[comp.mail.sendmail] Theoretical address questions

NAHAJ@cc.utah.edu (John Halleck) (06/26/90)

We were looking at some theoretical problems associated with parsing addresses 
with rewrite rules (as sendmail does).  A number of questions have come up
that we can not answer here.  Comments (Via E-mail please) are solicited on
the following questions:  I'll summarize answers if people are interested.

1.  The most common address formats have the property that (in the absence of
    nesting two or more formats) the "next" host is on the opposite end of the
    address from the user name.  (E.G.  host!host!host!user it is the host
    on the opposite end from the user name that is used.    In the source route
    flavor   @a,@b,@c,@d:user@site  it is the site A that is "next" and not
    site D.

    The question is:   Is there an address format in use anywhere that does
    not have this property?   I.E.  Is there a format such as

           user .. host .. host .. host

    Where the FIRST host is the next hop?

2.  Are there any address formats in common use other than the following:

    @host:stuff
    @host,@host,...@host:stuff

    site!stuff

    stuff@site

    stuff%site

    site::stuff

    name at site

    site:user

    site^user

-John Halleck
nahaj@cc.utah.edu
nahaj@utahcca.bitnet

rsalz@bbn.com (Rich Salz) (06/29/90)

In <73126@cc.utah.edu> NAHAJ@cc.utah.edu (John Halleck) writes:
>1.  The most common address formats have the property that (in the absence of
>    nesting two or more formats) the "next" host is on the opposite end of the
>    address from the user name.
That's an interesting observation.  I've never heard that explicitly
expressed before.  Neat!

>    @host:stuff
>    @host,@host,...@host:stuff
Ack, source routes.  Ignore them.

>    site!stuff
UUCP

>    stuff@site
The real thing.

>    stuff%site
Hmm, this is a local addres of some kind, to be interpreted on the local
host.

>    site::stuff
Looks like DECNET.

>    name at site
Outdated format -- do not bother to support it.  Odds are pretty good that
your sendmail can't really handle it anyway.

>    site:user
Berknet.  Ignore it.

>    site^user
Real old (v6? berknet? uucp/csh?).  Ignore it.
-- 
Please send comp.sources.unix-related mail to rsalz@uunet.uu.net.
Use a domain-based address or give alternate paths, or you may lose out.