[Ffmpeg-devel-irc] ffmpeg-devel.log.20151107

burek burek021 at gmail.com
Sun Nov 8 02:05:02 CET 2015


[01:03:56 CET] <cone-479> ffmpeg 03Michael Niedermayer 07master:c719cd6cf79e: avcodec/microdvddec: Check for string end in 'P' case
[03:02:41 CET] <cone-479> ffmpeg 03Michael Niedermayer 07master:43492ff3ab68: avcodec/jpeg2000dec: Clip all tile coordinates
[06:35:53 CET] <atomnuker> warning: 'ff_planar_sample_fmts_array' defined but not used gets spammed when compiling lavfi
[06:37:22 CET] <atomnuker> but wait, the array is defined as static, GCC should ignore any such unused
[06:39:23 CET] <durandal_170> for every filter?
[06:41:40 CET] <atomnuker> yeah
[06:41:53 CET] <atomnuker> clang seems fine though
[06:42:06 CET] <atomnuker> let me just try with non-git-master gcc
[06:51:32 CET] <atomnuker> yeah, gcc-5 doesn't spam that
[06:51:45 CET] <atomnuker> well, none of that matters now
[06:52:09 CET] <atomnuker> just a few seconds after verifying it works my machine shut down
[06:52:15 CET] <atomnuker> (it's been doing that for a while)
[06:52:39 CET] <atomnuker> and my ext4 root finally kicked the bucket after not being shut down normally in a while
[06:53:13 CET] <atomnuker> so yeah, my saturday noon will be very enjoyable as I reinstall
[06:54:05 CET] <atomnuker> thankfully I didn't keep anything important outside of /home
[06:55:25 CET] <atomnuker> I think I might ditch Debian and KDE though
[06:58:04 CET] <atomnuker> 591 days this Debian Unstable installation survived, this is even way before the systemd transition
[07:00:59 CET] <atomnuker> nah, can't be bothered, will just try to repair it
[07:01:24 CET] <jamrial> atomnuker: just use -Wno-unused-const-variable to mute that warning
[07:02:27 CET] <jamrial> for gcc 6 they made it warn about unused const variables, and since fmpeg has plenty of those we get a warn spam
[07:03:17 CET] <jamrial> we'll probably have to add a check to configure for that option before gcc 6 is released
[07:05:42 CET] <jamrial> atomnuker: see http://fate.ffmpeg.org/history.cgi?slot=x86_64-archlinux-gcc-experimental around september. the gcc change in question made the warning count go from 90 to 530 until i muted them :p
[07:08:57 CET] <jamrial> durandal_170: none of the samples at http://samples.mplayerhq.hu/game-formats/xma/ seem to work
[07:10:05 CET] <jamrial> i get errors about missing fmt tag, or sample rate with a value of 0 (ffmpeg tries to use wav demuxer for all of them)
[07:11:49 CET] <durandal_170> jamrial: that needs changes to wav demuxer
[07:13:06 CET] <durandal_170> there are much more 'normal' wavs that are already supported
[07:15:29 CET] <durandal_170> those wavs have XMA2 chunk instead of fmt one and fate samples have it at end instead of start of file
[07:16:50 CET] <durandal_170> also if they have more than 2 channels they are unsupported
[07:20:39 CET] <durandal_170> s/fate//
[07:24:37 CET] <jamrial> ah, i see
[08:09:20 CET] <atomnuker> jamrial: why would they start warning about them?
[12:25:00 CET] <cone-570> ffmpeg 03Will Kelleher 07master:7f7fa90f7b7f: hevc: extract SEI caption data
[12:52:38 CET] <kierank> got bored of vp9 fuzzing
[12:55:46 CET] <kierank> durandal_170: what should I fuzz no
[12:55:47 CET] <kierank> w
[13:01:34 CET] <atomnuker> kierank: ah, you haven't heard of the "Which FFmpeg decoder should I fuzz today?" game
[13:01:46 CET] <durandal_170> You must fuzz everything for 100 cycles
[13:02:06 CET] <atomnuker> it's really simple to play, just open up the libavcodec folder, hit your head on the keyboard and the file you pick is what you fuzz
[13:02:07 CET] <durandal_170> In alphabetical order
[13:02:35 CET] <kierank> meh, not going to fuzz game codecs etc
[13:04:33 CET] <durandal_170> wma codecs, ms codecs
[13:18:37 CET] <cone-570> ffmpeg 03Andreas Cadhalpun 07master:db374790c75f: jvdec: avoid unsigned overflow in comparison
[13:18:38 CET] <cone-570> ffmpeg 03Andreas Cadhalpun 07master:3e8e1a660ea1: apng: use correct size for output buffer
[13:24:41 CET] <kierank> any idea where the old gsoc ffmpeg dirac encoder got to?
