Re: [quicwg/base-drafts] TLS alert no_application_protocol is not always possible (#3580)

[Martin Thomson <[e-mail]>]

I see.  I hope that the application isn't the callback code, but that quicly is involved in this somehow; otherwise, you have just made H2O responsible for fulfilling QUIC requirements that you might assume quicly would be responsible for.

In general, I would assume that you have access to QUIC transport state in your callback from the TLS stack.  Let's say that you choke on the ALPN.  You cause TLS to issue a generic alert (which is fine, as Lucas points out).  At some later point that alert is going to hit QUIC code again so that it can be translated to a QUIC alert.  If you can save the reason from the callback (presumably you need this capability to save transport parameters) then you can construct the right message.

Of course, that's complicated.  The generic alert is fine.

-- 
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/3580#issuecomment-614953736