Joshua Oreman

> I think we had a miscommunication :-). When I say "bidirectional" I mean "full duplex" – handle A can send/receive to handle B, and handle B can send/receive to...

@goodboy - interesting ideas! It sounds like you're not proposing anything that introspects "what task am I running from?" per se, but rather suggesting a different way of tracking whether...

Other thoughts on the questions in the initial post: > In open_memory_channel, should we make max_buffer_size=0 default? Is this really a good choice across a broad range of settings? The...

I left a comment on #1457: I think we shouldn't drop the schedule point in either branch, only the cancellation point; and we shouldn't drop the cancellation point if the...

I really like the ergonomics-per-implementation-complexity-unit of `@producer`. I think the version as listed won't quite work -- the `async with send_channel:` needs to be done inside the nursery task, else...

> The alternative would be to take the exception that send raised, and throw it into the agen, so the yield acts exactly like a call to send. I believe...

We would probably also want to cancel the nursery inside `wrapper()` when the context gets exited. Closing the receive channel will deal with cleanup when `wrapper()` is suspended inside `send_channel.send()`,...

Tasks have had names for a while, and Cancelled exceptions don't know what scope will catch them anymore, so I'm not sure there's anything left to be done here.

> I don't actually know why @oremanj is so eager to get at partial results in this case; I gather he has some use case where he needs this feature,...

`subprocess.CompletedProcess` can't be constructed without at least some info: ``` >>> subprocess.CompletedProcess() Traceback (most recent call last): File "", line 1, in TypeError: __init__() missing 2 required positional arguments: 'args'...