Re: [quicwg/base-drafts] Fix: a connection is not identified by a *collection* (#1632)

[Mike Bishop <[e-mail]>]

I think we need to assume that CIDs are unique without regard to the client address, because otherwise you won't survive NAT rebinding.  The same will be true for server addresses if we allow server migration in a future version, so I'd like to lay the conceptual groundwork here.

I agree that a server could hypothetically reuse CIDs across its own addresses, but to me that seems more conceptually akin to multiple independent servers running on different ports.  I'm inclined to leave the uniqueness comments in, as @dtikhonov has done here.

I originally said "unlikely to be" unique, since IIRC we have at least one implementation that used zero-length addresses until there was a tuple overlap.  Zero-length can be unique if it's the only one.  :-)

-- 
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/1632#issuecomment-411133382