Andrew Godwin

OK, I'm very confused - your traceback's line numbers do, indeed, line up with a v3.8.1 release of Python, but the code at the end of your traceback is protected...

Yeah, we just need to work out what is causing it. Apache may well be the cause - not sure - but either way I don't think this is a...

Also, how are you deploying on apache - mod_wsgi?

OK, my suspicion is that `mod_wsgi` claims to a Python process that it is running in the main thread when, in fact, it is not (due to the main Apache...

Yeah, I suspect this is a bug in mod_wsgi somehow then. I'll have to look into it more when I get time.

Could you flesh this out a bit more with how you detected this? That would be valuable for whoever picks this up, so they can verify a fix.

Hm, asgiref does support Python 3.5 for now, so we need to do something, or do a major version bump. I think acceptable behaviour would be to throw a RuntimeError...

I don't mind too much; I generally use RuntimeError for things a bit more serious/unsolvable, but this could go either way. We can't mirror the underlying error, anyway, since it's...

@devxpy That's needed to signal to `asyncio.iscoroutinefunction` that it should return True. We'll need to keep that for anything that outputs an async-like function.

No, no, just a quick one-off test to make sure it's not massively worse. `timeit` or something will do and then paste the results here.