[Ffmpeg-devel-irc] ffmpeg-devel.log.20150403

[burek]

[00:07:10 CEST] <cone-056> ffmpeg 03Stephen Hutchinson 07master:3723a1831037: avisynth: Bump minimum required version to interface version 6
[00:13:02 CEST] <cone-056> ffmpeg 03Lukasz Marek 07master:c043def91f92: lavf/ftp: parse user and password once
[00:13:03 CEST] <cone-056> ffmpeg 03Lukasz Marek 07master:cd3d0d542235: lavf/ftp: remove fixed length buffers from context
[00:41:17 CEST] <Prelude2004c> sorry guys one more question that nobody seems to answer in ffmpeg... for encryption.. we want to use verimatrix for example.. anyone have any docs on ffmpeg could encrypt the HLS output using some key that it obtains from a verimatrix server for example
[00:41:45 CEST] <kierank> go ask verimatrix
[00:41:57 CEST] <kierank> how is ffmpeg meant to know how a secret method of getting keys is meant to work
[00:42:10 CEST] <Prelude2004c> ya i did.. no reply yet
[00:42:13 CEST] <Prelude2004c> not sure if htey know though
[00:42:16 CEST] <Prelude2004c> or maybe they do
[00:42:26 CEST] <kierank> they do but as soon as they see ffmpeg they'll run a mile
[00:42:40 CEST] <Prelude2004c> lol
[00:42:50 CEST] <kierank> that's why dvb-csa was good
[00:42:53 CEST] <kierank> it was vendor neutral
[00:43:09 CEST] <cone-056> ffmpeg 03Stephen Hutchinson 07master:b2c05867f50d: doc: Slightly revise AviSynth version info
[00:48:22 CEST] <Prelude2004c> echo '#EXT-X-KEY:METHOD=AES-128,URI='"$keyFile" into the M3u8 seems easy.. problem is that the file is written every 3 seconds with new content.. wont it remove that.. or is there a param inside the ffmpeg command for segment that allows you to place the key?
[01:09:14 CEST] <nevcairiel> someone put a patch onto the ML some time ago to implement aes encryption for segments using a keyfile you provided, but there were concerns about the patch and i dont think the author  ever responded again
[01:12:37 CEST] <Prelude2004c> developer question... when using the sgment_list options for HLS.. can an option be added to automatically add the " #EXT:AES-128:KEYxxxxxxx " in other words.. a custom line to the segment output ?
[01:13:46 CEST] <nevcairiel> the segments arent encrypted, so why would adding that line make even sense? :d
[01:14:25 CEST] <Prelude2004c> i want to encrypt the segments
[01:14:37 CEST] <Prelude2004c> so output to  " openssl ...." then out to .ts
[01:15:11 CEST] <Prelude2004c> when i do that i need to put the KEY URI inside the m3u8s.. but since its live content and it keeps changing every 3 seconds, it will not stay if i edit manually
[01:15:21 CEST] <Prelude2004c> it has to come from the code that is creating the m3u8 files
[01:15:27 CEST] <nevcairiel> aes encryption isnt super complicated, so a better option would be to convince someone to implement encryption right into the hls muxer
[01:15:47 CEST] <Prelude2004c> the HLS muxer i am using is built into ffmpeg
[01:15:53 CEST] <Prelude2004c> is it not?
[01:16:03 CEST] <Prelude2004c> i am using -f ssegment ....
[01:16:48 CEST] <nevcairiel> thats not the hls muxer
[01:16:51 CEST] <nevcairiel> the hls muxer is -f hls
[01:18:02 CEST] <Prelude2004c> my code looks like " -f ssegment -hls_wap 3 -hls_list_size 15 ...... -segment_format mepgts FIle.ts
[01:18:04 CEST] <nevcairiel> but the point remains, instead of supporting arbitrary random things for random use cases, why not simply support encryption right there, and the problem solves itself
[01:19:13 CEST] <Prelude2004c> we may have a situation where we don't want to enrypt a particular channel
[01:19:31 CEST] <nevcairiel> then you tell ffmpeg not to encrypt
[01:20:09 CEST] <Prelude2004c> right now i am not telling ffmpeg anything... is there some way to tell ffmpeg to encrypt with AES128 and a specific Key?
