08:15 <elenril> >call av_buffersrc_close() on the only input

08:15 <elenril> >av_buffersink_get_frame_flags() returns EAGAIN

08:15 <elenril> quality code

09:07 <durandal_1707> elenril: where?

09:12 <elenril> in my threading branch

16:17 <courmisch> elenril: it seems your want of RVQ hasn't been communicated well

16:18 <courmisch> elenril: new designs pitch half-precision or bastards thereof

18:10 <jkqxz> Does anyone ever write a value of 2^32-1 for uvlc in AV1?

18:11 <jkqxz> The standard is clear that >=32 zeroes followed by a one means 2^32-1.

18:12 <jkqxz> libaom thinks that 32 zeroes means 2^32-1, and does not look for a one at all in that case.

18:13 <Lynne> libaom is wrong, then

18:13 <jkqxz> dav1d is the same as libaom.

18:16 <jkqxz> SVT is a copy of the libaom code.

18:17 <jkqxz> gav1 errors out if it sees 32 zeroes.

18:24 <durandal_1707> michaelni: your narrative to bad mouth mine and kurosu work is going to fail, go work on your SDR hacks

18:40 <BBB> now now now

19:15 <Traneptora> jkqxz: sounds like you should submit a bug report to libaom and dav1d

19:16 <Traneptora> since the reference implementation being noncompliant on a valid stream is ????

19:18 <cone-018> ffmpeg Leo Izen master:86ed68420d3b: avcodec/librsvgdec: fix memory leaks and deprecated functions

19:25 <michaelni> durandal_1707, i have no problem with your work. The problem is that the code was pushed to git master in this condition. There are many developers who try to cleanup the codebase. Newly added code should not require extensive cleanup

19:28 <Traneptora> durandal_1707: tbf if "Fix is trivial" unclear why you can't just implement it in the original patch, so we never have a broken master?

19:32 <Lynne> jkqxz: what about svt-av1's?

20:20 <cone-018> ffmpeg Andreas Rheinhardt master:315c956cbd14: avcodec/pthread_frame: Remove ff_thread_release_buffer()

20:20 <cone-018> ffmpeg Andreas Rheinhardt master:6e4030a07b79: avcodec/av1dec, vaapi_av1: Remove excessive logmessages

21:27 <Lynne> kierank: do you have any comments about improving avtransport more? I still don't know how to specify much about the buffer model of mpeg/av1 in

23:09 <BBB> jkqxz: I believe the spec takes precedence here

23:09 <BBB> jkqxz: so while I agree that this is weird, maybe we should clarify this in the codecwg

23:09 <BBB> rather than outright disallowing it

23:09 <BBB> and then we can take the correct approach after the fact

23:09 <BBB> wdyt?