[Ffmpeg-devel-irc] ffmpeg-devel.log.20160504

[burek]

[00:36:27 CEST] <cone-282> ffmpeg 03Paul B Mahol 07master:3395ad48fa2f: avcodec/adpcm: replace char/short/long with int8_t/int16_t/int
[00:43:09 CEST] <durandal_1707> thardin: not enough magic in ffmpeg to use imagemagick
[00:43:57 CEST] <Shiz> magick*
[00:44:20 CEST] <llogan> i dont think the servers use it either
[01:13:52 CEST] <andrey_utkin> is there a way to automatically apply failing FATE tests' diffs to tests/ref/fate/, without manually re-running "make fate"?
[01:29:42 CEST] <jamrial> andrey_utkin: "make fate GEN=1" should regenerate ref files for you
[01:32:11 CEST] <andrey_utkin> thank you
[11:35:42 CEST] <cone-042> ffmpeg 03Paul B Mahol 07master:66eb5b809add: avcodec/adpcm: fix decoding of stereo non 4-bit ADPCM IMA WAV
[14:31:53 CEST] <Daemon404> [FFmpeg-devel] A Hackathon for Codec Develovers
[14:31:55 CEST] <Daemon404> what am i even reading
[14:34:50 CEST] <Compn> i dont see that mail.
[14:35:00 CEST] <petrurares> Hi there, I have a doubt. Any body know how to compute CMP_SHIFT with tiny_psnr?
[14:35:32 CEST] <petrurares> *CMP_SHIFT variable is used in tests
[14:36:54 CEST] <Daemon404> Compn, i do..
[14:37:12 CEST] <Daemon404> http://ffmpeg.org/pipermail/ffmpeg-devel/2016-May/193839.html
[14:37:15 CEST] <Daemon404> so does the archive
[14:38:29 CEST] <Shiz> i'm a develover
[14:38:40 CEST] <Shiz> nice urls
[14:39:12 CEST] <Shiz> what's with the asterisks
[14:45:03 CEST] <BBB> that is really scary
[14:45:11 CEST] <BBB> I mean, Im all for REing old codecs and stuff
[14:45:41 CEST] <BBB> but to use a hackathon as replacement for interview process, thats scary at best and slavery at worst
[14:48:18 CEST] <wm4> isn't that common
[14:49:33 CEST] <Daemon404> BBB, thats pretty standard in SF
[14:49:44 CEST] <Daemon404> and usually they get to keep anything you write too
[14:49:46 CEST] <Daemon404> it's retarded.
[14:49:53 CEST] <BBB> right, so slavery
[14:50:29 CEST] <BBB> if thats standard, Im clearly doing something wrong ;)
[14:50:51 CEST] <Daemon404> .. in SF
[14:50:54 CEST] <Daemon404> SF is "special"
[14:50:57 CEST] <Daemon404> i mean special in a bad way.
[14:52:50 CEST] <BBB> :)
[14:52:50 CEST] <BBB> so you lived in NY
[14:52:51 CEST] <BBB> whats your opinion of Bloomberg?
[14:55:11 CEST] <Daemon404> BBB, the company or the news outlet or?
[14:55:21 CEST] <BBB> the dude
[14:55:24 CEST] <BBB> michael bloomberg
[14:55:26 CEST] <Daemon404> oh
[14:55:37 CEST] <Daemon404> i didnt form any opinions on politics since i couldnt vote
[14:57:20 CEST] <kierank> our mayoral elections are tomorrow
[14:57:26 CEST] <DHE> there's a specific recent news article about him. that what you mean?
[15:00:32 CEST] <Daemon404> kierank, i think i CAN vote in the EU vote tho
[15:00:35 CEST] <Daemon404> and i definitely will be
[15:00:39 CEST] <Daemon404> if possible.
[15:00:41 CEST] <kierank> ah commonwealth I guess
[15:00:45 CEST] <Daemon404> yea
[15:00:54 CEST] <kierank> we are such hypocrites
[15:01:09 CEST] <Daemon404> what government isnt?
[15:01:18 CEST] <Shiz> what person isn't
[15:30:03 CEST] <atomnuker> anyone here who could upload a fate sample?
[15:36:03 CEST] <durandal_170> I will reply to that spammer to refrain from posting such stuff on our ml
[15:38:07 CEST] <wm4> in theory we allow recruiting posts
[15:38:22 CEST] <wm4> atomnuker: maybe michaelni or nevcairiel
[15:45:13 CEST] <RiCON> from the way they used an obfuscated URL thrice, it looks more like spam than informative
[15:56:35 CEST] <Compn> RiCON : i agree, tracking urls are the devil
[16:11:36 CEST] <wm4> rkern: want me to push your patches now?
