ietf-mta-filters
[Top] [All Lists]

Re: Small comment on 3028bis redirect

2006-06-14 14:45:52


I'm reading over the redirect action, and it makes reference to the
behavior of "a .forward file using sendmail under UNIX."

Could the intended behavior be made explicit? I'd to know if the
envelope sender on the new outgoing message should be the Sieve
recipient or the original sender.

This is implementation and situation dependent. There are reasons why the MAIL
FROM address may need to change and reasons why it may need to stay the same.

One factor is whether or not sieve acts before, during, or after final
delivery. If actions are taken prior to final delivery there's a good argument
to be made that redirect is equivalent to an MTA forward and the MAIL FROM
address and envelope id should not change. If, OTOH, redirect is done
client-side after final delivery there's a good argument to be made that you're
dealing with  a new submission to the transfer system and the message needs to
have a new MAIL FROM and envelope id.

And uh, since it seems somewhat obvious now that I think about it that
it cannot be the original sender, who is not likely to be vouched for by
the relaying MTA if, say, there's an SPF query... probably best to
clarify the language in that paragraph in section 4.2 of 3028bis.

The fact that SPF has issues in this regard is one of the reasons why it is an
experimental specification, not standards track. SPF doesn't play very nicely
with many sorts of MTA-level forwarding, including but not limited to some
implementations of sieve redirect. Some of these types of forwarding have been
part of email pretty much since the beginning. THis is why SRS exists, after
all.

In any case, while I would have no problem with adding some discussion of this
issue, I would strongly object to any recommendation that the MAIL FROM address
MUST, MUST NOT, SHOULD, or SHOULD NOT change.

                                Ned

<Prev in Thread] Current Thread [Next in Thread>