I beg to differ about the "normal" form of addresses.  I frequently
see ARPA mail addressed in the "phrase <local-part@domain>" format.  The
"normal" form is, however, totally irrelevant.  The question should not
be "what is normal?", but "what is valid?".  LISTSERV should make every
attempt to make sure it understands not just the normal, but the valid.
And if it's not valid, it should make a valiant attempt to turn it into
something valid.
 
Actually, there's nothing wrong with an address that looks like
 
   Eric <@WISCVM:ERIC@FRECP11>
 
The @WISCVM: is an RFC822 explicit path specification (see section 6.2.7
of the RFC).  The meaning is that this mail item must be routed through
WISCVM to get to its destination.  Mind you, the RFC discourages this
(also known as "source routing"), saying the transmission route should be
left to the mail transport service.
 
I think you have two choices:
 
  a) Ignore the explicit path specification.  Probably will work most of
     the time.
  b) Strip off the first explicit path specification and send the file to the
     transport service (Crosswell MAILER or whatever) at that site.  I.e.,
     make "Eric <@WISCVM:ERIC@FRECP11>" into "Eric <ERIC@FRECP11>" and send
     the file to WISCVM.
 
(I haven't tried that example, by the way.)
 
Remember, the general case is that you have
 
      phrase <@domain1@domain2@domain3...@domainN:local-part@domain>
 
which would mean that the mail item would need to be routed to
domain1, from there to domain2, from there to domain3, ..., from there to
domainN, from there to local-part at domain.  Each domain in the path
could strip the left-most domain from the explicit path specification and
send the file to that domain, trusting that that domain would know what
to do from there on.
 
Richard Schafer