[16:41:48 CEST] <rkern> yeah, sounds good
[16:57:20 CEST] <Fyr> guys, sorry to bother, but is it only new version, but the previous versions work
[16:57:21 CEST] <Fyr> http://trac.ffmpeg.org/ticket/3339
[16:57:21 CEST] <Fyr> ?
[16:57:31 CEST] <Fyr> just met exactly the same problem.
[16:57:51 CEST] <Fyr> I 'm not sure should I download the older version.
[16:58:08 CEST] <Fyr> if so, then what version should I use?
[17:16:43 CEST] <durandal_1707> The hash of commit that introduced bug is clearly mentioned in bug report
[17:24:13 CEST] <Fyr> right
[17:26:11 CEST] <Fyr> is -c copy 100% saves all the information? i.e. if I do this a thousand times with different containers, the final signal will 100% coincide with the original one?
[17:27:28 CEST] <durandal_1707> Only bitstream stuff
[17:27:57 CEST] <durandal_1707> metadata and format bits are recreated
[17:29:59 CEST] <Fyr> ok
[17:30:21 CEST] <Fyr> only delays will make them different. xD
[18:41:37 CEST] <cone-880> ffmpeg 03Rick Kern 07master:d7cf3610c2fb: lavc/videotoolboxenc: Use shared pixel buffer pool
[18:41:37 CEST] <cone-880> ffmpeg 03Rick Kern 07master:0d4bf3074d33: lavc/videotoolboxenc: Handle hwaccel format as input
[18:41:37 CEST] <cone-880> ffmpeg 03Rick Kern 07master:fbe0cf8ca7e2: lavc/videotoolboxenc: Require hardware encoding
[18:41:37 CEST] <cone-880> ffmpeg 03Rick Kern 07master:645df431449d: lavc/videotoolboxenc: Add entropy setting
[18:41:37 CEST] <cone-880> ffmpeg 03Rick Kern 07master:abe05892bee9: lavc/videotoolboxenc: Fix AVCodecContext.has_b_frames usage.
[18:41:38 CEST] <cone-880> ffmpeg 03Rick Kern 07master:2fbf39382dbf: lavc/videotoolboxenc: Fix DTS
[18:41:38 CEST] <cone-880> ffmpeg 03Rick Kern 07master:272dd437bc6b: lavc/videotoolboxenc: Handle out-of-memory and fix memory leak
[18:41:39 CEST] <cone-880> ffmpeg 03Rick Kern 07master:ca429e725353: lavc/videotoolboxenc: Add realtime encoding property
[18:41:40 CEST] <cone-880> ffmpeg 03Rick Kern 07master:4b806081b25f: lavc/videotoolboxenc: add concatentation properties
[18:41:40 CEST] <cone-880> ffmpeg 03Rick Kern 07master:9d8a38d20b40: lavc/videotoolboxenc: Support for forced I-frames
[18:41:42 CEST] <cone-880> ffmpeg 03Rick Kern 07master:c356b6865d37: lavc/videotoolboxenc: Support pixel aspect ratio
[18:41:42 CEST] <cone-880> ffmpeg 03Rick Kern 07master:dc750194b6c9: lavc/videotoolboxenc: Set colorimetry values
[20:34:41 CEST] <Daemon404> wm4, is the 0-sized issue with decoding or encoding?
[20:35:04 CEST] <Daemon404> woops wrong chan
[20:38:26 CEST] <wm4> yeah, decoding
[20:39:13 CEST] <durandal_1707> why we have probing of random shit in wav demuxer?
[20:39:49 CEST] <wm4> because people put random shit into wav
[20:39:51 CEST] <wm4> literally
[20:40:00 CEST] <Compn> yes.
[20:40:13 CEST] <Compn> ac3, dts, plus a million old codecs
[20:40:26 CEST] <durandal_1707> and we need to support idiots and ignore sane
[20:41:11 CEST] <wm4> it's what ffmpeg does
[20:41:12 CEST] <durandal_1707> file is properly marked as pcm 16 le
[20:41:22 CEST] <wm4> sounds like spdif fun
[20:41:33 CEST] <Compn> lots of spdif :P
[20:41:47 CEST] <durandal_1707> michaelni: ?
