[Ffmpeg-devel-irc] ffmpeg-devel.log.20190822

[burek]

[07:55:34 CEST] <thardin> test
[08:10:26 CEST] <BradleyS> thardin: beep boop
[08:10:50 CEST] <thardin> had some auth troubles
[08:11:01 CEST] <thardin> this channel doesn't require registration I noticed
[09:29:35 CEST] <cone-692> ffmpeg 03Xuewei Meng 07master:c87237d10511: libavfilter: Update derain filter doc.
[10:40:59 CEST] <durandal_1707> hello leechers! whats up!
[10:49:11 CEST] <thardin> did you know: if I run ffmpeg on my 486 then decoding takes a long time
[10:49:20 CEST] <thardin> this is a serious DoS issue which must be addressed post-haste
[10:51:13 CEST] <durandal_1707> ffmpeg obviously sucks, use something else, please
[11:23:18 CEST] <cone-692> ffmpeg 03Peter Collingbourne 07master:9bcb1cb6ed50: Add assembly support for -fsanitize=hwaddress tagged globals.
[11:26:12 CEST] <kierank> "I can implement this if theres consensus that this should be done."
[11:26:17 CEST] <kierank> funny how the fuzz patches go in without consensus
[11:26:24 CEST] <kierank> but when they want to be reverted they need consensus
[11:47:52 CEST] <thardin> the people's patches
[12:01:12 CEST] <durandal_1707> the people's patches republic
[12:36:15 CEST] <durandal_1707> cehoyos: you have not added av1 to mplayer
[12:36:34 CEST] <cehoyos> You have not ported dav1d to FFmpeg;-)
[12:40:40 CEST] <durandal_1707> when you add bunch of missing decoders to mplayer
[12:50:04 CEST] <cehoyos> When I feel like it... (or when I need them)
[12:54:03 CEST] <durandal_1707> cehoyos: when you write vfm/dshow wrapper for ffmpeg?
[12:54:14 CEST] <cehoyos> Never
[12:54:51 CEST] <durandal_1707> do not make Compnn sad
[13:05:26 CEST] <cehoyos> I wouldn't even know where to start
[14:03:09 CEST] <durandal_1707> https://t.co/redI0lSUhW?amp=1 your best friends
[15:18:49 CEST] <JEEB> wbs: would it make sense to write subtitle AVPackets into a buffer in mov_write_packet, and then in mov_flush_fragment/mov_write_trailer call a function that goes through the list and generates a single AVPacket containing the whole mess and writes that in?
[15:19:07 CEST] <JEEB> if 14496-30 says you have to have a single sample for the duration of the whole document/fragment
[15:19:21 CEST] <JEEB> just trying to think of a """sane""" way of doing it right now
[15:19:39 CEST] <JEEB> we already have write_single_packet kind of function utilized for 3gpp text
[15:21:07 CEST] <JEEB> if you see the == AV_CODEC_ID_MOV_TEXT parts
[15:23:52 CEST] <wbs> JEEB: yes, that sounds sensible
[15:24:25 CEST] <JEEB> ok, cool
[15:24:44 CEST] <JEEB> (I am starting to have a headache of headaches for 14496-30)
[15:25:01 CEST] <JEEB> but at least it feels like I might be getting over it if all goes well :)
[16:20:22 CEST] <kierank> this EPG code is not good
[16:20:28 CEST] <kierank> epg has nothing to do with side data
[16:21:00 CEST] <kierank> or Frames
[16:21:19 CEST] <kierank> yet another thing being forced into the ffmpeg paradigm
[16:24:38 CEST] <JEEB> as a player listening for EPG side data packets from the demuxer kind of makes sense. esp. since you get updates for it. Not sure if making a new stream and pushing out Epg AVPackets would be any better? I haven't looked at the code and thus don't know how much of a PITA it would be utilize in its current implementation state
[16:24:56 CEST] <JEEB> I think the last implementation I saw of Epg parsing was only getting stuff like current program metadata
[16:25:26 CEST] <JEEB> since decoding the whole Epg is kind of "uhhh" if you're now adding that much work to the demuxer. not sure how VLC handles it.
[16:40:34 CEST] <snoring_cat> Does anyone know if there is a way to show a filters available arguments from the command line?  running ffmpeg -filters just shows filter names, not arguments.  Would like to do something like ffmpeg -filters crop.
[16:43:39 CEST] <jamrial> snoring_cat: ./ffmpeg -h filter=name
[16:43:58 CEST] <jamrial> also, this is a question for #ffmpeg
[16:46:24 CEST] <snoring_cat> jamrial, thanks.  That is what I was looking for.
[16:46:38 CEST] <snoring_cat> Initially I tried to ask the question on the #ffmpeg, but was getting a message saying "cannot send to nick/channel"
[18:16:34 CEST] <juandl> Anyone knows VP9, that could help me with something?
[18:19:26 CEST] <juandl> I'm not sure I understand the dequantization process for skipped blocks, in the analyzer I'm using skipped blocks show the base quant even though they have a segment id with alt q enabled
[18:37:56 CEST] <cone-676> ffmpeg 03Michael Niedermayer 07master:5937f0550304: avformat/vividas: Check av_xiphlacing() return value before use
[18:37:56 CEST] <cone-676> ffmpeg 03Michael Niedermayer 07master:ab56e62e8f7e: avcodec/vp5/6/8: use vpX_rac_is_end()
[18:37:56 CEST] <cone-676> ffmpeg 03Michael Niedermayer 07master:493438fafc5c: avformat/realtextdec: free queue on error
[18:40:55 CEST] <thilo> durandal_1707: So do you want a dev meeting or not?
