Re: [idn] Prefix for testing AMC-Z implementations

[John C Klensin <klensin@jck.com>]

--On Monday, 24 September, 2001 22:12 -0700 Mark Davis
<mark@macchiato.com> wrote:

> I don't see why this is an issue at all. Any two letters could
> mean something somewhere to someone. The fact that they happen
> to be country codes should be irrelevant.

Yes.  It makes a good deal of sense to me that we avoid using
two-letter IS 3166-1 codes as DNS labels, especially at the
second level in widely-known generic domains.  On the other
hand, it seems to me that banning any two-character sequence
that corresponds to a 3166-1 alpha-2 code from appearing
anywhere in a label would be excessive, unreasonable, too
constraining, and not accomplish very much.

Whether having those two letters as the first two letters of a
label --whether followed by a hyphen or not-- is a significantly
different situation, I don't know, but I suspect it is not.  And
use of the 3166 "individual use" codes doesn't seem to me to
help a lot either, since may leave too small a selection set
(and encourage cybersquatting on those prefixes).

FWIW, I could make a much stronger case for reserving the
two-letter ISO 636 codes than I can for 3166 ones.  We might
actually need the former if the "above DNS" approaches to
accomodating the need for language tagging (where that need
exists) fails and we eventually have to evolve a way to do that
in the DNS labels.

   john