[20:42:50 CEST] <nevcairiel> the typical case is just dts in wav because people ripped a DTS CD, which is technically the same format
[20:42:55 CEST] <nevcairiel> its not spdif or anything like that
[20:44:47 CEST] <Compn> if you just want to do a 6 codec ffmpeg (xvid, h264, h265, aac, dts, ac3) i think that project would make things easier :P
[20:45:43 CEST] <durandal_1707> somehow ffmpeg thinks pcm is h264 video
[20:46:16 CEST] <wm4> that's strange, normally I'd expect it detects it as mp3
[20:46:40 CEST] <michaelni> how can i reproduce the pcm-h264 misdetection ?
[20:46:44 CEST] <Compn> raw pcm ?
[20:46:47 CEST] <Compn> header pcm ?
[20:47:21 CEST] <durandal_1707> michaelni: see user ml list
[20:48:14 CEST] <durandal_1707> requesting probe higher than h264 returns fixes it but that is another hack
[20:48:55 CEST] <durandal_1707> It shouldnt probe video codecs at all
[20:49:43 CEST] <Shiz>  +wm4  because people put random shit into wav  
[20:49:44 CEST] <Shiz> https://shiz.me/junk/code/riffogg.py
[20:49:46 CEST] <Shiz> yup
[20:55:53 CEST] <wm4> how satanic
[20:56:03 CEST] <Shiz> visual novels
[20:58:22 CEST] <TD-Linux> I think that might be worse than HDCD
[20:59:39 CEST] <TD-Linux> Shiz, at least you had the taste to write a python script instead of sending a patch to ffmpeg :)
[21:00:42 CEST] <Shiz> yet here i am trying to write an ffmpeg patch for a format that embeds atrac3+ in mpeg private_stream_1 headers
[21:01:31 CEST] <Shiz> s/headers/packets/
[21:02:22 CEST] <durandal_1707> good luck, nobody wants to improve our mpegts demuxer
[21:02:29 CEST] <Shiz> neither do i
[21:02:32 CEST] <Shiz> (also it's ps)
[21:03:42 CEST] <Compn> be happy someone re'd atrac3+ :P
[21:03:55 CEST] <Shiz> yes, i looked at the decoder and was not happy
[21:04:32 CEST] <wm4> atrac? more like detrac
[21:10:54 CEST] <llogan> i removed the "hackaton" from archives
[21:11:19 CEST] <durandal_1707> michaelni: have you found sample?
[21:11:55 CEST] <durandal_1707> llogan: what about snack and pop spam?
[21:12:03 CEST] <llogan> what's the URL?
[21:12:17 CEST] <michaelni> durandal_1707, i found the h264-wav file, yes
[21:14:18 CEST] <durandal_1707> llogan: [FFmpeg-user] Video Engineer Career Opportunity in Toronto 
[21:15:44 CEST] <durandal_1707> no gym, only snacks - fat devs
[21:20:03 CEST] <llogan> durandal_1707: that one is not as bullshitty but i removed it too
[21:20:44 CEST] <durandal_1707> thx
[21:31:28 CEST] <Compn> huh i'm not getting any of these mails
[21:36:28 CEST] <llogan> probably in your spam
[21:37:59 CEST] <BBB> it was definitely spammy
[21:51:43 CEST] <wm4> xine...
[21:55:11 CEST] <Compn> is xine worse than mplayer wm4 ? 
[21:55:12 CEST] <Compn> ehe
[21:57:47 CEST] <wm4> no clue, last time I used xine must have been 13 years ago?
[21:58:47 CEST] <llogan> when was the last time you used RealPlayer?
[21:59:27 CEST] <llogan> they still make money by convinsing state/provincial governments to buy their "helix server" then charge $10K USD for "training"
[21:59:41 CEST] <llogan> my taxes at work
[22:00:15 CEST] <drv> the worst part of xine is the weird name with a dumb pronunciation
[22:01:06 CEST] <llogan> "zeen"?
[22:01:46 CEST] <wm4> I thought it was like "fine"
[22:01:48 CEST] <drv> http://www.xine-project.org/faq#pronounce says "[ksi:n]", which makes no sense to me
[22:02:01 CEST] <Compn> last realplayer use was sometime around 2001 probably
[22:02:32 CEST] <Compn> they pronounce the x
[22:02:35 CEST] <Compn> kyne
[22:02:37 CEST] <Compn> ?
