Re: [idn] Requirements I-D

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

--On Tuesday, 16 May, 2000 19:48 -0600 mark.davis@us.ibm.com
wrote:

> Perhaps you misunderstand me (or I you).
> 
> The problem as stated to me was how to extend the current
> caseless comparison of domain names to the full range of
> Unicode characters. My answer assumes that  DNS names are
> *stored* in a case-folded manner. Whatever the user typed in
> would also be case-folded when by the time the strings are
> compared for identity. I was not suggesting that every possible
> match be tried!

Understood.

> I am also at all not arguing that there *should be*
> country-specific case folding; entirely the contrary -- that
> there be a single case-folding algorithm that is
> language-independent.

I don't believe, for reasons that others have explored on the
list, that a "case-folding algorithm" (in the "drop one bit in
ASCII" sense) is feasible.  Instead, one would need mapping
tables and would need to understand that some of the "foldings"
are not reversible (e.g., ones that remove accents or
diacriticals when going from "lower case" to "upper case".
Those mapping tables would presumably need to be updated each
time a character was added to the code set, so they had probably
best be embedded in the DNS files/tables or, at worst, in the
servers -- if every client machine on the network needs to be
updated each time new characters are added, I think we have a
non-solution.

I'm also concerned about the problem that acceptable
mapping/folding rules may differ from locale to locale with the
same characters.  Perhaps that is in the "bad, but not terrible"
category, but what this is supposed to be all about is letting
people use their own characters and languages in natural ways.

So I was looking for a way to deal with the problem without
embedding the rules in code we'd have trouble changing and that
was where the "put in everything" idea came from -- just an
idea, although obviously not a very good one.

    john