> In message <[EMAIL PROTECTED]>,
> "Roger Fajman" <[EMAIL PROTECTED]> wrote:
>
> >This suggests that there could be messages other than bounces that
> >have a null MAIL FROM.
> >
> >LISTSERV's "probe failed" message sounds to me like an "error
> >reporting message".
>
> You are obviously using a VERY loose definition of the term.
>
> If I decide to write a letter (or have some automated bot do it for me)
> to some sysadmins somewhere, telling him that he's a bonehead for con-
> tinuing to run an open mail relay, is that an ``error reporting message''
> also?? Well... you could call it that.
>
> Where do you draw the line?
>
> I believe that the only sensible place to draw it is to say that the term
> ``error reporting messages'' (in the context of the relevant SMTP RFCs)
> refers to error reporting messages having to do with (and/or generated by)
> the SMTP mail transport system itself, not by mere clients thereof. Other-
> wise, these ``error reporting messages'' become fair game, and _everybody_
> can start calling _their_ messages ``error reporting messages''. (``You're
> a dork, and you ice isn't cold enough!'' There! Now _that's_ an error
> message! :-) Or how about ``Your request to Majordodo failed because it
> was unintelligible and unparsable.''? Or how about ``This is the vacation
> program. Joe will read your mail when he gets back from the Bahamas in
> two weeks. Until then, live in envy.''?)
>
> I am merely a client of the SMTP mail transport system. So are _all_ of
> the various mailing list packages, including Listserv. I don't screw
> around and go out of my way to make my messages look like bounces (even
> though doing so might appear to be convenient for me in some special
> cases... just as it appears to be for some spammers) and I don't believe
> that Listserv should be crossing this line either. Again, Listserv isn't
> a part of the SMTP transport system, it is a mere client of that system,
> as am I.
>
>
> -- Ron Guilmette, Roseville, California ---------- E-Scrub Technologies, Inc.
> -- Deadbolt(tm) Personal E-Mail Filter demo: http://www.e-scrub.com/deadbolt/
> -- Wpoison (web harvester poisoning) - demo: http://www.e-scrub.com/wpoison/
I believe that the original motivation for the presence of the null MAIL FROM
in RFC 821 is the prevention of loops. That's why RFC 821 says that error
messages should use it, why DSNs use it, and why MDNs use it. It doesn't
matter what's generating the message. MDNs are generated by user agents,
not SMTP servers. By your definition, they would not be allowed to use
a null MAIL FROM.
Now LISTSERV's probe failed message is not likely to cause a loop if it
did not have a null MAIL FROM, but only because LISTSERV has other measures
to keep that from happening.
There are other cases when a list server might need to use a null MAIL
FROM. For example, suppose that it forwards copies of error messages
received as a result of messages sent from the command processing
address to the list server maintainers. Such messages might cause
a loop if they bounce and the MAIL FROM is not null.
My main point is that there are things other than bounce messages that
are specified by Internet standards track documents to use a null MAIL
FROM. They can't be considered to be spam.
