--On Samstag, März 13, 2004 20:13:45 +0100 Daniel Roethlisberger
<daniel(_at_)roe(_dot_)ch> wrote:
[...]
<MAILER-DAEMON(_at_)sitename> is very common for UUCP-setups as UUCP does not
allow empty envelope senders.
Then those systems forwarding bounces from such UUCP setups to the open
Internet seem to violate a largish number of RFC, for example RFC 2821,
section 6, unless they rewrite bounces' return paths to <> before
forwarding them. I don't think working around brokenness like that is
necessarily a good thing.
But RFC2821 is about SMTP which is *NOT* used then, it is UUCP. Sections
3.8.3 and 3.8.5 recommend rewriting but do require it. Section 6 is not
applicable as the bounce has not been generated by an SMTP system but
rather an UUCP system as is being gatewayed into SMTP space.
But anyway, wrong list?
Maybe not. We assume that bounces always have an empty return path in
SPF algorithms, but there are a few systems/networks out there that
still use different protocols internally and are gatewayed from/to SMTP
space. We should avoid breaking communication with these systems (X.400
is already broken enough, IMVHO, though still used by several government
networks here in Germany) if we want to encourage adoption.
And indeed the users in these gatewayed environments would benefit much
from SPF: they have well defined mail sending systems already (the
gateways) and (in case of UUCP) often very little possibility of
checking for invalid recipients during SMTP transaction (on the gateway
system).
Ralf