[ On Tuesday, March 18, 2003 at 20:49:25 (+0000), Giuliano Gavazzi wrote: ]
> Subject: Now we talk! Was: [Exim] Reaction to rude 554 greeting
>
> I am not sure where you are getting with this, but since the object
> is highly hypothetical..
The archives of the lists which discussed the formation and editing of
RFCs 2821 and 2822 over the years are publicly available. I haven't
actually searched them on this very subject but I have reasons to
believe my conjectures are correct.
> in passing, just how can you or they or anyone say that it is not a
> DNS issue and that is a sending system's configuration problem?
Because you're placing the DNS in the wrong position in this discussion.
The DNS and what MX records may point anywhere are completely unrelated
to any of this issue.
> The
> receiver can only determine its own state and not the sender's. If
> the RFC really meant that, that it would be in error.
Meant "what" exactly? Although section 3.1 is surprisingly quite clear
on its own, I've also given it to you in different words and the
combination should make things so clear that even someone not familiar
with SMTP should be able to understand the issue now.
> As you can see, I was not quoting out of context, as that is
> *exactly* the snipped *I* quoted in my *first* message, and that I
> repeated in a subsequent...
When you quoted that paragraph you did effectively quote it out of
context because as far as I could tell it was not related to the
particular points of the argument you were trying to make at the time.
> Perhaps you should have given your interpretation before?
Why should someone like myself have to re-word something that really is
already quite clearly worded? Why did you lead the argument off into
totally bogus directions claiming various things that section 3.1 makes
very clear are false. Why did you then go about quoting other unrelated
portions of the RFC, sometimes completely out of context, in an attempt
to justify those bogus directions?
> What I do agree mostly is that a bounce must be generated, only that
> I said it should be directed to the postmaster of the client site.
> Why? Because the end user(s) will just be confused (and as an Hotmail
> user, you would be confused too). Why the postmaster, because he is
> the only one in the position to correct the configuration (if needed)
> or contact the relevant administrator if the problem is not in his
> smtp server.
Sorry, but your desire goes contrary to the entire design of error
handling in SMTP. There is absolutely no provision in SMTP for
specifying any alternate destination for any error report. Error
reports _MUST_ always be sent back to the sender address.
Users _must_ be trained to ask their own _local_ postmasters for help if
they don't understand a problem. This is partly why the RFCs require
that a human responsible for the mail system at a given site read the
mail sent to "postmaster" at that site. The standard mailbox name is
not just for remote people, but for local users to use as well!
It's probably OK for a given postmaster to configure his or her mailer
to send copies of all bounces to the <postmaster> mailbox, but anyone
doing that at even a medium-sized site these days will be swamped with
far more junk than they care to see, and also swamped with a great deal
more responsibility than they are likely prepared to deal with (in at
least the USA and Canada very strict limitations are placed on what
people who can intercept communications streams can do with the
information they might learn as a result, and while a postmaster is
always capable of intercepting all e-mail at a site, having all bounces
copied to the postmaster mailbox will effectively force them to read a
lot more private mail than they would normally otherwise see).
--
Greg A. Woods
+1 416 218-0098; <g.a.woods@???>; <woods@???>
Planix, Inc. <woods@???>; VE3TCP; Secrets of the Weird <woods@???>
