> -----Original Message-----
> From: RJ Atkinson [mailto:rja@inet.org]
...
> Stepping back from the details a bit, I'd like to recast
> some of this discussion in terms of requirements language.
>
> Would there be agreement that, in the DNS context,
> - some normalisation form is needed ?
I'd say yes.
> - a single normalisation form is preferable to having several ?
Single in what sense? We can use one for strings that are to be matched
(the names themselves), and another one for auxilliary texts (if those
are to be normalised at all). If we are to have caseless matching (via
a normal form, rather than collation keys, which is much more complex
and less stable), then that cannot be used for auxilliary texts, where
case should not be normalised avay.
> - The IETF should avoid inventing a new normalisation form,
> provided a technically sufficient one can be identified
> elsewhere (e.g. in an ISO standard) ?
Yes, but the only place to seek one is from the Unicode consortium.
ISO does not define any normal forms for text at all.
However, the normal forms currently defined by the Unicode consortium
do not do any kind of case mapping. And I think we all agreed that
we should continue to have the DNS "case insensitive", in some sense,
when matching. Such a 'caseless' normal form is not yet defined.
Kind regards
/kent k