[Ffmpeg-devel-irc] ffmpeg-devel.log.20160421

[burek]

[00:54:52 CEST] <cone-241> ffmpeg 03Michael Niedermayer 07master:9f36ea57ae6e: avcodec/avpacket: Fix off by 5 error
[01:16:17 CEST] <Daemon404> where exactly does this 5 come from...
[01:20:44 CEST] <rcombs> 4 bytes size, 1 byte type
[02:31:46 CEST] <BBB> durandal_170: is tm2rt in any way related to tm2?
[02:33:41 CEST] <durandal_170> Dunno, not much
[03:32:11 CEST] <KGB> [13FFV1] 15ablwr opened pull request #17: CSS stylesheet now locally available (06master...06master) 02https://git.io/vwCkP
[08:43:40 CEST] <pfelt> evening all.  in ffmpeg_opt.c what is the intended difference between init_complex_filters() and configure_complex_filters() ?  my filter init() function is getting called twice with two different contexts from those two functions and its messing my code up
[09:16:46 CEST] <pfelt> grumble.  i just found this in code:   /* this graph is only used for determining the kinds of inputs * and outputs we have, and is discarded on exit from this function */
[09:17:02 CEST] <pfelt> that isn't good.  it screws up the fifo i'm trying to use
[13:22:07 CEST] <durandal_1707> michaelni: could you backport few commits?
[13:26:25 CEST] <michaelni> durandal_1707, sure, which ?
[13:29:12 CEST] <durandal_1707> michaelni: latest takdec and apedec
[13:31:33 CEST] <KGB> [13FFV1] 15michaelni closed pull request #17: CSS stylesheet now locally available (06master...06master) 02https://git.io/vwCkP
[14:02:11 CEST] <cone-648> ffmpeg 03Michael Niedermayer 07master:0de99ab06f70: avcodec: Add avpriv_codec_get_cap_skip_frame_fill_param()
[14:02:11 CEST] <cone-648> ffmpeg 03Michael Niedermayer 07master:a0b92788a885: avformat/utils: Remove use of caps_internal as it is not public API
[14:10:22 CEST] <michaelni> durandal_1707, 9149e9c0baaec122bc3da925d6068dffa60b5427 and 38797a8033d061ade58b30b8ac86da222fe42a84 ? 
[14:10:45 CEST] <Daemon404> durandal_1707, why dont you learn to use cherry pick
[14:10:48 CEST] <Daemon404> instead of just complaining
[14:10:50 CEST] <Daemon404> and being lazy
[14:11:00 CEST] <Daemon404> it's getting kind of annoying.
[14:12:34 CEST] <durandal_1707> michaelni: yes
[14:14:50 CEST] <durandal_1707> Daemon404: I don't like to mess with biblical branches in fear I will fsck smthing up
[14:15:04 CEST] <durandal_1707> *nonlocal
[14:15:20 CEST] <Daemon404> that's a bs excuse. it's a cherry pick of a localized file, which you wrote.
[14:15:51 CEST] <wm4> michaelni, Daemon404: ok if I push the decode API patches?
[14:16:05 CEST] <Daemon404> wm4, fine by me.
[14:17:00 CEST] <michaelni> wm4, should be ok
[14:18:13 CEST] <wm4> should I squash patch 2/3 into 1/3?
[14:18:34 CEST] <michaelni> no oppinion ...
