Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.3 4.3bsd-beta 6/6/85; site topaz.RUTGERS.EDU
Path: utzoo!watmath!clyde!cbosgd!cbdkc1!desoto!packard!topaz!root
From: root@topaz.RUTGERS.EDU (Charles Root)
Newsgroups: net.mail
Subject: Re: a common sendmail.cf bug
Message-ID: <3581@topaz.RUTGERS.EDU>
Date: Sun, 8-Sep-85 22:32:41 EDT
Article-I.D.: topaz.3581
Posted: Sun Sep  8 22:32:41 1985
Date-Received: Tue, 10-Sep-85 03:25:33 EDT
References: <9600010@uiucdcsb> <10306@ucbvax.ARPA>
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 18

Alas, @DECWRL.ARPA:user@host.DEC is not legal according to The
Official Interpretation of RFC822.  In a route address, all of the
things after @ must be legal host names.  In order to a legal host
name, it must be registered with the NIC, or possibly these days part
of a domain that is sanctioned by NIC and has the necessary redundant
domain servers.  This syntax is not like the UUCP !.  With
foo!bar!baz, as long as you know what foo is, you don't need to know
anything about bar and baz.  It suffices for each host to know the
next one.  Under RFC822, all addresses are absolute.  The original
sender must know (or at least be able to find out using domain
servers) the Internet addresses for all of the hosts in the path.
Clearly this makes the syntax somewhat less useful that it first
appeared to be.  That is why % was invented in the first place.
Believe me, it was not because people were too lazy to implement the
RFC822 route address syntax.  I had a complete implementation of it
for TOPS-20, and was told I couldn't use it, at least not in any way
that accomplished anything useful.  The people behind TCP specifically
did not want to supply any equivalent of the UUCP !.