Re: [ietf-smtp] what's not a trace field

[Dave Crocker <[e-mail]>]

On 2/14/2021 5:30 PM, John R. Levine wrote:
> FYI, Delivered-To is not always added by an MDA.  Sometimes, as in the 
> example I sent a few messages back, it's added by the MTA when it 
> rewrites an address internally.


It's almost amusing to think that a field specific to delivery is 
affixed by a module that doesn't do deliveries.  But only almost.

The idea that an MTA does some deliveries, as an accommodation to the 
alias mechanism, encourages an approach in which it is warts, all the 
way down.  Let's fix one wart by adding another.

Preferring simplicity instead of warts, the better modeling view is that 
aliases are handled by an MDA function.

Some offline discussion has uncovered limitations in RFC 5598 -- 
especially with the architecture diagram -- and that it needs revision, 
to provide better handling of a case like this.  For example, the 
current diagram requires all deliveries to go through a message store, 
which obviously is not correct for the alias example.


d/
-- 
Dave Crocker
Brandenburg InternetWorking
bbiw.net