Re: [quicwg/base-drafts] Fixing max_ack_delay (#3938)

[Kazuho Oku <[e-mail]>]

@kazuho commented on this pull request.



> @@ -1129,6 +1129,10 @@ follows:
    smoothed_rtt = kInitialRtt
    rttvar = kInitialRtt / 2
    min_rtt = 0
+   // max_ack_delay is initialized to 0 and
+   // updated when transport parameters are exchanged.
+   // If a peer does not specify max_ack_delay,
+   // it is set to 25ms.

I'd argue that the spec is clear about the intent, and the current state of this PR is going to introduce a contradiction from what the spec says.

Quoting from [section 4.1](https://quicwg.org/base-drafts/draft-ietf-quic-recovery.html#section-4.1-1), "congestion control and round-trip time (RTT) measurement are unified across packet number spaces." As can be seen, persistent congestion by definition is a path-wide event. Not an event specific to the PN space in which a loss has been newly observed.

Based on that, my understanding is that consistently applying max_ack_delay exchanged by the Transport Parameter is the intent of [the formula found in in section 7.8](https://quicwg.org/base-drafts/draft-ietf-quic-recovery.html#section-7.8-3), and that the definition that currently exists in the WG draft is correct.

-- 
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/pull/3938#discussion_r458788016