[01:20:22 CEST] <Prelude2004c> i get the key from verimatrix ... then tell FFMPG to crate the .ts files with that key
[01:20:27 CEST] <nevcairiel> you should really read what i write carefully
[01:20:32 CEST] <nevcairiel> oh well, time to go sleep instead
[01:20:34 CEST] <Prelude2004c> so then players will find the key in the " EXT"
[01:20:47 CEST] <Prelude2004c> sorry i am trying but i dont always understand everything
[01:22:30 CEST] <Prelude2004c> where is the segmenter files .. maybe i can get a programmer to look at it
[01:24:32 CEST] <Prelude2004c> found it
[01:24:33 CEST] <Prelude2004c> :)
[07:52:51 CEST] <j-b> moroning
[09:20:04 CEST] <durandal_1707> morning
[12:53:15 CEST] <cone-952> ffmpeg 03Mariusz SzczepaDczyk 07master:26e0e393d0ca: lavf/avio: Add new types to AVIODirEntryType, bump minor version
[12:53:15 CEST] <cone-952> ffmpeg 03Mariusz SzczepaDczyk 07master:f10aa0e125c1: doc/examples: Handle new types in avio_list_dir
[13:27:59 CEST] <rcombs> anyone around willing to take a look at an HTTP I-D and comment? https://datatracker.ietf.org/doc/draft-combs-http-indeterminate-range/
[13:40:32 CEST] <cone-952> ffmpeg 03Steve Dierker 07master:acbb90407a27: libavformat/mxf.c: Relaxed ProRes pattern
[15:46:47 CEST] <wm4> trac sure is slow...
[15:54:49 CEST] <wm4> ugh, this is fucked up... AV_CODEC_ID_TTF, AV_CODEC_ID_OTF
[15:55:09 CEST] <wm4> the matroska demuxer translates these from mime types... and my own code has to translate this back to mime types
[15:59:11 CEST] <wm4> ah the real mime type is also available
[15:59:16 CEST] <wm4> so the font codec IDs are useless
[16:15:52 CEST] <JEEBsv> :2
[16:18:13 CEST] <wm4> and seriously, only the matroska code uses the TTF/OTF codec IDs in the whole ffmpeg source tree
[16:24:58 CEST] <nevcairiel> well its kinda handy to have the codec ids
[16:25:06 CEST] <nevcairiel> easier than string comparisons in my code =p
[16:27:42 CEST] <wm4> nevcairiel: even then it could just be AV_CODEC_ID_FONT or so
[16:28:02 CEST] <wm4> also lol they're already finding issues in the new directory listing API
[17:03:51 CEST] <JEEBsv> hmm, is this kind of raw->mpegts muxing failure a feature or is something possibly wrong in the input file? http://pastebin.com/tjP5U4W0
[17:04:11 CEST] <JEEBsv> seems to fail even if -r is set before -i
[17:28:47 CEST] <BtbN> https://github.com/BtbN/FFmpeg/commit/4119dbe2b3553398d60207e8743cc147e117285a Is there some better way to do this? I don't realy like that huge chain of if(!strcmp...
[17:30:28 CEST] <wm4> woah
[17:30:38 CEST] <wm4> just make it an array
[17:30:41 CEST] <nevcairiel> put in an exception for 1b, and otherwise use a string to number conversion?
[17:31:06 CEST] <nevcairiel> ie. 10 afaik
[17:31:15 CEST] <nevcairiel> *ie. 1.0 is 10 afaik
[17:31:33 CEST] <nevcairiel> that is assuming the nvenc defines just follow the spec values =p
[17:32:16 CEST] <BtbN> It looks like they are, but i'm not sure if i want to rely on that.
[17:32:48 CEST] <BtbN> For HEVC they are all over the place, or is NV_ENC_LEVEL_HEVC_51 = 153 specified that way?
