Andrew Godwin

To be precise - when you call sync code, it uses a fixed-size threadpool, and when you then call async code, it blocks that thread while the async loop does...

`async_to_sync` doesn't make a threadpool - `sync_to_async` does, so none of those calls are using a threadpool (`async_to_sync` will instead try to find an event loop to run it on)....

If you're using `SyncConsumer`, which I presume you are by the code, then `websocket_disconnect` is being run inside `sync_to_async` for you already. This means that when you run something inside...

It's more timing I'm interested in. If you add more threads, is it longer until the error happens? If you reduce them, does it happen faster?

Hmm OK, I'll try to take a look in the next week or two. Thanks for investigating the threads setting - that rules out it being a thread deadlock.

Finally getting back to this (I must have let it slip through last year, sorry). You're totally right about this - we should see if there's a way to either...

Unfortunately I don't think this is a bug in Channels, and so we can't help you on here - you're better off going to one of the mailing lists/stackoverflow or...

No worries - take the time you need to get everything working! I might suggest you mark the PR as a draft, though, so that way you can move it...

I'm not really feeling like switching typechecker right at this moment, unfortunately. Sorry for delaying this yet further!

Will current versions of mypy-in-the-wild choke on the new stuff, btw, or will they ignore it and not typecheck it? Considering the type annotations here are mostly for the benefit...