On Thu, 23 Jan 2003, D. North wrote:
> Understood. -- Perhaps I'm abusing the purpose of BSMTP a little then....
> I use the pipe transport to route mail to SpamAssassin (SA), then use BSMTP
> to re-inject it into exim for delivery. I'm having a hard time conceptually
> with a percieved inconsistency in the changes that are applied to the message
> in the process, because I want the message sent to SA to be re-injected into
> exim exactly as it was sent out to SA, but with the SA modifications (only) added.
That is a perfectly valid thing to want to do. Hmm. I hadn't realized
that this might cause problems with header lines that contained
non-fully-qualified addresses. (Of course, header lines coming from
external host should not contain such things, but ...)
> > For the normal SMTP port, investigate sender_unqualified_hosts and
> > recipient_unqualified_hosts (or was it receiver_unqualified_hosts in
> > Exim 3?)
>
> These do not seem to affect the 'To:' header rewriting though,
> which is where the rub is...
In my tests on Exim 4.12, they do affect To: header rewriting. A quick
check of the 3.36 code suggests that this behaviour has not changed.
From the 4.10 manual:
recipient_unqualified_hosts Type: host list* Default: unset
This option lists those hosts from which Exim is prepared to accept
unqualified recipient addresses in message envelopes. The addresses are
made fully qualified by the addition of the "qualify_recipient" value.
This option also affects message header lines. Exim does not reject
unqualified recipient addresses in headers, but it qualifies them only if
the message came from a host that matches "recipient_unqualified_hosts".
(I keep talking about Exim 4 because it has been out for almost a year
now, and Exim 3 is frozen and obsolescent.)
Thoughts:
It is probably a bad idea to set recipient_unqualified_hosts=* in
general.
In Exim 4, you might be able to check for unqualified addresses in
header lines in a DATA ACL (or local_scan() function) and reject the
message on that basis at that stage.
I should perhaps add a feature to -bS (and -bs) to disable qualification
in header lines so that you can mimic a remote message more exactly.
I've noted that requirement.
Philip
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