[14:18:57 CEST] <nevcairiel> 2 is just a fix on top of 1, isnt it
[14:19:11 CEST] <wm4> in a way
[14:20:16 CEST] <wm4> I'll squash it
[14:21:29 CEST] <wm4> hm I guess I'll add a version bump too
[14:21:33 CEST] <wm4> it was skipped, apparently
[14:29:52 CEST] <cone-648> ffmpeg 03Aaron Boxer 07release/3.0:5d79566ab3dd: avcodec/j2kenc: Add attribution to OpenJPEG project:
[14:29:53 CEST] <cone-648> ffmpeg 03Marios Titas 07release/3.0:47f0d80ee01b: avfilter/src_movie: fix how we check for overflows with seek_point
[14:29:54 CEST] <cone-648> ffmpeg 03Michael Niedermayer 07release/3.0:6103a8453fad: Changelog: Make formating consistent
[14:29:55 CEST] <cone-648> ffmpeg 03Michael Niedermayer 07release/3.0:325d0b64d5a4: avcodec/pngdec: Fix alpha detection with skip_frame
[14:29:56 CEST] <cone-648> ffmpeg 03Michael Niedermayer 07release/3.0:14fdebc4ffcb: avcodec/bmp_parser: Ensure remaining_size is not too small in startcode packet crossing corner case
[14:29:57 CEST] <cone-648> ffmpeg 03Ivan 07release/3.0:76573c5239fb: avcodec/h264: Fix for H.264 configuration parsing
[14:29:58 CEST] <cone-648> ffmpeg 03Michael Niedermayer 07release/3.0:264c9fe6aada: avcodec/avpacket: Fix off by 5 error
[14:29:59 CEST] <cone-648> ffmpeg 03Paul B Mahol 07release/3.0:6fdd122b11e5: avcodec/apedec: fix decoding of stereo files with one channel full of silence
[14:30:00 CEST] <cone-648> ffmpeg 03Paul B Mahol 07release/3.0:7626fb6cbf5f: avcodec/takdec: add code that got somehow lost in process of REing
[14:30:50 CEST] <Daemon404> oh shit i broke libnut
[14:31:27 CEST] <Daemon404> er
[14:31:33 CEST] <Daemon404> hey michaelni where do i *get* libnut?
[14:31:42 CEST] <Daemon404> google provides a bunch of weird github forks
[14:33:54 CEST] <cone-648> ffmpeg 03wm4 07master:7fc329e2dd62: lavc: introduce a new decoding/encoding API with decoupled input/output
[14:33:55 CEST] <cone-648> ffmpeg 03wm4 07master:656b07b5a969: lavf: use new decode API
[14:34:16 CEST] <wm4> unfortunately I have no idea how to fix ffmpeg to use the new API
[14:35:02 CEST] <wm4> and I suppose removing the DTS "trick" on draining would not be accepted
[14:35:16 CEST] <Daemon404> hack, you mean hack
[14:36:02 CEST] <Daemon404> the fact that i cannot even find the repo, really does say something about libnut
[14:36:41 CEST] <Daemon404> git://git.ffmpeg.org/nut seems to be it
[14:36:48 CEST] <Daemon404> but it is literally only reference from multimediawiki
[14:36:52 CEST] <Daemon404> which is hardly canonical
[14:37:21 CEST] <Daemon404> it isnt listed anywhere.
[14:37:23 CEST] <Daemon404> ^ michaelni 
[14:37:33 CEST] <wm4> can't we just drop libnut support
[14:37:44 CEST] <Daemon404> oh... its a git mirror
[14:37:45 CEST] <wm4> is there even a single half-way non-insane argument for keeping it
[14:37:47 CEST] <Daemon404> of svn://svn.mplayerhq.hu/nut
[14:37:50 CEST] <Daemon404> even better
[14:38:05 CEST] <Daemon404> wm4, the only argument i can think of is that it helps to have two implementations of nut to keep it "honest"
[14:38:14 CEST] <wm4> lol.
[14:39:11 CEST] <Daemon404> holy crap that svn co is slow
[14:40:10 CEST] <nevcairiel> why didnt you use the git mirror
[14:40:49 CEST] <Daemon404> i wanted to see if upstream even still existed
[14:42:26 CEST] <Daemon404> real    0m25.744s
[14:42:32 CEST] <Daemon404> for ~10 files
[14:42:57 CEST] <Daemon404> ... huh... there is no configure... just config.mak
[14:43:00 CEST] <Daemon404> which you edit by hand
[14:43:01 CEST] <Daemon404> ghetto.
[14:44:06 CEST] <wm4> the server must be located in someone's basement, running on dialup
[14:44:38 CEST] <Daemon404> yes that sounds like mplayer's svn server
[14:45:11 CEST] <Daemon404> wm4, btw it took almost 2 weeks for someone to notice libnut.c didnt even compile.
[14:45:23 CEST] <wm4> as expected
[14:47:14 CEST] <Daemon404> ff_parse_specific_params
[14:47:16 CEST] <Daemon404> nice fucn name
[14:55:24 CEST] <Daemon404> shit wrong email in my git config
[15:59:43 CEST] <Daemon404> oh hey
[16:00:16 CEST] <Daemon404> BBB / michaelni - i got persmission to share a sample which triggers a race in the h264 decoder (causing broken output)
[16:04:30 CEST] <BBB> awesome!
[16:04:42 CEST] <BBB> can you file a trac ticket and poke me with it friday?