[13:24:49 CET] <kierank> i.e where can i find it
[13:52:31 CET] <kierank> upgrading avdev to ubuntu 14.04
[13:52:34 CET] <kierank> some annoying boost packages missing
[14:12:16 CET] <cone-570> ffmpeg 03Michael Niedermayer 07master:183720eb6efe: avformat/segafilm: Only add index entries when the stream exists
[14:37:17 CET] <cone-570> ffmpeg 03Ganesh Ajjanagadde 07master:91bc4de2a4fc: doc/resampler: fix some trivial typos
[15:01:49 CET] <BBB> kierank: what happened to vp9 fuzzing?
[15:01:52 CET] <BBB> kierank: Im waiting!!!
[15:05:07 CET] <funman> kierank: fuzz the fuzzer
[15:06:31 CET] <kierank> BBB: it got painfully slow
[15:06:37 CET] <kierank> like 1-2 months to finish
[15:07:06 CET] <kierank> well to do one cycle
[15:57:41 CET] <BBB> kierank: compared to what?
[15:57:52 CET] <kierank> h264
[15:57:57 CET] <BBB> kierank: and is that because the decoder is complex, or is each decoder so slow?
[15:58:07 CET] <BBB> well, ok, vp9 is a lot more complex than h264...
[15:58:10 CET] <kierank> I think the fuzzer managed to create some files which were really slow to decode
[15:58:12 CET] <BBB> if you remove inline?
[15:58:19 CET] <BBB> does that help?
[15:58:36 CET] <BBB> you can also set artificial resolution limits, e.g. no more than 320x240 or so
[15:58:42 CET] <BBB> that should help quite a bit
[16:11:13 CET] <cone-570> ffmpeg 03Nicolas George 07master:b16e56931271: lavfi: add testsrc2 test source.
[16:11:14 CET] <cone-570> ffmpeg 03Nicolas George 07master:432e27e6e7b6: fate: add mpdecimate test.
[16:11:15 CET] <cone-570> ffmpeg 03Nicolas George 07master:7037f9430b62: lavfi: add realtime filter.
[16:11:16 CET] <cone-570> ffmpeg 03Nicolas George 07master:47ea04ff1058: lavu/opt: enhance printing durations.
[17:01:39 CET] <philipl> fritsch: I'm about to head over.
[17:01:44 CET] <philipl> Leaving in a few minutes.
[17:02:07 CET] <fritsch> philipl: cool - kodi team is already there
[17:02:15 CET] <fritsch> so just take a big sign with you :-)
[17:02:19 CET] <philipl> haha.
[17:03:47 CET] <fritsch> damn, that I did not go ...
[17:03:55 CET] <fritsch> but birthday of wife - would not have been good
[17:06:30 CET] <philipl> Indeed. I wouldn't survive that.
[17:06:56 CET] <fritsch> but it would be good if you meet our head of dvdplayer once in person
[17:07:06 CET] <fritsch> good basis for the future
[17:07:52 CET] <philipl> Yep. I didn't realise there would be anyone from Kodi there, so I'm very glad to hear it.
[17:08:12 CET] <philipl> Well, time for me to go.
[17:27:08 CET] <philipl> fritsch: Is he online anywhere. Wont be easy to find otherwise
[17:29:48 CET] <fritsch> take the mic next to you and ask: Rainer Hochecker please come to this microphone.
[17:30:04 CET] <fritsch> nope - do you have twitter?
[17:30:59 CET] <philipl> It's just breakfast right now.
[17:31:01 CET] <fritsch> https://twitter.com/natedrives <- if you ping him, nate or PR guy - he can arrange a meeting, he is also there
[17:31:20 CET] <fritsch> and fernet does not have mobile internet in US I assume
[17:31:51 CET] <philipl> There is wifi in the building.
[17:32:03 CET] <fritsch> ah - then he checks emails :-)
[17:32:07 CET] <fritsch> fernetmenta at kodi.tv
[17:36:10 CET] <philipl> let's see if that works
[17:37:30 CET] <fritsch> hehe
[18:23:55 CET] <Daemon404> 1-2 months O_o
[18:25:09 CET] <kierank> yeah it ground to a halt for some reason
[18:25:22 CET] <kierank> I think afl was making really really slow files
[18:25:23 CET] <Daemon404> is it hitting swap
[18:48:50 CET] <kierank> dunno might have been
[18:49:30 CET] <kierank> killed it to see if i can merge that old dirac encoder
[18:57:29 CET] <cone-570> ffmpeg 03Nicolas George 07master:d92e0848d9a5: lavfi: remove astreamsync.
[18:57:30 CET] <cone-570> ffmpeg 03Nicolas George 07master:79c1be124e59: lavfi/avf_concat: return immediately after requesting a frame on input.
[18:57:31 CET] <cone-570> ffmpeg 03Nicolas George 07master:67d3f5296ea6: lavfi/fifo: do not assume request_frame() returns a frame.
