David Woodhouse writes:
It's just a single transition through the RFC821 transport system which
we're attempting to authenticate. So I agree with you -- yes, using
RFC822 headers (such as the Sender: address) _is_ a layering violation.
What I have asked repeatedly of the people who want a crypto
identity for auth/authz the last "sender" is why STARTTLS is
not adequate (ie, get to the 80/20 rule). Nobody ever
answers that.
I think MASS solves a different problem -- one that cannot
be addressed by STARTTLS.
Mike