Path: utzoo!attcan!uunet!husc6!mailrus!umix!honey
From: honey@umix.cc.umich.edu (Peter Honeyman)
Newsgroups: comp.mail.uucp
Subject: Re: host!user@sun.com (was smail wants you to register a domain)
Message-ID: <4608@umix.cc.umich.edu>
Date: 8 Sep 88 02:12:44 GMT
References: <1289@mcgill-vision.UUCP>
Reply-To: honey@citi.umich.edu (Peter Honeyman)
Organization: Center for Information Technology Integration, Univ of Michigan
Lines: 29
UUCP-Path: {uunet,rutgers}!umix!honey

mouse points out that when sun is relaying uucp mail into the
internet, they are compelled to generate a from header that
looks like From: something@sun.com.  with this i'm sure we all
agree.

mouse goes on to make the case that the uucp name of sun's
sender should be accounted for in the reconstructed from
header.  the logical extension of his argument is to copy the
uucp from line (you know, the first line) into the from header.

that way, sun can look at paul's from line, which says

	From vixie!paul some date remote from decwrl
	
and make a from header that says

	From: decwrl!vixie!paul@sun.com
	
ignoring paul's bogus from header entirely.  (I'd suggest
something like X-From-at-sun.com-was: vixie!paul.)  of course,
if paul's from header were already legitimate 822, sun shouldn't
touch it.

although my mailer doesn't do this (it's lazy, like sun's, but
at least i use pathalias, so the replies mostly work), i
suspect that using the unix from line instead of an existing
non-822 address in a from header is more likely to succeed.

	peter