Re: Working Group Last Call for draft-ietf-sieve-refuse-reject-00.txt

[Kjetil Torgrim Homme <[e-mail]>]

On Tue, 2005-08-23 at 14:08 -0400, Mark E. Mallett wrote:
> On Mon, Aug 22, 2005 at 12:41:05PM -0700, Matthew Elvey wrote:
> > How's this replacement?
> > 
> > (The first sentence is unchanged.)
> > 
> >   The "refuse" action refuses delivery of a message by sending back
> > 
> > the 550 SMTP response code to an SMTP client. This extension can be 
> > supported only by a Sieve implementation capable of sending the 550 over 
> > an SMTP connection between Administrative Units.
> 
> Well, I'm probably being anal (again).  To me, the use of "between
> Administrative Units" and "over the open Internet" are birds of a
> feather.

the current draft doesn't say "over the open Internet", so I don't see
how this is relevant?

"Administrative Units" would have to be defined, though.

> I don't think it's important to specify where the message is
> transported, or between whom-- only that the message is turned away by
> the receiver during transport, thus avoiding a later asynchronous
> notification via a separate email message.  e.g. I think refusal should
> apply to a transmission to the SMTP server running on localhost.

is an e-mail system which accepts all e-mail but sticks it in a
directory where it is subsequently picked up by a spam classifier, which
in turn resubmits the message via SMTP on localhost where the Sieve
script is run (*deep breath*) allowed to offer "refuse"?  I'd say no,
since the 550 will not be able to reach the previous e-mail system.

an MTA which doesn't double as a MDA will have to use a callout to the
next hop in the system and wait for its response before passing the
answer back to the other e-mail system.  (a system which does queueing
in some circumstances is fine, but queueing shouldn't be normal
behaviour.)
-- 
Kjetil T.