Re: iesg comment re message submission in draft-ietf-grip-isp-expectations-03.txt

[Patrik Fältström <paf@swip.net>]

At 14.10 -0700 00-05-30, John Myers wrote:
>With different port numbers, you can apply to the submit port a policy
>of refusing all unauthenticated transactions.  On the same port, one
>typically needs to accept unauthenticated transactions to local
>addresses as this is needed for mail relay.  A policy of refusing all
>unauthenticated transactions leads to a more consistent and
>understandable user experience when unauthenticated mail submissions are
>attempted.

Ok, so "understandable user experience" is what we talk about here. 
Nothing technical.

That was what I wanted to know.

If we talk about this, should the draft talk abut "strongly 
recommend" use of the separate port?

I.e. what I myself is currently implementing at Tele2 is the 
following algorithm on port 25:

   if the user is authenticated
     accept delivery of mail to any recipient
   else
     if peer address is within Tele2 "friendly address space"
       accept delivery of mail to any recipient
     else
       accept only local delivery
     fi
   fi

My point was that I didn't understand how this algorithm ended up 
being simpler (technically) with the use of a separate port number.

But as I said before, I think the new text suggested by Rendall is ok.

Ship it!

   paf