[16:04:51 CEST] <BBB> if michaelni fixes it earlier, thats ok also
[16:05:05 CEST] <Daemon404> BBB, i will file a ticket
[16:05:17 CEST] <Daemon404> however i cannot guarantee carl wont close it as FUCKYOU
[16:07:15 CEST] <BBB> hahahahhahahaha
[16:07:17 CEST] <BBB> well, ok
[16:07:27 CEST] <BBB> if the file is somewhere, Ill look anyway
[16:07:32 CEST] <BBB> I dont mind carl so much
[16:07:40 CEST] <BBB> as long as he doesnt delete the file
[16:08:49 CEST] <Daemon404> the file will be hosted elsewhere
[16:08:57 CEST] <Daemon404> its probsbly too large for trac upload
[16:09:05 CEST] <Daemon404> (266 mb)
[16:15:08 CEST] <Daemon404> BBB, ^
[16:16:30 CEST] <fritsch> Daemon404: i wonder how often you end up in /dev while navigating to your ~/dev directory
[16:16:51 CEST] <Daemon404> not often
[16:16:57 CEST] <fritsch> hehe, but it happens
[16:16:57 CEST] <Daemon404> since i type cd ~/dev/<...>
[16:17:02 CEST] <Daemon404> itll fail with /dev
[16:17:02 CEST] <Daemon404> ;)
[16:18:34 CEST] <JEEB> I usually have ~/projects/
[16:19:04 CEST] <JEEB> (and I just learned that irssi can tab-autocomplete random paths)
[16:19:26 CEST] <Daemon404> ... so it can
[16:19:27 CEST] <Daemon404> wtf
[16:19:46 CEST] <JEEB> I knew it did it with stuff like /script or /config
[16:19:47 CEST] <Daemon404> 10 years of use and i didnt know this.
[16:19:52 CEST] <JEEB> inorite
[16:20:58 CEST] <durandal_1707> Why people use libnut when there is superior implementation already in lavf
[16:22:08 CEST] <cone-648> ffmpeg 03Derek Buitenhuis 07master:be5fde92ff6c: libnut: Pass stream to ff_parse_specific_params
[16:22:09 CEST] <BtbN> because they are nuts
[16:23:27 CEST] Action: JEEB slowly claps
[16:24:11 CEST] <wm4> no, because they think the lib is needed
[16:24:19 CEST] <wm4> because if not, why would we support it?
[16:24:40 CEST] <wm4> (the truth is that we support it because we never delete anything, not even totally useless things)
[16:24:55 CEST] <Daemon404> also our wiki for building includes roughly 100000 libraries
[16:25:01 CEST] <Daemon404> so everyone follows those instructions
[16:27:10 CEST] <atomnuker> wm4: we delete stuff, 2 aac encoders, libstagefright, something else I can't remember...
[16:28:10 CEST] <Daemon404> but not libutvideo
[16:28:40 CEST] <wm4> is that other aac encoder really deleted
[16:32:16 CEST] <atomnuker> libaacplus was
[16:37:16 CEST] <rcombs> and libvo_aac
[16:43:46 CEST] <wm4> but not libfaac or libfdk
[16:44:03 CEST] <Daemon404> libfdk should not be removed
[16:44:27 CEST] <atomnuker> the latter has it's place, the former will be gone as soon as I find time to rewrite the fast coder
[16:55:28 CEST] <cone-648> ffmpeg 03Martin Storsjö 07master:fab8156b2f30: avio: Copy URLContext generic options into child URLContexts
[16:55:29 CEST] <cone-648> ffmpeg 03Derek Buitenhuis 07master:f8e89d8a297e: Merge commit 'fab8156b2f30666adabe227b3d7712fd193873b1'
[17:02:08 CEST] <cone-648> ffmpeg 03Martin Storsjö 07master:136c3438bbdb: tcp: Use rw_timeout for setting the connect/listen timeouts
[17:02:09 CEST] <cone-648> ffmpeg 03Martin Storsjö 07master:3ee2ec5ec1e3: unix: Use rw_timeout for setting the connect timeout
[17:02:10 CEST] <cone-648> ffmpeg 03Derek Buitenhuis 07master:ba2d37e9ada5: Merge commit '136c3438bbdb56a5d2f1f0f486f180641dc6dda0'
[17:02:11 CEST] <cone-648> ffmpeg 03Derek Buitenhuis 07master:7e3d3a6a895f: Merge commit '3ee2ec5ec1e39a438f89302d949c93a1b5d365a2'
