Johannes Weiss

Thank you @3a4oT , I filed https://bugs.swift.org/browse/SR-12008 for this.

Wow awesome! Thanks @3a4oT

@0xTim this is blocked on this: https://bugs.swift.org/browse/SR-12008 I believe.

CC @tomerd

ah of course, we have to consume everything from the kernel here and currently we deliver it through the pipeline even though we have `.autoRead = false`... Maybe we should...

> Buffering in the `Channel` is really tough, I think, because it forces the `Channel` to have a lifecycle past the FD closure. Right now NIO is super aggressive at...

> Yes, I agree, with the current model `autoRead=false` is not very useful, but it's worse than that, because it means that `read` is of limited utility as well. I...

> @weissi Your proposed design for dealing with this is fairly reasonable. It does interact poorly with our `registered`/`unregistered` events, but frankly those are already of extremely limited utility and...

> Yeah, the issue is that while this is a very nice story about how `unregistered` works, `unregistered` is right now strictly ordered with `channelInactive`. This would break that strict...

ouch, thanks @allright , we'll look into that