[Top] [All Lists]

Re: Dotless domains

2009-01-25 09:08:41

--On Saturday, January 24, 2009 18:13 -0800 Steve Atkins
<steve(_at_)blighty(_dot_)com> wrote:

On Jan 24, 2009, at 6:01 PM, John C Klensin wrote:

--On Sunday, January 25, 2009 12:30 +1100 Mark Andrews
<Mark_Andrews(_at_)isc(_dot_)org> wrote:

    Dotless hostnames are in the local namespace and can *never*
    be made to work *reliably* in a global context.

    Note the use of non-heirachical names is undoing the changes
    introduced by RFC 921 and will introduce problems RFC 921
    was trying to remove/prevent.

Yeah.  Since I have not been able to find a single hint in
ICANN's new TLD plans that those TLDs would be restricted to
delegation-only uses, tell it to ICANN.  Or tell it to
whomever is supposed to be supplying adult supervision to
ICANN  :-(

Not an SMTP problem.  SMTP requires FQDNs, without exception,
and does not permit single-component ones.

5321 explicitly says otherwise  ("A domain name (or often just
a "domain") consists of one or more components, separated by
dots if more than one appears. In the case of a top-level
domain used by itself in an email address, a single string is
used without any dots.").

The whole point of this thread is to dig a little deeper into
why that decision was made and what the implications are, I

Arggh.  At a minimum, we missed a place where the document
should have been made completely clear and consistent.

Aside: While I strongly agree with the decision --with both 2821
and 5321-- to not make more changes than were necessary, one
result is that we have ended up with nearly the same
information, stated differently, in multiple places.  The
problem actually goes back to the 821 decision to describe the
commands and the details of their operations separately, but
accretion has made it much worse.  If I saw any hope of getting
it right and consistent with the current spec, a rewrite from
scratch to produce a more compact document with no repeated
information would be really attractive, especially since the
result would probably be only half or 2/3 the length of 5321.


<Prev in Thread] Current Thread [Next in Thread>