[18:57:32 CET] <cone-570> ffmpeg 03Nicolas George 07master:d0b82d798d08: lavfi/af_join: partially fix scheduling.
[18:57:33 CET] <cone-570> ffmpeg 03Nicolas George 07master:f53c4b6a186f: lavfi/vf_framepack: fix scheduling.
[18:57:34 CET] <cone-570> ffmpeg 03Nicolas George 07master:a08fb3983f2e: lavfi/af_amix: mostly fix scheduling.
[18:57:35 CET] <cone-570> ffmpeg 03Nicolas George 07master:785ac437be6b: lavfi/af_asyncts: remove looping on request_frame().
[19:29:06 CET] <cone-570> ffmpeg 03wm4 07master:b07cbf672420: mmaldec: add vc1 decoding support
[19:29:07 CET] <cone-570> ffmpeg 03wm4 07master:a55fbfa4c496: mmaldec: correct package buffering accounting
[19:56:10 CET] <Daemon404> kierank, wut
[19:56:33 CET] <kierank> using the machine to do a merge, too slow on my pc
[19:56:48 CET] <Daemon404> ... but why
[19:57:01 CET] <kierank> dirac low delay is quite good
[19:57:07 CET] <kierank> i need it for something
[19:57:14 CET] <Daemon404> why not use libschro
[19:57:26 CET] <kierank> slooow
[19:57:36 CET] <Daemon404> lol and ours was faster?
[19:57:42 CET] <Daemon404> it certainly wasnt ... correct
[19:57:52 CET] <kierank> we'll see
[19:58:00 CET] <kierank> the encoder was very basic lossless vlc coding
[19:58:12 CET] <kierank> I seem to recall that guy got bullied by diego
[19:58:16 CET] <kierank> and he basically quit
[19:58:45 CET] <Daemon404> he should have aligned his =
[19:58:47 CET] <kierank> during the height of the cosmetic crusades
[19:58:49 CET] <kierank> yes
[19:58:58 CET] <kierank> literally all his commits are cosmetics on the decoder merge
[20:00:32 CET] <kierank> is there a way in git of resetting conflicts to use local HEAD?
[20:01:12 CET] <Daemon404> git checkout file?
[20:01:20 CET] <Daemon404> not sure that works during conflicts
[20:01:43 CET] <kierank> error: path 'libavcodec/diracdec.c' is unmerged
[20:01:55 CET] <kierank> it's all just bloody cosmetics that conflict
[20:02:49 CET] <Daemon404> aha
[20:02:56 CET] <Daemon404> git checkout HEAD -- fiel
[20:13:08 CET] <kierank> FF_COMMON_FRAME
[20:13:11 CET] <kierank> blast from the past
[20:21:19 CET] <Daemon404> lul
[20:43:02 CET] <durandal_170> hmm, I'm in, if it produce compliant files I'm for assimilation
[20:44:28 CET] <durandal_170> next on the list: FSB container
[20:44:42 CET] <JEEB> next version of the KGB container?
[20:53:23 CET] <kierank> durandal_170: well it doesn't build
[21:05:37 CET] <durandal_1707> kierank: have repo?
[21:05:52 CET] <kierank> i've given up with the nenjordi repo
[21:05:58 CET] <kierank> I'm now using svn gsoc from 2007!
[21:06:48 CET] <Daemon404> your best bet would just be to do it entiterly by hand
[21:06:58 CET] <Daemon404> entirely*
[21:09:43 CET] <BBB> I remember that
[21:09:55 CET] <BBB> that whole cosmetic phase was cruelly damaging to the project
[22:17:37 CET] <ubitux> what coeffs are we supposed to use in simd code then?
[22:17:40 CET] <ubitux> oups
[22:25:00 CET] <cone-570> ffmpeg 03Michael Niedermayer 07master:7cea3430a56f: avcodec/takdec: Use memove, avoid undefined memcpy() use
[23:01:14 CET] <Daemon404> wm4, ... but why would anyone do that (or why would it be supported)
[23:01:20 CET] <Daemon404> it boggles the mind
[23:02:49 CET] <wm4> well I didn't say which software
[23:02:54 CET] <wm4> but it's mplayer
[23:03:01 CET] <wm4> so, maybe it doesn't count?
[23:03:55 CET] <wm4> it doesn't have a very good reason to do so either
[23:04:14 CET] <Daemon404> well i guessed it was mplayer
[23:04:31 CET] <Daemon404> why else would anyone say it should work
[23:32:50 CET] <cone-570> ffmpeg 03Timothy Gu 07master:4b80b895a9db: pixblockdsp: x86: Condense diff_pixels_* to a shared macro
[23:37:13 CET] <Compn> lol
[23:39:03 CET] <Compn> kierank : did you find the gsoc encoder then ?
[23:39:09 CET] <kierank> yes
[00:00:00 CET] --- Sun Nov  8 2015


More information about the Ffmpeg-devel-irc mailing list