[22:02:42 CEST] <llogan> drv: Kissy Colon N
[22:02:48 CEST] <drv> haha
[22:02:51 CEST] <drv> new official name
[22:24:17 CEST] <cone-333> ffmpeg 03Michael Niedermayer 07master:3646ef6f7c0c: avformat/h264dec: Check pps_id/sps_id fields from parameter sets
[22:24:17 CEST] <cone-333> ffmpeg 03Michael Niedermayer 07master:824906a5416b: avformat/utils: Do not detect video codecs when audio is expected
[22:29:13 CEST] <jamrial> michaelni: will golomb work from lavf on shared builds?
[22:29:35 CEST] <jamrial> get_ue_golomb_31() apparently uses an ff_ prefixed table from golomb.c
[22:30:05 CEST] <jamrial> two, even
[22:30:43 CEST] <durandal_1707> something uses it?
[22:30:53 CEST] <durandal_1707> it can't work
[22:31:10 CEST] <jamrial> durandal_1707: check michaelni's latest h264dec commit
[22:34:47 CEST] <Daemon404> yes im not sure how this will work
[22:35:10 CEST] <Daemon404> (and to fix a stupid bug none the less)
[22:35:12 CEST] <durandal_1707> hmm it works here
[22:35:52 CEST] <durandal_1707> dunno about Windows
[22:38:10 CEST] <durandal_1707> when to apply lbr decoder?
[22:39:30 CEST] <jamrial> i was hoping someone else would comment on it as well
[22:39:37 CEST] <jamrial> or try stuff like fuzzing
[22:47:15 CEST] <michaelni> jamrial, didnt see that, ill fix it
[22:47:31 CEST] <michaelni> that is golomb
[22:49:27 CEST] <Daemon404> hmmm it links here too
[22:49:30 CEST] <Daemon404> but i dont udnerstand how it does
[22:49:35 CEST] <Daemon404> the table is private to avcodec
[22:51:31 CEST] <jamrial> same, tried on mingw and it linked
[22:51:48 CEST] <Daemon404> "my code works, and i dont know why"
[22:52:05 CEST] <jamrial> michaelni: ^
[22:52:13 CEST] <jamrial> it works magically :p
[22:53:29 CEST] <Daemon404> sure feels wrong tho
[23:00:06 CEST] <jamrial> Daemon404: http://pastebin.com/5uwqYhki
[23:00:27 CEST] <Daemon404> i have no context for this paste
[23:01:00 CEST] <jamrial> try compiling with hevc decoder disabled and hevc parser enabled
[23:01:45 CEST] <jamrial> that function doesn't exist. looks like it should be ff_h2645_packet_split() instead, which does exist
[23:02:05 CEST] <Daemon404> ... thats not confusing at all
[23:02:12 CEST] <Daemon404> -_-
[23:02:27 CEST] <Daemon404> test and send a patch if you can please
[23:02:37 CEST] <jamrial> it was called like that before the merge in question i guess
[23:02:55 CEST] <jamrial> and sure
[23:04:00 CEST] <Daemon404> one thing i dont like about doing merges: half the codebase is my fault
[23:04:22 CEST] <jamrial> haha
[23:04:58 CEST] <jamrial> commit them as Guy Incognito
[23:05:21 CEST] <Daemon404> j-b will hurt me if i use a pseudonym
[23:05:26 CEST] <Daemon404> duh
[23:08:23 CEST] <drv> Elvis "The King of Merges" Presley
[23:17:09 CEST] <michaelni> jamrial, it works due to libavformat/golomb_tab.c, but i think its better if the tables arent used from avformat, ill remove the _31 calls 
[23:17:43 CEST] <jamrial> oh, no wonder
[23:18:07 CEST] <cone-333> ffmpeg 03Michael Niedermayer 07master:03e8c1b84398: avformat/h264dec: Avoid get_ue_golomb_31( ) due to ff_ prefixed table from golomb.c
[23:21:20 CEST] <Daemon404> ah
[23:21:27 CEST] <Daemon404> why didnt my grep find that
[23:21:31 CEST] Action: Daemon404 must be tired
[23:27:04 CEST] <cone-333> ffmpeg 03James Almer 07master:fc1c836def08: avcodec/hevc_parser: fix packet_split function name
[23:40:28 CEST] <cone-333> ffmpeg 03James Almer 07master:fe483ac42854: avformat/mpegtsenc: fix usage of AVStream.codec
[00:00:01 CEST] --- Thu May  5 2016