07:31 <fflogger> [editedticket] nafmo: Ticket #10732 ([avcodec] avcodec_flush_buffers() not resetting E-AC-3 decoder) updated https://trac.ffmpeg.org/ticket/10732#comment:15

09:40 <fflogger> [newticket] guest: Ticket #11662 ([undetermined] ffplay does not detect when the remote side closes the connection) created https://trac.ffmpeg.org/ticket/11662

10:36 <cone-395> ffmpeg Desmond Liu master:060fc4e3a5ac: doc/scaler.texi: add missing gamma option

11:37 <TheVibeCoder> see the channelmap filter in Librempeg

12:14 <kierank> TheVibeCoder: do not discuss competing project

12:14 <kierank> leadership will come for you

13:41 <fflogger> [editedticket] Balling: Ticket #10732 ([avcodec] avcodec_flush_buffers() not resetting E-AC-3 decoder) updated https://trac.ffmpeg.org/ticket/10732#comment:16

13:44 <TheVibeCoder> see the channelmap filter in Librempeg

14:49 <kasper93> have you ever considered sending your patches upstream? You seem to be very talented coder.

14:55 <TheVibeCoder> Librempeg is upstream

14:55 <TheVibeCoder> FFmpeg is downstream

15:04 <llyyr> why would he help competition

15:07 <kasper93> and the competition is about what?

15:12 <Compn> not competition just another fork

15:13 <Compn> in a long line of forks

15:19 <TheVibeCoder> the best fork

15:23 <ePirat> mkver, fix for copy_tls_opts_dict checking on the list.

15:23 <ePirat> mkver, there are a ton more instances of unchecked av_dict_set but checking all of them and erroring properly probably requires quite a bit of refactor of this code

15:24 <ePirat> the failure cases are already right now so hard to follow

15:30 <kierank> TheVibeCoder: who owns librempeg.org

15:31 <mkver> ePirat: I only complained about these one because they seem important security-wise.

15:31 <ePirat> to be fair we are in a bad place security wise anyway :D

15:31 <ePirat> openssl doesnt even check the hostname

15:32 <ePirat> mkver, I wonder if it would make sense to have an error state in AVDictionary to make these checks easier, similar to what AVBprint does, given we have a ton of unchecked ones…

15:36 <ePirat> also why is -tls_verify 1 not the default…

15:36 <ePirat> seems illogical that you pass a https url to ffmpeg and it doesnt verify anything

15:36 <kierank> lol

15:37 <ePirat> (and even if you set tls_verify, if you use openssl, it doesnt check the hostname)

15:39 <kasper93> yeah, there is no reason to not check things in present year

15:39 <kasper93> though I'm sure people will complain if you change that

15:48 <ePirat> probably…

15:51 <TheVibeCoder> jamrial_: fixed ac3 bug/artifact

15:51 <ePirat> wbs, do you think it makes sense to keep the hlsproto around?

15:56 <jamrial_> TheVibeCoder: something i introduced?

15:57 <TheVibeCoder> jamrial_: https://trac.ffmpeg.org/ticket/11578

17:50 <fflogger> [newticket] DaveFlater: Ticket #11663 ([avfilter] lv2 omitted values printed as NaN in help) created https://trac.ffmpeg.org/ticket/11663

18:37 <ePirat> pross, in adpcm_sanyo_expand3 sign is uninitialized if none of the cases match

18:38 <ePirat> add I meant, not sign…

18:43 <TheVibeCoder> jamrial_: not interested in fixing bugs?

18:43 <jamrial_> TheVibeCoder: yes, i sent a patch for another ticket in fact :p

18:54 <wbs> ePirat: no, probably not. originally, it was useful for comparing how things work in some cases (there were some odd cases that worked better with hlsproto), but with 15 years of development on top of the demuxer, it's clear that the hlsproto approach was wrong, so let's bury it

18:57 <jamrial_> TheVibeCoder: i can't even hear the click the ticket talks about...

18:57 <TheVibeCoder> first play

18:57 <TheVibeCoder> make sure seeking is not ever used

20:28 <indecisive_turtl> Lynne: I'm not sure why, but there is some strange behavior when converting int16_t to int32_t. On the shader for estimating slice sizes, it reads a buffer of int16_t coefficients. If those values are negative, it will work fine as long as the type is kept int16_t, the moment it's promoted to int32_t it becomes unsigned

20:28 <indecisive_turtl> Have you experienced this before?

20:41 <Lynne> maybe averne would know?

20:41 <indecisive_turtl> Only way to make it work correctly seems to be inserting a bitfieldExtract(coeff, 0, 16) to make it sign extend hmm

20:42 <indecisive_turtl> Though could also be renderdoc debugger acting up, cause I feel that would break a lot of things if it was a real bug

21:20 <pross> ePirat: thanks

21:21 <ePirat> pross, sorry, had not noticed your patch before I pinged you…

21:24 <pross> one of my tests is failing on softworkz's x86 mac build machines, but fate no longer prints the contents of the .err file. how would i go about fixing this? does FFmpeg have any x86 mac machines i can test on?

21:58 <JEEB> &31

22:32 <ePirat> pross, maybe try emailing softworkz

22:32 <ePirat> maybe he can add printing the err file or something