[17:04:04 CEST] <cone-648> ffmpeg 03Diego Biurrun 07master:b298b36fc008: fate: Only run SRTP test if SRTP code is enabled
[17:04:05 CEST] <cone-648> ffmpeg 03Diego Biurrun 07master:3dfbf32b95ae: build: Drop redundant removal of compiled object files
[17:04:06 CEST] <cone-648> ffmpeg 03Derek Buitenhuis 07master:420806f0d14a: Merge commit 'b298b36fc008ad94a24929fe770c8189d96bcac4'
[17:04:07 CEST] <cone-648> ffmpeg 03Derek Buitenhuis 07master:53107052b2a1: Merge commit '3dfbf32b95ae69a9b15cd4912bc1c68fa16b4093'
[17:04:41 CEST] <cone-648> ffmpeg 03Diego Biurrun 07master:439929859ae0: testprogs: Clean up #includes
[17:04:42 CEST] <cone-648> ffmpeg 03Derek Buitenhuis 07master:32c044cbc640: Merge commit '439929859ae0eb9542d3bb8a0c856bd5a1d1ec48'
[17:50:20 CEST] <cone-648> ffmpeg 03Diego Biurrun 07master:3b08d9d932ee: testprogs: K&R formatting cosmetics
[17:50:21 CEST] <cone-648> ffmpeg 03Derek Buitenhuis 07master:5f587b1daf90: Merge commit '3b08d9d932eef09403074d5af31e10d8011e840b'
[17:52:24 CEST] <cone-648> ffmpeg 03Anton Khirnov 07master:a638e9184d63: vf_fade: make sure the slice end is always in the frame
[17:52:25 CEST] <cone-648> ffmpeg 03Derek Buitenhuis 07master:aa40df483b21: Merge commit 'a638e9184d63e57e67901f34afe919fd56fd3ac4'
[17:53:28 CEST] <Daemon404> who is our qsv person
[17:53:37 CEST] <Daemon404> qsvdec_h2645: switch to the new BSF API
[17:53:40 CEST] <Daemon404> looks nontrivial to merge
[17:57:13 CEST] <jamrial_> Daemon404: Ivan Uskov maybe?
[18:02:06 CEST] <nevcairiel> like those people are going to be of much help =p
[20:15:47 CEST] <cone-648> ffmpeg 03Paul B Mahol 07master:c85d04251df6: avcodec: add TrueMotion 2.0 Real Time decoder
[20:24:43 CEST] <cone-648> ffmpeg 03Michael Niedermayer 07master:7b7c338e9a73: avfilter: Add AV_OPT_FLAG_FILTERING_PARAM to where it was missing
[21:09:32 CEST] <wm4> how does h264 code interlaced video? separate fields, or something worse?
[21:10:20 CEST] <JEEB> the cable tv mafia made there be two modes
[21:10:23 CEST] <JEEB> MBAFF and PAFF
[21:10:35 CEST] <JEEB> and all the implementors rejoiced
[21:10:39 CEST] <JEEB> (not)
[21:10:47 CEST] <wm4> yeah I heard it's complex
[21:12:00 CEST] <nevcairiel> mbaff sounds good in concept, but implementation wise its of course complex
[21:13:27 CEST] <atomnuker> afaik MBAFF is more efficient
[21:22:19 CEST] <nevcairiel> definitely can be
[21:34:58 CEST] <iive> it would have been quite a bummer if it was complex, slow and inefficient
[21:35:06 CEST] <iive> also, interlace must die
[21:42:51 CEST] <fritsch> next gen live tv standard has it whiped
[21:42:56 CEST] <fritsch> hevc 10 bit uhd progressive
[21:50:12 CEST] <Shiz> 12 bit!1
[22:02:21 CEST] <iive> well, iso14496-2 (Mpeg4 ASP) also had it wiped out... but it came back.
[23:21:14 CEST] <pfelt1> afternoon all.  i've gotten pretty far on my filter to allow me to swap streams without killing ffmpeg, however i've hit a small problem.  every time i swap streams my audio and video get out of sync by about 1 second.  only thing i can think of is that the audio data is in a buffer waiting to be output (something like multiple frames per packet).  could that be the case, and if so, is there a way to just snarf all the remaining frames from the buffer and
[23:39:03 CEST] <pfelt1> hmm.  actually that can't be it.  the av_frame_get_best_effort_timestamp() is resetting to 0.  it's just that the decoder is able to pull audio before video
[00:00:00 CEST] --- Fri Apr 22 2016