Re: XML Blueberry

[David Brownell <david-b@pacbell.net>]

> > It almost feels to me like those tables should be the responsibility of
> > the Unicode folks (or some similarly lucky but separate intermediary
> > with time on its hands) to maintain, and that some means of
> > incorporating them by reference might have avoided this entire
> > discussion.
> 
> In essence they are:  Unicode sets the character properties, and
> XML has rules saying which types of characters can be used for what,
> and what the exceptions are.

That's not what Appendix B says ... it says normatively "these are the
legal name/namestart/... character codes" (paraphrased) and everything
else is vaguely explanatory.  (It's weak on explanations for _why_ those
exceptions to Unicode tables should have been made.)  The Unicode
tables (which change as Unicode evolves) aren't normatively referenced;
the result of applying an algorithm (mentioned, non-normatively) to those
tables is normative.

On any platform with a decent Unicode library, it'd certainly be easier
to use that directly -- java.lang.Character primitives are sort of in that
category.  Then when the platform upgrades its level of Unicode support,
the documents accepted/generated as legal XML would change; not a
good thing.  Which is the point you made in a later followup.

- Dave