On Mon, Sep 09, 2013 at 09:00:01AM -0700, Todd Lyons wrote:
> > > [email protected]
> > > <-- [email protected]
>
> I think it's a terminology issue. It's not exim "rewriting" the email
> address. I believe a more technically correct phrase would be
> "normalizing". A domain is not supposed to have an MX record set to a
> CNAME.
You're mistaken. The illegal configuration is:
example.com. IN MX 0 mail.example.com.
mail.example.com. IN CNAME smtp.example.com.
smtp.example.com. IN A 192.0.2.1
However, the below (which is the OP's situation) is valid (since
April 2001, RFC 2821):
example.com. IN CNAME example.net.
example.net. IN MX 0 smtp.example.net.
smtp.example.net. IN A 192.0.2.1
More than a decade ago (prior to RFC 2821), envelope recipient
addresses of the form [email protected] (with example.com as
above) were expected to be canonicalized (normalized if you will)
to [email protected] since RFC 821 requires primary names in
all contexts where domains are used. This was relaxed in RFC 2821
with the express purpose of allowing <[email protected]> in
MAIL FROM: and RCPT TO:.
MTAs are now expected to not automatically canonicalize the domain
part of envelope recipient addresses based on the presence of a
DNS CNAME alone. Of course explicit rewriting rules in the MTA
configuration can rewrite local addresses at will, and remote
addresses at their peril.
> Behavior when such an event occurs is undefined. Some MTA's
> merely fix the sending domain from the (invalid) CNAME to the
> (standards compliant) A record. I *know* sendmail does this as I'm a
> list owner on a mailing list machine which has this particular
> configuration. Read on for exim behavior in this scenario:
Sendmail (in best-practice and default configurations) no longer
automatically canonicalizes the domain part of envelope recipient
addresses. The same is true of Postfix.
> Yes it has been changed, but no it's not incorrect (IMHO).
Your opinion is contrary to long established standards.
--
Viktor.
--
## List details at https://lists.exim.org/mailman/listinfo/exim-dev Exim
details at http://www.exim.org/ ##
