[Ietf-not43] IDN's and IRIS

[Eric A. Hall]

on 9/3/2003 8:50 AM Stephane Bortzmeyer wrote:

> On Wed, Aug 27, 2003 at 05:24:26PM -0400,
>  Andrew Newton <anewton at ecotroph.net> wrote 
> 
>>At the output of nameprep is where certain judgements are made with 
>>regard to code point equivalences, thus creating these things called 
>>domain variants.
> 
> An important point: these judgements are registry-specific, partly
> because they strongly depend on the natural language(s) commonly in
> use for a given ccTLD. For gTLD, finding the right bundle may require
> that the client sends the language used.

Right. The operator of the zone makes delegations according to their
policy, and all such policies are unknowable for every client.

> I suggest that the whole issue of variants is too complicated and not
> enough understood at the present time to be incorporated in the
> protocol.

It just needs to be correctly focused. Trying to put partition-specific
logic into either the client or the protocol is a guarantee for failure.

Since these are zone-specific policy issues, let the operators deal with
variants however they are going to, and let the clients and protocol stick
to ~equality matching.

There is no need whatsoever to have clients attempt to guess at what
variants may or may not exist for a domain name, or if a domain name is a
variant or a canonical entry. Let queries match or fail, and have servers
inform the client of the entry status if this is desirable for some
reason. There's no reason to include charsets or anything else in the
query, just normalize it and see if there is a match.

-- 
Eric A. Hall                                        http://www.ehsco.com/
Internet Core Protocols          http://www.oreilly.com/catalog/coreprot/