RE: [idn] Re: alpha v0.3

[Karlsson Kent - keka <keka@im.se>]

> -----Original Message-----
> From: RJ Atkinson [mailto:rja@inet.org]
...
> At 18:26 02-02-00 , Karlsson Kent - keka wrote:
> >  
> >We should refer to the properties characters have 
> >in Unicode 3.0.0, rather than any older version. 
> 
> Disagree.  
> 
> Existing IETF standards are oriented to ISO 10646 
> already.   As this is an IETF activity, we should 
> continue to reference ISO 10646 rather than any 
> version of UNICODE.  If the differences are as
> minor as you indicate, then this ought not bother
> you.  If the differences are significant, than
> to reference UNICODE would be to increase probability
> of issues in interworking with existing IETF standards.

The problem with ignoring Unicode here is that 10646
does not have ANY character properties AT ALL (nor does
any other ISO document).  No general categories
(normative in Unicode), no case mappings (informative
in Uniocde), no normalisation forms, no nothing.  Thus,
without Unicode, and only 10646, we will have GREAT
problems in specifying which characters are allowed in IDNs,
and we would also have GREAT problems in specifying
which case mapping to use (which apparently most of
us want in one way or another), and we would have GREAT
problems in specifying normalisation.

And if we did battle these problems ourselves, we are sure to
come up with something that is incompatible with every other
implementation of 10646 on this planet.

Why do you want us to have all those rather huge problems?

There are some initial moves to make SC2/WG2
to also (in synchrony) specify character properties,
as well as the BiDi algorithm.  I don't want the IDN
effort to just hang around the several years its going
to take before those things appear in an ISO standard
(if ever).

		Kind regards
		/kent k