How does the returned python object look when None is returned? Perhaps we should add that to the docs?

I thought that Closed channel would result in the TryRecvError::Closed error.

Can it be None?

I thought that Closed channel would result in the TryRecvError::Closed error.

Can it be None?

Ah sorry, I was confused. There is no None, but there is TryRecvError::Empty instead. It looks like this currently also mapped to a Python exception with a string message. Is there a good way to differentiate an empty from a closed channel this way? This would be useful imo because you might want to handle a closed channel differently than an empty one (e.g. break vs wait for a while).

I thought that Closed channel would result in the TryRecvError::Closed error.
Can it be None?

Ah sorry, I was confused. There is no None, but there is TryRecvError::Empty instead. It looks like this currently also mapped to a Python exception with a string message. Is there a good way to differentiate an empty from a closed channel this way? This would be useful imo because you might want to handle a closed channel differently than an empty one (e.g. break vs wait for a while).

Yeah, so for now it only going to be text based but in the future we could add custom exception

Ok, I guess you could work around this by checking the new is_empty before calling try_recv.

Maybe it would still be better to return an Option instead, which would map to Python's Optional. So basically return None on empty and throw an exception only if finished. This would be more intuitive in my opinion. It would also match the drain method more closely, which also returns an empty value if no events are ready, instead of throwing an exception.

I don't feel strongly about this though, so feel free to merge this as-is if you prefer.

Ok, I guess you could work around this by checking the new is_empty before calling try_recv.

Maybe it would still be better to return an Option instead, which would map to Python's Optional. So basically return None on empty and throw an exception only if finished. This would be more intuitive in my opinion. It would also match the drain method more closely, which also returns an empty value if no events are ready, instead of throwing an exception.

I don't feel strongly about this though, so feel free to merge this as-is if you prefer.

Sounds good, to me. I think it's a bit ugly to have

try:
  node.try_recv()

which would make the second try tautological.