On Wed, 31 Dec 2003, Paul Jarc wrote:
> Ok, I'll try that.  If I find that some clients can't handle it, I
> guess I'll add "anonymous"-only LOGIN.  In that case, do you think I
> should also allow AUTH=PLAIN with "anonymous" as the username?

I don't see much point to doing that.  LOGIN with anoymous as the userid
is enshrined for legacy purposes (see RFC 3501 section 6.2), but not use
of a userid of anonymous in a SASL mechanism such as PLAIN.

A valid implementation could accept anonymous as a userid to the LOGIN
command but not as a userid to a SASL mechanism.

> I
> guess some clients might send a real username and password belonging
> to another service, only to be rejected here, but sniffers would still
> get the password.  But I don't know whether advertising AUTH=PLAIN
> would make this any more likely.

That is the risk taken with LOGIN and with AUTH=PLAIN.

The safest thing is LOGINDISABLED and no AUTH=PLAIN.

-- Mark --

http://staff.washington.edu/mrc
Science does not emerge from voting, party politics, or public debate.
Si vis pacem, para bellum.

Reply via email to