[quicwg/base-drafts] Should we allow ACK-only packets to be declared lost? (#3451)

[Martin Thomson <[e-mail]>]

After some discussion, the "in-flight" condition here leads to an odd requirement.

> A packet is declared lost if it meets all the following conditions:
> * The packet is unacknowledged, in-flight, and was sent prior to an acknowledged packet.
> * [other stuff]

This says that lost ACK-only packets cannot be declared lost.  This buries a potential signal that might drive a congestion controller.

This comes from TCP.  In TCP, detecting loss of an ACK is impossible, but in QUIC, we can.  But this requirement essentially forbids that.  Implementations don't seem to be consistent on this point.

@ianswett says that if we change this, we should also mandate the acknowledgment of ACK-only packets when sending an ACK for other ACK-eliciting packets.

We could simply say that this is good and recognize that this requirement is consistent with not requiring ACKs for these.  Or we could take the signal and declare them lost.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/3451