Re: [ietf-smtp] Possible cont4ibution to moving forward with RFC5321bis SMTP

[Keith Moore <[e-mail]>]

On 1/2/20 1:06 PM, Dave Crocker wrote:

> On 1/2/2020 9:18 AM, Keith Moore wrote:
>> no NAT between the MSA and the MX SMTP server.
>
>
> "MX Server"?
>
> Since MX is used for destination side delivery, I'm not sure which MTA 
> you mean, since MSAs aren't modeled as talking to the delivery site. 
> Please clarify. 

I mean a server that accepts incoming mail on behalf of a mail domain.  
Normally they are advertised using MX records.  Sorry if that wasn't 
clear; I thought "MX server" would be an unambiguous shorthand.

To me the point at which a message crosses a boundary between mail 
domains - i.e. from an MTA forwarding outgoing mail from one domain, to 
an MTA accepting incoming mail for what is likely a different domain, is 
interesting.   It's especially important for the standards to get that 
interface right, because that interface is what lets us send mail from 
arbitrary senders to arbitrary recipients and have a reasonable 
expectation of that mail getting there.

If MTAs A and B are both managed by the same mail domain, and A sends 
mail to B, it's not a problem whether B accepts EHLO with ip domain 
literals.   If B accepts them, fine, because it's only internal traffic 
anyway.   If B rejects EHLO with ip domain literals, that's also fine, 
because that same domain can presumably also arrange for A to use a DNS 
name there.    It's when the mail is relayed between mail domains that 
the standards need to provide guidance, because when sending domain and 
receiving domain have different assumptions about what is right, it 
causes unnecessary failures.

Keith