[Ffmpeg-devel-irc] ffmpeg-devel.log.20151102

burek burek021 at gmail.com
Tue Nov 3 02:05:02 CET 2015


[00:26:22 CET] <atomnuker> kierank: up for a pint sometime this week?
[00:26:40 CET] <kierank> maybe
[00:27:52 CET] <atomnuker> k
[00:27:57 CET] <atomnuker> do tell when you know
[00:31:58 CET] <kierank> maybe thurs/fri
[00:45:42 CET] <Plorkyeran> j at BBB> do you still need samples with full code coverage? <-- one of the main awesome things about afl is that you don't
[00:46:00 CET] <Plorkyeran> better starting points will still cut down on the amount of cpu time needed, of course
[01:36:23 CET] <cone-603> ffmpeg 03Ganesh Ajjanagadde 07master:c03044c86ad8: avutil/eval: minor typo
[01:52:45 CET] <BBB> so if the goal is just for ffserver like codepaths, should this code live in internal libav* api?
[03:53:18 CET] <cone-603> ffmpeg 03Timothy Gu 07master:dee74405312a: vf_boxblur: Templatize blur{8,16}
[08:44:48 CET] <cone-207> ffmpeg 03Paul B Mahol 07master:c293ef258cbb: avformat/ipmovie: put video decoding_map_size into packet and use it in decoder
[09:03:02 CET] <cone-207> ffmpeg 03Paul B Mahol 07master:83d20a6a7e8f: avformat/ipmovie: add context to av_log()
[09:13:06 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:b64fe493717c: avformat: always unref the packet after parsing
[09:13:07 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:55abb94b7d9a: avformat: unref packet after storing it in internal packet queue
[11:07:46 CET] <cone-207> ffmpeg 03Paul B Mahol 07master:e6cccf9daa64: avformat/segafilm: implement seeking
[11:07:47 CET] <cone-207> ffmpeg 03Paul B Mahol 07master:0cb09b0511f4: avformat/segafilm: set audio stream duration
[11:07:48 CET] <cone-207> ffmpeg 03Paul B Mahol 07master:85a5cdb4fa74: avformat/segafilm: set video stream duration
[11:43:52 CET] <cone-207> ffmpeg 03Martin Storsjö 07master:567ca142952c: configure: Add -D_CRT_NONSTDC_NO_WARNINGS when building with msvc
[11:43:53 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:9dc9b7e682fc: Merge commit '567ca142952c5be57e52c149c815dfe5d6ac6d41'
[11:44:54 CET] <cone-207> ffmpeg 03Martin Storsjö 07master:00b62968d079: os_support: Don't try to return the service name as a string in getnameinfo
[11:44:55 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:6b5d6c93dda9: Merge commit '00b62968d079e63bf22028f253ac297292436ebe'
[11:45:45 CET] <kierank> Timothy_Gu: https://trac.ffmpeg.org/ticket/1869
[11:45:53 CET] <kierank> that's an easy bug to fix in v210 if you are interested
[11:54:22 CET] <cone-207> ffmpeg 03Kirill Gavrilov 07master:1720791e36f9: png: read and write stereo3d frame side data information
[11:54:23 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:1fc13a626951: Merge commit '1720791e36f9cc24c05efea5bb275ab52156ce50'
[11:55:07 CET] <cone-207> ffmpeg 03Vittorio Giovara 07master:8487987b87a2: h264: Print user data SEI under normal debug verbosity
[11:55:08 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:ed6d451d934c: Merge commit '8487987b87a252963b27b6adfd82887ea5933918'
[11:55:52 CET] <cone-207> ffmpeg 03Vittorio Giovara 07master:b4a82e740743: h264: Print the complete user message
[11:55:53 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:3ff6d40cbcab: Merge commit 'b4a82e740743744136e07054525560351e4892ae'
[11:56:29 CET] <cone-207> ffmpeg 03Vittorio Giovara 07master:f53583ea6e2e: omadec: Fix position of opening parenthesis
[11:56:30 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:b0b752334c29: Merge commit 'f53583ea6e2e3fda4e4aa363f9a62cff4285affb'
