multistream/negotiated: Propagate poll_close on unreceived protocol messages #62
Conversation
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
Signed-off-by: Alexandru Vasile <alexandru.vasile@parity.io>
| // However, for the Litep2p the negotation must conclude before closing the | ||
| // lazy negotation of protocol. We'll wait for the close until the | ||
| // server has produced a message, in this test that means forever. |
There was a problem hiding this comment.
I'm curious why it was implemented like this in the first place. Aren't we in a Chesterton's fence situation here?
There was a problem hiding this comment.
Yep indeed, I wonder if this affects something in substrate. Proly would be good to have it running in versi for a while. I'll wait a bit for that and group this PR with the clippy PR (and any others) for testing
There was a problem hiding this comment.
cc @altonen would love to get your feedback on this if you remember why the change was needed 🙏
There was a problem hiding this comment.
I honestly don't even remember touching the multistream-select code. Did you check the behavior in this test against the latest multistream-select that libp2p uses or some other version? I wonder if it could be related to this: paritytech/substrate#14691
This PR ensures that lazily negotiated protocols can be closed before the specific protocol message has been received.
Closes: #60
cc @paritytech/networking