[17:33:06 CEST] <cone-952> ffmpeg 03Himangi Saraogi 07master:24326930d87d: filtfmts: Avoid null pointer dereference
[17:35:31 CEST] <nevcairiel> but yeah, then put it into a array of structs and iterate over them
[17:36:32 CEST] <nevcairiel> or put it into the avoptions and let it handle the string -> number conversion
[17:36:40 CEST] <nevcairiel> but thats probably equally evil
[17:36:53 CEST] <BtbN> specialy as h264 and h265 share the options, but need diffrent values
[17:36:57 CEST] <BtbN> for the same inputs
[18:11:41 CEST] <cone-952> ffmpeg 03Rong Yan 07master:af5fec9538d8: avcodec/ppc/fdctdsp: POWER LE support in ff_fdct_altivec()
[18:11:53 CEST] <BtbN> https://github.com/BtbN/FFmpeg/commit/1c411964ca204a0b9e09671a71712bdb58fe0eef better
[18:19:12 CEST] <wm4> BtbN: much better IMHO
[19:37:16 CEST] <cone-952> ffmpeg 03Michael Niedermayer 07master:98c7729c133e: ffmpeg: last frame duplication heuristic
[19:50:56 CEST] <cone-952> ffmpeg 03Michael Niedermayer 07master:6eb6956d067a: ffmpeg: Fix indentation
[20:34:43 CEST] <cone-952> ffmpeg 03Andreas Cadhalpun 07master:1f47c2399da2: mpeg4: only show the divx_packed warning, when a packed frame is present
[20:45:11 CEST] <cone-952> ffmpeg 03Lou Logan 07master:44cae912522d: cmdutils: fix "Commmand" typo
[21:09:19 CEST] <cone-952> ffmpeg 03Martin Storsjö 07master:7d097a0fc57f: mpegtsenc: Take max_delay into account when buffering multiple audio packets into one PES packet
[21:09:20 CEST] <cone-952> ffmpeg 03Michael Niedermayer 07master:5b97681eaa5a: Merge commit '7d097a0fc57f0fa8385962a539c657c2f40b5ed0'
[21:29:53 CEST] <cone-952> ffmpeg 03Anton Khirnov 07master:1ae7afd256f9: hevc: split out setting AVCodecContext parameters
[21:29:54 CEST] <cone-952> ffmpeg 03Michael Niedermayer 07master:a7a6194ec410: Merge commit '1ae7afd256f9af1eb4f63f9adcf03d581ce4e2b5'
[21:36:37 CEST] <cone-952> ffmpeg 03Anton Khirnov 07master:7693ba0a0eec: hevc: export stream parameters from extradata
[21:36:38 CEST] <cone-952> ffmpeg 03Michael Niedermayer 07master:d386a523aefe: Merge commit '7693ba0a0eecdcdba71b7fbd9a4a12d1ba7b82aa'
[21:58:12 CEST] <cone-952> ffmpeg 03Anton Khirnov 07master:ce0bc09ee258: hevc: do not change the VPS if its contents are the same
[21:58:13 CEST] <cone-952> ffmpeg 03Michael Niedermayer 07master:41dde62d7f41: Merge commit 'ce0bc09ee2580d49fec90a6eb0de2ba1b580c854'
[22:23:39 CEST] <cone-952> ffmpeg 03Anton Khirnov 07master:df528b11ac60: hevc: make sure no dangling pointers remain around on VPS/SPS change
[22:23:40 CEST] <cone-952> ffmpeg 03Michael Niedermayer 07master:235863463abc: Merge commit 'df528b11ac607de13a7c438f2a51f2119f71a03c'
[23:01:05 CEST] <cone-952> ffmpeg 03Anton Khirnov 07master:b2965ec7761c: FATE: add a test for parameter changes in HEVC
[23:01:06 CEST] <cone-952> ffmpeg 03Michael Niedermayer 07master:0cb46c1e0ece: Merge commit 'b2965ec7761c99e8ef9c217a28a517e37f010480'
[23:01:07 CEST] <cone-952> ffmpeg 03Michael Niedermayer 07master:20ba28af945a: tests/fate/hevc: Use a bitexact scaler in the hevc param change test to avoid failing on non x86
[23:46:19 CEST] <kierank> sooo ffmpeg is keeping ER
[23:46:25 CEST] <kierank> yet h264 is crash central
[23:53:56 CEST] <wm4> features are more important than stability
[23:55:35 CEST] <jamrial> not many players or users must be using slice threading if it's as unstable as you say it is
[23:55:40 CEST] <jamrial> otherwise trac would be flooded by now
[23:57:22 CEST] <kierank> erm
[23:57:25 CEST] <kierank> look at my reports
[23:57:28 CEST] <kierank> I have so many crashes
[23:57:44 CEST] <kierank> literally as soon as there's packet loss on the link, boom
[23:58:51 CEST] <kierank> and one other person I know who does the same tried it and said it was unusable
[23:58:55 CEST] <jamrial> yes, your reports, but nobody else's. what i mean is that it's an obviously not very used feature if nobody noticed it until now
[00:00:00 CEST] --- Sat Apr  4 2015