[11:58:54 CET] <cone-207> ffmpeg 03Rémi Denis-Courmont 07master:d35d0c723e3c: vdpau: remove dysfunctional H.263 support
[11:58:55 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:d15adeacf3e6: Merge commit 'd35d0c723e3c8fc8cde76bf677f67928f5e179a8'
[11:59:09 CET] <cone-207> ffmpeg 03Luca Barbato 07master:a0562e531723: configure: Add a SONAME entry for the android target
[11:59:10 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:859c4ae402ac: Merge commit 'a0562e531723923b632684c7b51a9dd584bf534f'
[11:59:20 CET] <cone-207> ffmpeg 03Martin Storsjö 07master:407ac22322e5: w32pthreads: Map MemoryBarrier to __sync_synchronize on mingw
[11:59:21 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:c1b712e078b2: Merge commit '407ac22322e5ce67996ec54ef619cafa4c9ceb78'
[11:59:33 CET] <cone-207> ffmpeg 03Diego Biurrun 07master:8161220eee15: h264_parser: Rename close() to h264_close()
[11:59:34 CET] <cone-207> ffmpeg 03Hendrik Leppkes 07master:23dffc7b54c2: Merge commit '8161220eee152dad8b2ea9e2755c78c8e127f747'
[12:45:45 CET] <wm4> some user is still getting "mp3: Header missing" with mp3 in mp4, and ffmpeg git
[12:45:51 CET] <wm4> wasn't this supposed to be fixed recently?
[12:52:28 CET] <cone-207> ffmpeg 03Ganesh Ajjanagadde 07master:430f943e63cc: avfilter/vf_frei0r: use av_strtod instead of strtod for added flexibility
[13:21:47 CET] <JEEB> oh fun, minor input corruption leading to segfault somewhere on the audio side... when I'm lucky. hopefully not thread-time dependent. one run under gdb failed to crash it, now running under valgrind.
[13:22:16 CET] <kierank> JEEB: which codec?
[13:25:09 CET] <JEEB> not sure about the input since it's running on a coworker's box atm, but I think it's AAC
[13:25:23 CET] <JEEB> and then fdk-aac as the encoder
[13:25:42 CET] <kierank> k
[13:27:49 CET] <kierank> J_Darnley: I can't upvote you because of HN noslack
[13:27:51 CET] <kierank> but +1
[13:28:06 CET] <J_Darnley> I'm sorry, what?
[13:28:23 CET] <kierank> "C++ would make it a horrible project to work on."
[13:28:27 CET] <J_Darnley> oh
[13:28:32 CET] <kierank> there's a thing on HN to stop you spending too much time on it
[13:28:37 CET] <J_Darnley> Yes, I know
[13:28:38 CET] <kierank> which blocks you upvoting for 3 hours
[13:28:44 CET] <kierank> but they are all downvoting you
[13:28:44 CET] <J_Darnley> I've got mine turned on
[13:28:47 CET] <kierank> but I want to upvote you
[13:28:55 CET] <J_Darnley> Its only internet points
[13:29:01 CET] <kierank> it's a good c++ troll though
[13:29:43 CET] <J_Darnley> Did I feed it?  :(\
[13:53:03 CET] <Daemon404> reading HN comments is bad for yooue health
[13:53:33 CET] <Daemon404> i see the usual "C is literally hitler" people have shown up
[13:53:50 CET] <Daemon404> your*
[13:59:15 CET] <JEEB> lol
[14:26:53 CET] <JEEB> kierank: btw it was ac3 input, not aac
[14:27:01 CET] <kierank> k
[14:27:11 CET] <JEEB> and seems to be timing-specific so coredump enablement and looping it is
[14:27:12 CET] <kierank> JEEB: naughty naughty as well - pay your dolby racket money :)
[14:28:47 CET] <ubitux> Daemon404: first one being "that wouldn't happen with c++"
[14:28:57 CET] <Daemon404> that was pretty funny
[14:29:05 CET] <Daemon404> especially his rant-reply to J_Darnley 
[14:29:09 CET] <kierank> nooo it's going to drop of the front page
[14:29:12 CET] <kierank> off*
[14:29:35 CET] <Daemon404> why didnt you write your fuzzing scripts in rust, kierank 
[14:29:40 CET] <Daemon404> would have stayed much longer
[14:31:09 CET] <JEEB> :D
[14:31:36 CET] <nevcairiel> why didnt you write ffmpeg in rust, we wouldnt have a need to fuzz then!
