Re: [idn] Requirements I-D

[RJ Atkinson <rja@inet.org>]

At 20:08 15-05-00 , A. Vine wrote:

>While I understand your desire to make case-folding locale-independent, I
>recommend you consider a couple of issues. 1. Case-folding in a
>real-world-language usage sense is _not_ locale independent, which means 
>that 2.
>no matter what standard chart of case folding you come up, _including_ the
>existing ASCII case-folding, you will leave several languages/locales with
>incorrect folding.  In addition to potentially causing problems and 
>confusion in
>the domain naming world, it will also serve to annoy people, committees, and
>national standards bodies from various parts of the world.

I think I do understand the issues here as that matter arises within the set
of languages that I speak/read.  DNS is NOT a text-processing situation
where case-folding really must be locale-dependent to behave correctly.

Rather this is a case where case-folding is not involved with text processing,
merely with the uniqueness of a Domain Name/Hostname.  In such a situation,
the locale-dependent attributes are not adversely affected to the same extent
(not even nearly to the same extent) as in a text-processing context.

Doing case-folding in a locale-dependent way cannot be interoperable globally,
so that is a non-starter on backwards-compatibility and interoperability 
grounds,
so I don't see that we really have a choice here, to be honest.

If you have a specific counter-proposal, including explanation of why that
proposal is interoperable globally and fully backwards compatible, please
outline it here or (better) in an I-D.

Regards,

Ran
rja@inet.org