>    Much of the standard error handling that works is on the basis of
>"standard practices" rather than standards.  In essence the alaska system
>was forging mail by using that From line when it should probably have
>included the original message within it's own set of RFC822 headers.
 
>   However, RFC1123 (Host Requirements) is pretty clear of sending
>errors to the MAIL FROM address and not letting you off the hook just
>because SMTP had accepted the message:
 
The discussion that was attached to this was interesting and informative, but
I am still seeking an answer to the question:
 
What simple precaution can I, as a list owner, take to try to prevent
this kind of problem from recurring.
 
So far I have heard only two clear suggestions:
1. Ask my systems administrator to serve off (if that's the term) the
sp[ecific site (in this case alaska.edu) that caused the problem.
2. Use "filter" to ward off the effects of words like "postmaster"
 
I have seen LISTSERV documentation that mentions use of "SAFE" in the header.
It was difficult to understand what the actual effect is.  It seemed to imply
that maybe, just maybe, use of SAFE in the header would by itself
screen out the effects of wrongly stamped addresses such as "postmaster."
But that documentation is a wonderful literary docuemtn, open toa great
variety of interpretations.
 
Soooooooo... what's the best precaution? I've gotten messages from other list
owners indicating that they too have fallen prey to this hazard.
a good, clear, simple solution would be of benefit to a good many
lists.
 
Yours in continuing perplexity and occasional bemusement,
Marc Abrahams                                           ([log in to unmask])
Editor, The Annals of Improbable Research (AIR)
Former editor (1990-1994), The Journal of Irreproducible Results