Would you accept a documentation patch describing Hyper's cancellation behavior?
I think Hyper is largely cancel safe:
- dropping an HTTP/1 request causes the connection to be closed
- dropping an HTTP/2 request leaves the shared connection in a good state and cancels the unwanted response (using RST_STREAM).
- otherwise Hyper doesn't really have a bunch of hidden state to screw up
- well, except the legacy Client support, but actually that's also carefully implemented to do the right thing on cancellation
I don't know about the server side, but that's OK, I can read code. If you're interested.
It doesn't seem like guide material. I was thinking a single # Cancel safety section at the crate level.
Would you accept a documentation patch describing Hyper's cancellation behavior?
I think Hyper is largely cancel safe:
I don't know about the server side, but that's OK, I can read code. If you're interested.
It doesn't seem like guide material. I was thinking a single
# Cancel safetysection at the crate level.