[Ffmpeg-devel-irc] ffmpeg-devel.log.20161209

burek burek021 at gmail.com
Sat Dec 10 03:05:03 EET 2016


[01:15:59 CET] <cone-633> ffmpeg 03Alex Converse 07master:d3795926876b: aacdec: Allow SBR after DRC.
[01:16:00 CET] <cone-633> ffmpeg 03Alex Converse 07master:51a055b20656: aacdec: Rename elem_type_prev to che_prev_type.
[01:17:10 CET] <rcombs> sweet
[02:00:38 CET] <cone-633> ffmpeg 03Michael Niedermayer 07master:cff1c0edaa79: avcodec/ffv1enc: Fix size of first slice
[02:00:39 CET] <cone-633> ffmpeg 03Michael Niedermayer 07master:38a7834bbb24: avcodec/ffv1enc: Allocate smaller packet if the worst case size cannot be allocated
[14:42:11 CET] <nevcairiel> BtbN: http://fate.ffmpeg.org/report.cgi?time=20161209035641&slot=x86_32-msvc11-windows-native is broken from some cuvid things
[14:43:46 CET] <BtbN> msvc2012, hm. Is that with the converter thing?
[14:43:52 CET] <nevcairiel> yes
[14:44:01 CET] <nevcairiel> 2013 is the first to not need it
[14:44:57 CET] <BtbN> Because that error makes no sense.
[14:45:26 CET] <BtbN> https://github.com/FFmpeg/FFmpeg/blob/master/libavcodec/cuvid.c#L291  
[14:45:35 CET] <BtbN> first element of that struct is https://github.com/FFmpeg/FFmpeg/blob/master/libavcodec/cuvid.c#L71
[14:45:41 CET] <wm4> does anyone know why the heck libnpp support is non-free?
[14:45:51 CET] <BtbN> Because libnpp is not part of the nvidia driver
[14:45:53 CET] <nevcairiel> because the library is nonfree?
[14:45:58 CET] <wm4> but why
[14:46:06 CET] <BtbN> you have to ship the dynamic libraries yourself
[14:46:08 CET] <wm4> what reason could they possibly have?
[14:46:11 CET] <BtbN> so it's clearly not a system library
[14:46:17 CET] <wm4> oh, hm
[14:46:23 CET] <nevcairiel> they dont really care for cuda stuff to be redistributable
[14:46:28 CET] <nevcairiel> thats not their target audience
[14:46:52 CET] <BtbN> Most distributions have the libnpp libraries packaged
[14:46:57 CET] <wm4> what's their target audience?
[14:47:04 CET] <BtbN> but still, not a system library according to GPL
[14:47:09 CET] <nevcairiel> professional use
[14:47:22 CET] <BtbN> libnpp are performance primitives, for science and stuff with CUDA
[14:47:50 CET] <wm4> nevcairiel: so does it require some sort of commercial license?
[14:48:04 CET] <BtbN> no
[14:48:11 CET] <nevcairiel> technically cuda is under a proprietary license agreement
[14:48:16 CET] <nevcairiel> but it oesnt cost anything
[14:48:18 CET] <nevcairiel> +d
[14:48:18 CET] <BtbN> But it's not a system library, and thus makes ffmpeg non-free when using it
[14:49:01 CET] <wm4> dumb
[14:49:08 CET] <nevcairiel> why do you care about libnpp
[14:49:50 CET] <wm4> just came up
[14:50:03 CET] <Rathann> BtbN: which distros have it? probably only source-based ones like gentoo
[14:50:15 CET] <wm4> though if it were free we'd probably use vf_scale_npp for transcoding when scaling is needed
[14:50:20 CET] <BtbN> Rathann, Ubuntu and Gentoo are the ones I know of.
