_info_header
[22:34:48 CEST] <cone-437> ffmpeg 03Andreas Cadhalpun 07master:3ff1af2b0db7: nutdec: check chapter creation in decode_info_header
[22:54:10 CEST] <cone-437> ffmpeg 03Andreas Cadhalpun 07master:361702660d2c: nutdec: fix memleaks on error in nut_read_header
[23:23:12 CEST] <cone-437> ffmpeg 03Andreas Cadhalpun 07master:7c24ca1bda2d: nutdec: fix illegal count check in decode_main_header
[00:00:00 CEST] --- Wed Apr 29 2015