SCRAM-SHA-256-PLUS support

[Neustradamus]

Dear @ircv3 team, @grawity, @jwheare,

It is possible to add -PLUS variant to have SCRAM-SHA-256-PLUS in more SCRAM-SHA-256?

Thanks in advance.

Linked to:

• State of Play scram-sasl/info#1

[progval]

Do you have an example use-case and motivation for IRC that would benefit actual users? RFC5056 mentions iSCSI delegating authentication to IPSec (which is not applicable to IRC as far as I can tell) and RFC5929 does not give any motivation for using it with TLS.

(And please, there is no need to add comments to other random SASL PRs/issues in ircv3-specifications to link to this one.)

[grawity]

RFC5929 does not give any motivation for using it with TLS.

On the other hand, my general impression is that TLS was meant as the primary use case of the -PLUS variants, with channel bindings ensuring that the host terminating your TLS connection is the same host handling SCRAM authentication, i.e. no MITM proxies. (Since SCRAM provides mutual authentication, a fake TLS server wouldn't be able to finish the auth handshake.)

I don't think channel bindings make much sense with traditional split ircd/services architecture, where the TLS channel is deliberately terminated elsewhere than SASL auth.

But either way, -PLUS variants (including for SHA256) are already specified outside of IRCv3. If some ircd (especially one with integrated services) wants to implement channel bindings, they can already do that.