Path: utzoo!censor!geac!jtsv16!uunet!tut.cis.ohio-state.edu!armadillo.cis.ohio-state.edu!lum
From: lum@armadillo.cis.ohio-state.edu (Lum Johnson)
Newsgroups: news.newusers.questions
Subject: Re: questions
Message-ID: <57019@tut.cis.ohio-state.edu>
Date: 7 Aug 89 20:44:53 GMT
References: <4051@udccvax1.acs.udel.EDU> <1710@frog.UUCP>
Reply-To: Lum Johnson <lum@cis.ohio-state.edu>
Followup-To: comp.mail.headers
Distribution: usa
Organization: The Ohio State University, IRCC/CIS Joint Computing Laboratory
Lines: 71

In article <1710@frog.UUCP> jp@frog.UUCP (John Pimentel) writes:
>Typically, paths are:
>	user@system.type   or	user@system.company   or   user@system.school
>However, some involve some relays in the path:
>	user@system.company.relay.network
>The rest of the names in the path are the systems transfering information
>from point A to point B.

The address (or fully qualified domain address specification):

  user@system.company.relay.network

is different than the path that is apparently perceived as equivalent:

  network!relay!company!system!user

The latter actually specifies a path (thus the name) consisting of
machines through which information will be transferred;  the former does
not specify any path at all - it serves only to _identify_ a host.
A path to the host may be found, if appropriate, but usually the concept
of path itself is inappropriate when speaking of Internet addresses.

The point of the domain system is that `foo.bar.bletch.edu' specifies
exactly one host: `foo.bar.bletch.edu', in domain `bar', a subdomain of
domain `bletch', a subdomain of domain `edu'.  `edu', `bletch', and
`bar' are domains, and they are almost certainly not (also) hosts;
there is only one `bletch' subdomain in the `edu' domain, only one `bar'
subdomain in the `bletch' domain, and only one `foo' host in the `bar'
subdomain.

The first few examples:

  user@system.type  or  user@system.company  or  user@system.school

are actually illegal on the Internet, unless `type' is one of the
following:

	.COM	.EDU	.GOV	.MIL	.NET	.ORG

or a two-character "country code" of which there about a hundred.
(These were mistakenly tacked on late in the design as a fix for the
political objections of the European PTTs.  A commercial organization
in France and one in the US should not be in separate domains;
they should both be in the .COM domain.  But that's another rant.)

Some pseudo-domains (eg, .BITNET and .UUCP) are tolerated, since
otherwise it would be still more difficult to move mail between the
Internet and other networks.  No RFC requires anyone to recognize such
pseudo-domains or do anything useful with them, however;  it merely
works better if the rules are not enforced too strictly all the time.

The reasons that .BITNET and .UUCP are pseudo-domains rather than
authentic domains are:  these "domains" have no domain structure (ie,
it is always `foo.bitnet', never `foo.bar.bitnet', because the `bitnet'
pseudo-domain does not (and can not) contain any subdomains - to
oversimply, the name space is "flat" rather than "hierarchical");
and these "domains" are not registered as such with NIC.DDN.MIL (nee
SRI-NIC.ARPA), with whom all legal domains are registered.

This is not really the newsgroup for discussing questions as difficult
as this anyhow;  I'm redirecting this to `comp.mail.headers', which
seems to be where such things are discussed.

Or you may send mail to `postmaster@osu-20.ircc.ohio-state.edu'.

Lum
-=-
-- 
Lum Johnson      lum@cis.ohio-state.edu      lum@osu-20.ircc.ohio-state.edu
"You got it kid -- the large print giveth and the small print taketh away."
-------