[14:31:58 CET] <nevcairiel> (or apparently c++ if your quotes are to be believed)
[14:34:15 CET] <Daemon404> my favourite response was "why would you care if it crashes on invalid input?"
[14:34:22 CET] <Daemon404> i dunno, cause i use it as a library?
[14:34:32 CET] <nevcairiel> thats just a clueless idiot
[14:34:37 CET] <nevcairiel> the world is full of them
[14:35:03 CET] <kierank> Daemon404: yeah
[14:35:12 CET] <kierank> that one was a bit jaw dropping
[14:35:15 CET] <JEEB> lol
[14:35:21 CET] <JEEB> that is quite clueless indeed
[14:35:24 CET] <Daemon404> we all wrap ffmpeg in a ruby script, kierank 
[14:35:27 CET] <Daemon404> as is the HN way
[14:37:28 CET] <Daemon404> im scared to look at that page
[14:37:33 CET] <Daemon404> it was a disaster last time i peeked
[14:40:53 CET] <saste> Daemon404, do you mean the MSVC page?
[14:41:21 CET] <Daemon404> saste, yes
[14:41:24 CET] <Daemon404> lets see
[14:41:38 CET] <Daemon404> hmm no it's less insane now
[14:41:42 CET] <Daemon404> or its a different page
[14:41:51 CET] <Daemon404> there was one which was batch scripts copypasted like 50 times over
[14:41:56 CET] <Daemon404> made by MS employees
[14:42:14 CET] <nevcairiel> that was another page
[14:43:01 CET] <saste> Daemon404, if you have comments/suggestions/flames about that page they're welcome
[14:43:20 CET] <Daemon404> saste, my comments were made when i rejected it from out texi docs ;)
[14:43:29 CET] <Daemon404> our*
[14:43:47 CET] <saste> Daemon404, ok ;-)
[15:04:27 CET] <kierank> Daemon404: have you used asan
[15:04:32 CET] <kierank> this guy on twitter is  saying I should use it
[15:05:06 CET] <Daemon404> i havent, but plenty in here have
[15:05:46 CET] <kierank> what does it do for fuzzing
[15:07:38 CET] <Daemon404> easier detection of out of bounds access i suppose
[15:07:44 CET] <Daemon404> or use after free
[15:36:29 CET] <cone-207> ffmpeg 03Derek Buitenhuis 07master:9ae73d0644cd: hlsenc: Only write PAT/PMT once per segment
[15:37:24 CET] <nevcairiel> old patch is old
[15:38:36 CET] <Daemon404> nevcairiel, indeed
[15:39:00 CET] <nevcairiel> why are you doing evil string printering instead of using av_opt_set_double though?
[15:39:39 CET] <Daemon404> nevcairiel, the option was not a double when i wrote that patch
[15:39:41 CET] <Daemon404> i guess i forgot
[16:16:26 CET] <Compn> ehe r3d 
[17:10:07 CET] <ubitux> Daemon404: re hlsenc: why not use av_opt_set_int?
[17:10:19 CET] <Daemon404> ubitux, double is correct now
[17:10:27 CET] <Daemon404> and does set_int even work for TYPE_FLOAT?
