Re: [quicwg/base-drafts] Long Header Packets and Routing Connection IDs (#2834)

[Kazuho Oku <[e-mail]>]

@ianswett IIUC, the design principle has been to allow the server to initiate a CID change _for the packets that the client sends in response_, when the server receives the incoming packet for the first time. Generally speaking, I think that having the capability to switch the CID just once is sufficient.

That's why we allow server's Initial to change the CID.

However, because we'd also want a way to validate reachability, we added Retry. Then, because we thought that having the capability to redistribute the load using Retry would be a good idea, we added the capability to switch CID in Retry.

We can forbid the change of CID using Initial packets when Retry is used. But IIRC the question was what we gain from having such prohibition. Assuming that it's just an addition of complexity without any gain, we just allowed CID to change at most twice, once by Retry, and then by Initial.

While I agree that the design is not clean-cut, I think it's very pragmatic and fullfills our needs.

-- 
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/2834#issuecomment-505225402