Josh Holmer

This is cursed. It's still trying to pull clap_derive 4.5.0 even though we've told it not to. Weird.

I guess the renovate.yml file doesn't apply to the Cargo.lock dependencies... man, what a headache. I may just manually bump these for now, and think about looking for a better...

1. There currently isn't 2. Tiles are a tradeoff between speed (both encode and decode speed) and quality. The general recommendation is to use as few tiles as you can...

Oh, huh... at one point I'm pretty sure it showed the values, I guess that broke at some point... I opened #3146 to track this. Regarding the pixel format, rav1e...

On a fully stripped binary on x64 linux, before was 3.8mb and after was 5.1mb

AWCY: https://beta.arewecompressedyet.com/?job=master-89787077db93ffe0b0a0e413479e2746b927dc96&job=estimate-entropy%402023-01-19T04%3A22%3A07.860Z

@maj160 what improvements did you see in your benchmarking? Mine is only showing 1-2% speedup for this change.

> And the AWCY link shared above suggests there is a hit in performance across all the resolution in Obj1-Fast and there is no speedups, is that latest and updated...

> Just to be clear, this will give us approx 0.5-1.8% bdrate loss with approx 1-2% speedup? Yes. As you mentioned, because of that I'd be hesitant about blanket applying...

This is going to be rather painful to debug given that it only seems to be failing on x64 Macos. Why that is, I have no clue... I'll fix it...