[14:50:41 CET] <Rathann> actually Debian too, but in non-free
[14:50:44 CET] <BtbN> http://packages.ubuntu.com/xenial/libnppc7.5
[14:51:00 CET] <wm4> ubuntu probably has it because debian does
[14:51:08 CET] <nevcairiel> dev headers in nvidia-cuda-dev, fwiw
[15:20:33 CET] <cone-307> ffmpeg 03James Almer 07master:77aa8005ece5: fate: improve fate-mov dependencies
[15:20:33 CET] <cone-307> ffmpeg 03James Almer 07master:06c62ce09c57: fate: add a monoscopic spherical matroska test
[15:45:03 CET] <BBB> omg that converter thingy
[15:45:17 CET] <BBB> of all things I ever did, thats probably the most evil of all
[15:45:38 CET] <BBB> (and everyone else that helped me, I didnt do it alone)
[15:45:44 CET] <BBB> but its still quite evil
[16:00:59 CET] <cone-307> ffmpeg 03Derek Buitenhuis 07master:4413e950b263: h264_slice: Wait for refs to be available before we use them in error concealment
[16:12:58 CET] <Zeranoe> So I set up a Cloudflare account with a rule to always cache the zips on their CDN. Initial results look promising.
[16:37:27 CET] <kierank> anyone familiar with mpeg-4 asp?
[17:45:52 CET] <cone-307> ffmpeg 03Michael Niedermayer 07master:ee2a6f5df8c6: avformat/oggdec: Skip streams in duration correction that did not had their duration set.
[17:45:53 CET] <cone-307> ffmpeg 03Michael Niedermayer 07master:8258e3638514: avcodec/mpeg4videodec: Fix undefined shifts in mpeg4_decode_sprite_trajectory()
[18:33:51 CET] <kierank> durandal_1707: how is pixlet going?
[18:35:34 CET] <durandal_1707> kierank: parsing completed,  need to figure how to combine layers and than clean up code
[18:35:47 CET] <kierank> combining layers should be addition
[18:35:57 CET] <kierank> after simple  transform
[18:36:06 CET] <kierank> like cfhd
[18:36:27 CET] <kierank> I will work on mpeg-4 studio profile I think
[18:37:06 CET] <durandal_1707> whats up with speedhq?
[18:42:09 CET] <kierank> durandal_1707: big gpl violations from xvid
[18:43:07 CET] <wm4> whaaat
[18:43:10 CET] <durandal_1707> it just some sharing :)
[18:46:16 CET] <kierank> wm4: the (i)dct is copied and pasted from xvid
[18:47:53 CET] <wm4> so _we_ violate it?
[18:49:20 CET] <kierank> no, they violate it
[18:49:27 CET] <kierank> they have copied the code *from* xvid
[18:49:53 CET] <wm4> who copied it where? sorry I think I'm a bit tired
[18:49:55 CET] <Shiz> xvid -code-> speedhq
[18:49:57 CET] <Shiz> wm4
[18:50:14 CET] <wm4> ah
[19:10:30 CET] <blb> kierank: that's been addressed
[19:10:58 CET] <kierank> eh?
[19:11:05 CET] <blb> the shq stuff
[19:11:31 CET] <kierank> deleting doesn't just mean the gpl is unviolated
[19:11:50 CET] <blb> i'm just saying it's been addressed
[19:12:15 CET] <kierank> i'm sure they think that
[19:12:25 CET] <blb> i know it
[19:12:34 CET] <nevcairiel> you cant fix the past no matter what you do, can only fix the now
[19:12:51 CET] <kierank> they've not acknowledged the gpl violation
[19:13:02 CET] <kierank> just gave a political response that they should have ignored it in the first place
[19:13:12 CET] <kierank> but they'll delete the code as a gesture of godwill
[19:13:19 CET] <blb> ?
[19:14:26 CET] <kierank> you seem to know what happened
[19:14:26 CET] <wm4> AFAIK even in ffmpeg we have cased all further harassment if the GPL violation was accidentally and then fixed as it was pointed out
[19:14:30 CET] <wm4> but IANAL
[19:14:37 CET] <kierank> it wasn't an accident
[19:14:39 CET] <blb> yes. i'm @ newtek :P
[19:14:46 CET] <kierank> well I have the emails
[19:14:51 CET] <kierank> and that's not what was said
[19:15:01 CET] <kierank> 6:12 PM <"kierank> they've not acknowledged the gpl violation
[19:15:01 CET] <kierank> 6:13 PM <"kierank> just gave a political response that they should have ignored it in the first place
[19:15:01 CET] <kierank> 6:13 PM <"kierank> but they'll delete the code as a gesture of godwill
[19:15:45 CET] <kierank> I have had enough of broadcast companies like newtek shitting on open source and then using it either legitimately or otherwise
[19:15:48 CET] <blb> and i'm saying, it's been addressed
[19:16:12 CET] <kierank> we can all play with euphemisms
[19:16:26 CET] <blb> i don't shit on open source :[
[19:16:29 CET] <kierank> your ceo does
[19:16:41 CET] <kierank> I have caught newtek with two gpl violations now
[19:17:04 CET] <blb> those were accidental, heck, remnants of old shit, unused code paths.
