I categorize senders in three categories: trusted, normal, blocked. Trusted
systems are allowed to send, blocked systems are not, normal systems are
subjectable to further checks.
We will need to balance between filtering in the handler, and in the
pipeline, but as one possible future scenario ... a server signs in that
isn't in a block-list. As soon as it starts sending messages, the pipeline
kicks in. If a matcher says that a message is spam, information is added to
a short-circuit system. At that point, James will reject all RCPT TO
commands that are not for postmaster, and *only* for postmaster (the RFC
mandates that postmaster accept e-mail without extremely strong reasons for
rejection), with a notice that e-mail from their system is forbidden (with
indication of why), and that they can contact the postmaster if they feel
that this is in error. E-mail that is intended for postmaster might be
limited to a single, size-limited, message per session.
Note that by spoofing their origination IP address, courtesy of
unconscionable negligence on the part of ISPs, a sender may be able to get
past an IP-based filter. Filters in the above system won't be just
IP-based.
--- Noel
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
