Re: [Emailcore] Delivered-To issues

[Dave Crocker <[e-mail]>]

On 1/2/2021 6:37 AM, Jeremy Harris wrote:
> On 02/01/2021 13:48, Dave Crocker wrote:
>> Any site performing the delivery transition of responsibility, from 
>> the MHS to the addressee's control, slaps a Deliver-To: header field, 
>> with that one address on it.
> 
>> delivery ... takes place /within/ the MDA
> 
> I think this implies that the address is site-meaningful and not
> necessarily externally-meaningful nor qualified.  For example,
> a message received with an envlope RCPT of "[e-mail]"
> is locally aliased to "fred", has "Deliver-To: fred" prepended
> and then added to fred's mailbox.


It's possible that one or another address -- notably the last one -- 
might only be site-meaningful, I suppose.

I'd be inclined to have the spec mitigate against that, though, by 
requiring it list the address in a way that is globally useful.  That 
is, after all, what the right-hand-side domain name is for.

(Note that the left-hand side is only ever site-meaningful, for Internet 
mail.  There is never a global interpretation of the LHS.)

Also note that 'delivery' is a transition out of the handling system. 
Just prior to that moment, the address being used is a classic Internet 
mail address, which is globally meaningful.  And it is to that address 
that the mail is being delivered.

Discussions have tended to imply or assert that delivery occurs at 
various other places or that things that occurr afterwards are, 
nonetheless, inside the MHS.

If a message reaches the RCPT TO address, anything that happens 
afterwards, such as being sent on to a new address, happens after 
delivery and involves a new submission.

d/
-- 
Dave Crocker
Brandenburg InternetWorking
bbiw.net