[18:48:58 CEST] <kurosu> he wanted an irc meeting, and I suspect he would never show in person anywhere
[18:49:02 CEST] <kurosu> so...
[18:49:15 CEST] <kurosu> *he most probably wanted
[18:53:28 CEST] <thilo> kurosu: he didn't mention IRC. For me, I'd prefer a personal meeting like others proposed for good reasons.
[18:54:46 CEST] <kurosu> I'm not disputing what's best
[18:55:24 CEST] <kurosu> only that, even if he still wanted it, he's unlikely to attend a F2F
[18:59:17 CEST] <thilo> kurosu: sure, I assume the same. Yet, I want to know
[21:47:55 CEST] <cone-201> ffmpeg 03Marton Balint 07master:6ee40dcb64c9: avformat/mxfdec: do not ignore bad size errors
[21:47:55 CEST] <cone-201> ffmpeg 03Marton Balint 07master:b2e37e3eb27b: avutil/imgutils: remove dead assignment
[21:47:55 CEST] <cone-201> ffmpeg 03Marton Balint 07master:3a09dbdb4a2f: avcodec: remove some dead assignments
[21:48:26 CEST] <thardin> heck yeah mxf fixes
[21:52:19 CEST] <thardin> hmm I think I can speed the cinepak encoder up by first computing the largest possible codebook then deriving the smaller ones from it
[22:03:33 CEST] <thardin> .. and not recomputing the v4 codebook every time
[22:28:07 CEST] <thardin> which my original code did, but rl undid that for some reason
[22:41:46 CEST] <cone-201> ffmpeg 03Marton Balint 07release/4.2:370c346d5dfd: avformat/mxfdec: do not ignore bad size errors
[23:00:21 CEST] <thardin> max_pixels seems like a decent solution
[23:22:12 CEST] <thardin> it's a specific limit instead of this wishy-washy percentage nonsense
[23:31:52 CEST] <cehoyos> kierank: http://mogurenko.com/2019/08/21/ffmpeg-gpl He appears to claim now that you did not ask him? Or do I misunderstand?
[23:33:15 CEST] <kierank> cehoyos: http://mogurenko.com/2019/05/17/prores-quality/#comment-9
[23:33:20 CEST] <kierank> that's what I did
[23:33:24 CEST] <kierank> I did nothing on github
[23:34:13 CEST] <kierank> "as its basically skeleton copied from prores_anatoly with whole logic replaced by calling functions from private static library"
[23:34:20 CEST] <kierank> cehoyos: he doesn't understand he's still in violation
[23:34:23 CEST] <cehoyos> Yes, I saw that and I am surprised now that he claims that it would have been nice if he were asked. Not sure if I understand...
[23:34:49 CEST] <kierank> he thinks he is compliant with that post
[23:34:54 CEST] <kierank> even though his library is private
[23:34:59 CEST] <thardin> he could provide the library as a .a file and be within the bounds of the lgpl at least
[23:34:59 CEST] <kierank> he just doesn't understand at all
[23:35:30 CEST] <kierank> he shipped gpl
[23:35:42 CEST] <thardin> yeah he done goofed
[23:36:58 CEST] <thardin> getting a better prores encoder into the codebase would be good tho
[23:38:25 CEST] <Compnn> just stop antagonizing the guy and ask him to send updated patch :D
[23:38:41 CEST] <thardin> he doesn't seem opposed to that. just not super into the review process
[23:38:54 CEST] <cehoyos> Compnn: Isn't this what Kieran did?
[23:40:12 CEST] <Compnn> oh nevermind i see he wants to make xprores encoder
[23:40:20 CEST] <Compnn> and charge money for it
[23:41:09 CEST] <Compnn> and c++
[23:42:22 CEST] <durandal_1707> some body please keep that ffmpeg ,  i wanna re it
[23:43:38 CEST] <thardin> Compnn: he could stick the agpl on it and still make a ton of money
[23:43:41 CEST] <thilo> durandal_1707: AFAICT you were offline earlier. Wanted to know if you want a dev meeting or not?
[23:44:12 CEST] <thardin> and be within the license' good graces does to the forgiveness clause
[23:44:18 CEST] <thardin> due*
[23:47:22 CEST] <durandal_1707> thilo: i wanted remote meeting at least, but got no blessings
[23:49:05 CEST] <thilo> durandal_1707: I read the ML thread like there is a desire to do it
[23:49:13 CEST] <thilo> durandal_1707: would you attend in person somewhere instead of remote?
[23:49:50 CEST] <durandal_1707> i cant attend in person anywhere random
[23:50:30 CEST] <thilo> any place you could attend?
[23:51:35 CEST] <durandal_1707> only my place, but there are not any oss meetings there. only microsoft
[23:55:08 CEST] <thilo> don't tell me you're on windows... how do I read that, you just can't / don't travel or don't you want to meet others in person?
[23:57:08 CEST] <durandal_1707> i do not travel, too expensive and dangerous
[23:57:48 CEST] <thardin> what continent are you located in?
[23:58:01 CEST] <durandal_1707> greenland
[23:58:06 CEST] <thardin> ah
[23:59:54 CEST] <durandal_1707> im in europe most devastated part
[00:00:00 CEST] --- Fri Aug 23 2019