Re: [quicwg/base-drafts] Describe PMTU probing that includes source connection ID for routing … (#2402)

[MikkelFJ <[e-mail]>]

> The probe is a part of the connection. Otherwise it will not elicit an ACK of the coalesced short packet.
But it will use a dedicated connection ID.

> I am a bit confused that we are discussing the cost of a single if statement.

Yes, but it still impacts the design because the first thing I would do would be to look up bytes 1-4 in a hash table and call a handler function that has no context for handshake. So it will now have to add an indirection and process packets it does not understand. It can't be the same logic as the one that handles handshakes, because it would start by verifying the packet.

Or that initial lookup would would have to first check the packet type and go slow path, but the context must be looked up by ID, and the general slow path expects context, or it will drop packets. But you can't do that here, because there is no context. So you have to special handle that.

So performance-wise it would be survivable, just pretty ugly.

-- 
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/2402#issuecomment-461019172