EL> But again, all of the RFC's were assuming that any connecting
EL> client would at least try and give you valid information,
EL> not some forged information.
The problem is that the only information that one can _know_, with
certainty, to be _forged_, are the domain names that _one's own mail
system uses_ and non-unique domain names such as "localhost." and
"localdomain.". There's no way that one can know that any SMTP Relay
client's use of (say) "ibank.barclays.co.uk." is not in fact entirely
legitimate. The only (semantic) requirement for the argument to
a "HELO"/"EHLO" command is that it be globally unique (across the
whole of the SMTP-based Internet mail transport network) to a single
mail system, so that server-side loop detection works. It's perfectly
possible for the owner of "barclays.co.uk." to have made a private
agreement with the owner of some other mail system for that mail
system to use the domain name "ibank.barclays.co.uk." for loop
detection. It's also perfectly possible for the unique domain name
used to identify a mail system to not have mappings (to IP addresses,
content DNS servers, or SMTP Relay servers) in the public DNS database.
It's _solely_ an arbitrary unique identifier for server-side loop
detection, not a host name, a server rôle name, or a mailbox domain
name, after all.
<URL:
http://homepages.tesco.net./~J.deBoynePollard/FGA/smtp-avoid-helo.html>
