Re: [quicwg/base-drafts] Largest acked in ACK frame MUST NOT decrease (#2205)

[ianswett <[e-mail]>]

There are a few things this precludes, and we we should understand those before doing this.

1) A very limited implementation(ie: hardware) can't always send only one ACK block without giving up on reordered packets.  I don't have such an implementation, but at some point I remember discussing this idea with people who seemed interested in it.
2) One cannot do multipath with QUIC v1 and it means the multipath design will have to use a PN space per path, as described in Christian's draft: https://tools.ietf.org/html/draft-huitema-quic-mpath-req-01#section-4.11  I think that's likely the right choice for other reasons, but I'm hesitant to make that decision now.
3) An RTT sample can only be taken based on the largest acked.  Therefore, if there is a large amount of reordering(ie: 1RTT), you don't get any RTT samples for an RTT.  If we care about this, we can probably solve it a different way in the recovery draft, but that likely means using ack delay sometimes and not others.

In regards to Martin's comment, I thought we already had text that said ACKs with lower packet numbers than the largest ACK packet number should be ignored, since they're old?

-- 
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/2205#issuecomment-448612503