Jake Howard

Possibly a race in the IO, or something else behaving weirdly. I'll try and look this evening.

I can't reproduce the issues with `cargo test` (running that case specifically), but `cargo tarpaulin` does show an issue. I wonder if it's something to do with how `tarpaulin` runs,...

So how can we progress this PR? It seems silly to hold it off because the coverage tool has a bug. Should we remove the coverage checker, merge this in,...

Nice to see the tests passing! Looks like it's not working because the watcher thread can't get an exclusive lock on the config - I think something is holding onto...

Which ones? I've looked and all of them seem to call `.await`

That won't fix the issue. The problem isn't needing a separate Redis instance. The problem is [running `rq_worker` in a child process](https://github.com/LibrePhotos/librephotos-docker/blob/main/backend/entrypoint.sh#L26) and not monitoring it.

Ah, I'd missed those - those should probably be moved to process manager.

Merged in https://github.com/wagtail/wagtail/commit/5cc28acc563ff258d02a2cbea3de3bad87dfbfe6 - thanks!

This does indeed look like a bug. I'd removed it because in testing it hadn't made a difference. It seems in some cases, it does. `FileResponse` looks like it assumes...

Glad to see you're interested in fixing this! I think the main thing will be working out in what case this causes issues currently, and attempting to write a test...