Re: [TLS] Review of draft-wouters-tls-oob-pubkey-00.txt

[Martin Rex <[e-mail]>]

Nikos Mavrogiannopoulos wrote:
> 
> The other goal of yours of sending raw public keys might as
> well be solved by defining a new certificate type RAW in the
> certificateType registry defined in RFC6091 and send anything you like
> in the "Certificate" message. You don't need WG consensus to do that.
> But do you really have a use case that makes all that effort
> worthwhile? Why would others be interested into implementing
> the RAW keys you are proposing?

I find the idea of extending rfc6091 with a new certificate type
for raw keys more appealing that a completely new TLS extension.

I also prefer the server key to part of the full TLS handshake, so that
the situation "client doesn't trust server key" or "client expected
different server key" can be reliably distinguished from other reasons
of a Finished message verification failure.

-Martin