Re: [quicwg/base-drafts] Handling of corrupt Retry packets (#3014)

[Kazuho Oku <[e-mail]>]

@ianswett 
> Functionally, a Retry or Initial with no payload should be identical, minus the packet format?
>
> * Both have a Token
> * Both have a way of validating the original CID
> * Both have no other payload

Ah thank you for the clarification. So the idea here is to allow the servers to send tokens in Initial packets (which is currently prohibited), when the payload length is zero.

I agree that that's a possibility, and that we could have taken that path.

OTOH, I am not sure if we should consider that at this late stage, even if we are to authenticate and encrypt the Retry packet. The reasons are:
* Use of the Initial packet does reduce the complexity. It just changes how the clients make the distinction between Initial vs. Retry.
* The existence of PN field is concerning, as the client cannot ACK a Retry.

>From the server's point of view, Retry is an action that happens before a connection is established, and I like the fact that it is given a distinct packet type rather than being defined as a subtype of a packet that could be part of a connection (i.e. carrying PN). And we have enough bits to express the four long header packet types.

-- 
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/3014#issuecomment-535746870