Unicode position on local mapping

[John C Klensin]

--On Wednesday, February 18, 2009 09:31 -0500 Andrew Sullivan
<ajs at shinkuro.com> wrote:

> On Tue, Feb 17, 2009 at 07:10:24PM -0500, John C Klensin wrote:
> 
>> Exactly.  But I read Jefsey's note as saying that it was up to
>> the registry whether to do that or not.  If it was not done,
>> then the two strings would effectively be treated as
>> completely different strings, as different as, say, "école"
>> and "school".
> 
> Right.  I thought registry operators regarded that as a
> feature, not a bug.

In general, they do.   But they are, substantially by
definition, used to thinking about the realities imposed by the
DNS.  Jefsey, if I correctly understand him, is trying to think
outside the DNS box and about a more general, and more precisely
localizable, system design.  

At some very fundamental level, I think the issues he is
concerned about are very real.  I believe that they can (and
should) be addressed by systems and applications layered on top
of the DNS (he and I even agree to say "presentation layer" when
we talk about those things, but that term involves a certain
amount of hand-waving in this context).   But, as soon as one
gets that layer and puts it in charge of localization (or, from
his perspective, multilingualization), then one immediately
faces two questions that are far out of this WG's scope: (i)
should such systems be layered on top of the DNS as we know it
or become a completely separate name resolution system designed
with different criteria? and (ii) do IDNs, especially
Unicode-based IDNs, that use the DNS have any role and what is
it?   I believe that he and I have very different answers to (or
at least hypotheses about) those questions.

      john