Re: [idn] Re: hostname history hell

[Eric Brunner-Williams in Portland Maine <brunner@nic-naa.net>]

> > as i thought, "3.*." was both valid, post-1123, and worked too.
> 
> There is a RFC 1591 rule/ comment that notes that no TLD has an
> all-numeric name and implies that one would never be approved.

1591 Domain Name System Structure and Delegation. J. Postel. March
     1994. (Format: TXT=16481 bytes) (Status: INFORMATIONAL)

informational, not normative, and reflecting a nation-state's policy of the
epoch, not the dns. modernly, see 2606.

2606 Reserved Top Level DNS Names. D. Eastlake, A. Panitz. June 1999.
     (Format: TXT=8008 bytes) (Also BCP0032) (Status: BEST CURRENT
     PRACTICE)

> 1591 is also, if I recall, where the "one character" prohibition
> originates.

I'm rather fond of 1591, partuclarly section 3. The word "duty" sits there
in splendid isolation.

Given a choice between a.isi.edu, and sections 2, or 4, of 1591, I'll pick
"a host". s/prohibition/advisory/

> people and applications might have trouble figuring out whether
> 10.250.250.1 is a DNS name or an [IPv4] address.

One things for sure, whatever ".1." is, it isn't a 3166 code point.

Use of numeric hostnames is not surprising, e.g., 3721.com, for lots
of reasons. Anyway, neither people nor apps are the dns.

> > when bill, donald, and i were working on what became 2929 we
> > discussed 2-octet lables (first left of ".") that were outside
> > of the current 3166 set, both as 7-bit and as 8-bit labels.
> > there was no reason other than convention, and now icann, to
> > not use ".3." as the first label after dot, at least in theory.
> 
> As long as 1591 is considered "convention", this is certainly
> correct.

2929, section 3.3, RR NAME Considerations, final para:

   A somewhat dated description of name allocation in the IN Class is
   given in [RFC 1591].  Some information on reserved top level domain
   names is in Best Current Practice 32 [RFC 2606].
   
>      john

eric