Re: [quicwg/base-drafts] Invariants extensbility (#1260)

[Mike Bishop <[e-mail]>]

It seems a contradiction in terms to says that we would add something in a future version of the invariants.  These are supposed to be things which are true of *all* QUIC versions, and something we add in the future would, by definition, not be true of QUIC versions which precede it.  That makes it difficult for version-agnostic network components to observe it.

I think the first line of defense is to simply define them as version-specific, with the intention of including them in all future QUIC versions.  This does require version awareness on the part of the observer.  A more extended version might be a new document defining additional invariants which are true for all versions conforming to a particular version pattern (first byte of 0x01, for example).  This requires some version-awareness, but wouldn't require the observer to understand *particular* versions.

But most fundamentally is the answer I suggested and no one argued with during the call for adoption:  If the invariants differ, you're defining a new protocol which is not QUIC.  Describe *that* protocol's invariants (and hopefully how to demux it with QUIC) in a new RFC.

-- 
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/1260#issuecomment-378007647