Jan Procházka
Jan Procházka
which is proving to be a mistake, because now I'll have to wait for CI to report all the _sequential_ failures one by one...
don't merge this yet, i left some ~~temporary~~ code in tests so i could get more failures in parallel (it was supposed to be temporary, but i think it's fine...
~~It seems like what you pushed in [`b914586`](https://github.com/rerun-io/rerun/pull/9787/commits/b9145869509bc9e7b3696952df9236be8ee6fd2e) didn't fix the failures :thinking:~~ (my fault, forgot to invert the condition)
We didn't mention performance at all in docs, I don't know if it makes sense at this point?
Current plan is to use https://github.com/rerun-io/rav1d-rs (a fork of `dav1d-rs` which replaces the underlying `dav1d` library with `rav1d`) to decode AV1 video. We're uncertain about how to support other codecs,...
(Note that I made the same mistake as in `re_mp4` and uploaded a bunch of binary files to `rav1d-rs`, which need cleaning up via history rewriting and either be turned...
Even after https://github.com/rerun-io/rerun/pull/7860 we still copy video data for each sample. But `Blob` is a `Arc` internally, so we should be able to avoid that as well. The decoder could...
Follow-up: https://github.com/rerun-io/rerun/issues/7878
There is not, unfortunately. I don't know what the convention for that is, but it feels wrong to read `serde`'s attributes for this. Implementing our own `rename` attributes would probably...