Kurt D. Zeilenga wrote:
And if they don't like CRAM-MD5 what they'll get is LOGIN or
PLAIN _without_ TLS, sigh.
I disagree with this statement. Today, many email client
and server supports TLS
Not my favourite old MUA, unfortunately. When I implement a
simple script I'm limited to a socket interface, and in that
case cram-md5 / digest-md5 / otp is the best I have. And the
server in question offers login / plain / cram-md5 for AUTH.
I think the best option for this protocol, given issues
raised by Simon regarding DIGEST-MD5, is TLS+PLAIN.
Where that's possible it's fine. I'm more interested in the
case where it's impossible. My understading of the draft is:
"Whatever you do stay away from PLAIN (or the obsolete LOGIN)
without TLS, use at least CRAM-MD5".
Maybe Brian's proposed compromise covers this concept somehow.
And he wanted "known weaknesses [citations]". That's about
today, not about some results of the not yet existing HASH WG
in 2006 or later.
Ietf mailing list