[19:17:11 CET] <Compn> blb : if you work at newtek, you probably shouldnt be talking about legal matters
[19:17:15 CET] <Compn> blb : could be used against you
[19:17:17 CET] <wm4> how does GPL code get into unused code paths
[19:17:18 CET] <Compn> or your company
[19:17:29 CET] <wm4> if the code was originally GPL, the GPL is still in effect, even if all code has been replaced
[19:17:34 CET] <blb> wm4: no idea
[19:17:57 CET] <blb> Compn: hi :P
[19:19:25 CET] <kierank> wm4:  https://usercontent.irccloud-cdn.com/file/vRUP4AYZ/
[19:20:02 CET] <wm4> templates...
[19:20:36 CET] <durandal_1707> you can't fight, they always can use code and obfuscate it
[19:20:57 CET] <kierank> yes, intel did that with satd from x264
[19:21:09 CET] <kierank> the guy literally invented a new method of satd for his masters
[19:21:24 CET] <kierank> and intel used intrinsics for exactly the same lines of code 
[19:23:24 CET] <wm4> neat, did they ever get sued?
[19:23:39 CET] <kierank> I don't know
[19:24:05 CET] <kierank> wm4: the email thread between steinar and newtek is magnificent
[19:24:51 CET] <kierank> they try to deny the violation and then steinar proves without a doubt the code is from xvid
[19:24:58 CET] <kierank> and then they say they will delete it as a gesture of goodwil
[19:25:16 CET] <wm4> lol
[21:02:51 CET] <kierank> anyone got a copy of ISO/IEC 14496-2?
[21:02:54 CET] <kierank> with studio profiles
[21:10:31 CET] <BtbN> michaelni, could the ffv1 size check fix be backported to at least 3.2?
[21:39:39 CET] <michaelni> BtbN, ok
[21:40:10 CET] <BtbN> nice, thanks.
[23:04:52 CET] <kierank> any idea why/how mpeg4video can use unchecked bitstream reader?
[23:14:00 CET] <kierank> durandal_1707: https://github.com/kierank/ffmpeg-sstp
[23:17:37 CET] <durandal_1707> kierank: michaelni should know, padding of some sort perhaps
[23:17:50 CET] <kierank> seems dodgy
[23:19:30 CET] <durandal_1707> get bits left can't be used?
[23:20:06 CET] <kierank> dunno, but the studio profile stuff breaks a lot of things
[23:20:13 CET] <kierank> it's like data inside data
[23:20:44 CET] <JEEB> o_O
[23:20:56 CET] <JEEB> I wonder which version of MPEG-4 Part 2 I have around
[23:21:36 CET] <atomnuker> kierank: define UNCHECKED_BITSTREAM_READER before including get_bits.h?
[23:21:52 CET] <kierank> JEEB: I only have a draft of the studio profile
[23:21:57 CET] <kierank> so the final version would be useful
[23:22:10 CET] <JEEB> ok, sorry. don't seem to have anything better on hand :/
[23:48:46 CET] <kierank> dammit nothing matches
[23:48:50 CET] <kierank> I really need that amendment
[23:49:08 CET] <kierank> atomnuker: yes, why is it allowed
[23:51:48 CET] <atomnuker> IIRC some decoders will never overread (like some wma one) so it's safe for them to enable this to get more speed
[23:52:36 CET] <kierank> yes but I'm adding stuff so I don't know if that's guaranteed any more
[23:56:26 CET] <atomnuker> oh it's already enabled?
[23:58:58 CET] <kierank> yes
[00:00:00 CET] --- Sat Dec 10 2016


More information about the Ffmpeg-devel-irc mailing list