Re: [quicwg/base-drafts] Improve ACK_ECN frame encoding (e.g., use bit-vector) (#1439)

[ianswett <[e-mail]>]

I'm going to reiterate: I don't think duplicate detection is a hard problem.  You have an IntervalSet that is used to generate an ack.  That goes back some distance in packet number space.  At some point, you drop older/smaller blocks off that data structure.  When you do, I'm suggesting you stop trying to decrypt packets with packet numbers than whatever packet number you've decided you're no longer waiting for(whether due to acks of acks or some other algorithm).  There might be some cases when this InteralSet needs to be a bit larger than what you serialize to an ack frame, but that's not too difficult to deal with.

The truncated packet number encoding is designed with increasing packet numbers in mind, so the only thing we're trying to allow is accepting packets within some reordering threshold.  Certainly a receiver can decide to have a longer reordering threshold than an RTT, but at some point not too much longer than an RTT, it's reasonable to assume the peer retransmitted the data anyway.

-- 
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/1439#issuecomment-397310140