Re: [quicwg/base-drafts] Rewrite text about Version Negotiation (#1039)

[Martin Thomson <[e-mail]>]

martinthomson commented on this pull request.



> -packets in anticipation of the Initial packet arriving later. This is
-especially useful for 0-RTT packets that routinely accompany Initial packets.
-
-Buffering ensures that data is not lost, which improves performance; conversely,
-discarding these packets could create false loss signals for the congestion
-controllers.  However, limiting the number and size of buffered packets might be
-needed to prevent exposure to denial of service.
-
-Servers MUST NOT send packets in response to these buffered packets until
-the initial packet arrives.
-
-In this version of QUIC, clients cannot send short form headers prior to a
-response from the server. Servers that do not support a version of QUIC that
-allows early short headers MUST either respond with a stateless reset or drop
-the packet silently.
+First, hosts try to associate the packet with an existing connection. If the

That's a good point.  It certainly makes sense for a server to push any stateless processing to the front.

I don't know if it makes sense to consider that as an optional amendment, or whether it isn't just easier to reorder the processing.  It depends on whether it is easier to say "a server might move any stateless processing of potential Initial packets to the start of this process", or whether it is easier to rearrange the order from the outset.

-- 
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/1039#discussion_r167749160