Re: homework, not an experiment, draft-crocker-email-deliveredto

"John Levine" <[email protected]> 4 Aug 2021 13:00:30 -0400
Newsgroups gmane.ietf.smtp
Organization Taughannock Networks
Message-ID <[email protected]>
It appears that Dave Crocker  <[email protected]> said:
>On 8/3/2021 10:10 PM, Viktor Dukhovni wrote:
>> In Postfix, the address that is presently recorded is the result of any
>> internal rewriting such as mapping"[email protected]"  to some internal
>> "[email protected]", that may not correspond to any public address
>> of the recipient. ...

>>    "Delivered-To:" FWS Mailbox CRLF
>>                     ; Mailbox is from [SMTP]

The thing after the FWS is an opaque token. Sometimes it may look like
a mailbox, sometimes not. And even if it looks like a mailbox, it
often is *not* a mailbox. You can't use it for delayed running of
procmail or redelivery via fetchmail or anything else that expects a
mailbox.

As a concrete example, if you send mail to [email protected], here's the Delivered-To

 Delivered-To: [email protected]

That pseudo-address reflects the detailf of my mail set up, but if you or I
tried to send mail to it, it wouldn't work. 

It would make sense for us to do a survey of MTAs and describe how they actually use
Delivered-To and write it up, which is a report rather than an experiment.  Take a look
at RFC 7085 in which Paul Hoffman and I did a survey of TLD apex DNS records for what I mean.

R's,
John