Casefolding Sigma (was: Re: IDNAbis PreprocessingDraft)

[Erik van der Poel]

On Jan 25, 2008 2:45 AM, Martin Duerst <duerst at it.aoyama.ac.jp> wrote:
> It may be true that in a world where there
> are only U-labels and A-labels, confusion would be reduced
> with IDNA200X. But then we have 'localized software', which
> introduces another layer, with something that I'll call
> L-labels here for convenience.

So now we have 4 kinds of labels, namely, A-labels, U-labels, V-labels
and L-labels. V-labels are variant labels that conform to the fixed
rules in IDNA2003, while L-labels follow the rules of a particular
local community. (L-labels are processed using locale-dependent rules,
while V-labels use fixed rules.)

The browser developers (and others interested in HTML) may wish to
extend V-labels to a 200X version using a new set of fixed rules that
follow the model of the original IDNA2003, i.e. case-folding, NFKC,
etc.

So we might even use the names V2003-labels and V200X-labels to
distinguish those. V200X-labels are handled differently by MSIE 7 and
Firefox 2. The former refuses to resolve such domain names, since they
contain characters that are unassigned in the version of IDNA that
MSIE 7 implements.

Erik