[17:11:00 CET] <ubitux> aren't you using an int format with snprintf?
[17:12:10 CET] <Daemon404> sure, but it is being passed as a string
[17:12:31 CET] <Daemon404> the option type is double atm, but in the past it was float.
[17:12:48 CET] <Daemon404> and INT_MAX / 2 - 1 is the limit mpegtsenc.c checks for.
[17:12:53 CET] <ubitux> oh, ok my bad
[17:12:58 CET] <nevcairiel> opt_set_double also works with float types, iirc
[17:13:00 CET] <ubitux> well there is a av_opt_set_double anyway
[17:13:13 CET] <Daemon404> already had this discussion with nevcairiel 
[17:13:24 CET] <ubitux> i should have backlog, ok
[17:13:28 CET] <ubitux> sorry for the noise
[17:14:37 CET] <nevcairiel> Daemon404: the "set" functions dont care what the underlying type of the option is, it will be converted if appropriate
[17:15:32 CET] <nevcairiel> although not all conversions are supported, you cant set a int to a string field, for example, but you can set an int to a double field
[17:15:33 CET] <Daemon404> aight
[18:27:48 CET] <Kjetil> Hi. I found some references to a ffmpeg-e-distrubtion-audio repo in an old irc-log, which was unfortunately hosted on gitorious. (Now defunct) Anyone got a clone of that repo?
[18:28:08 CET] <Compn>  whats the old url or any other info ?
[18:28:13 CET] <Compn> Kjetil
[18:28:45 CET] <Compn> oh dolby-e audio
[18:29:33 CET] <Compn>  kierank knows soemthing about it
[18:29:41 CET] <Compn> http://gitorious.org/ffmpeg-e-distribution-audio
[18:30:08 CET] <Compn> Kjetil : do you have a sample file that ffmpeg fails to convert or something ?
[18:30:09 CET] <Kjetil> If only archive.org had finished their work :/
[18:30:17 CET] <Compn> and can you share it with us if so?
[18:32:08 CET] <Compn> Kjetil : oh, yes, that gitorious project was authored by kierank ,  so ask him
[18:32:59 CET] Action: Compn goes afk
[19:17:24 CET] <cone-207> ffmpeg 03Tobias Rapp 07master:474665346616: avutil/file_open: avoid file handle inheritance on Windows
[21:31:05 CET] <J_Darnley> What picture is that guy with the png issue looking at? His image clearly begins with a white pixel.
[21:34:58 CET] <drv> probably confused about xxd output... it is in 2-byte groups by default for some reason
[21:36:29 CET] <J_Darnley> ... well now
[21:37:27 CET] <nevcairiel> endianness can confuse you
[21:37:40 CET] <nevcairiel> why it outputs a 16-bit pair is just weird though
[22:41:19 CET] <kylophone> Is there a preferred way of parsing AV_OPT_TYPE_STRING options? Seen a few different techniques in the code...
[23:33:02 CET] <kierank> BBB: is there such thing as a VP9 ES?
[23:33:15 CET] <nevcairiel> ivf :p
[23:33:24 CET] <BBB> right
[23:33:30 CET] <kierank> basically I don't want to fuzz probing but actually fuzz the decoder
[23:33:38 CET] <BBB> no
[23:33:39 CET] <BBB> just ivf
[23:33:42 CET] <BBB> sorry
[23:33:43 CET] <kierank> hmm ok
[23:33:56 CET] <nevcairiel> you can force-override probing decisions on the command line, cant you
[23:34:05 CET] <kierank> can skip 32 bytes I think
[23:36:56 CET] <Compn> yeah just force the decoder /demuxer
[23:36:59 CET] <Compn> and it skips probe IIRC
[23:37:20 CET] <Compn> check -v output to be sure.
[00:00:00 CET] --- Tue Nov  3 2015


More information about the Ffmpeg-devel-irc mailing list