Re: gcj 3.0/unicode

[David Brownell <david-b at pacbell dot net>]

> >      And I seem to remember that when I used the
> > 2.x tables I found errors in them.  OTOH the online docs pretty clear
> > say "Unicode 2.0" (something they didn't say before JDK 1.3, at least
> > as far as I remember).  Maybe we ought to downgrade.  Comments?
> 
> I'd say "downgrade".  I seem to recall the Unicode revision being part
> of the JLS, though that might be wrong.  If it's in the language spec,
> it's expected that apps will rely on it.  And Javadoc specs 2.0 any
> place it says more than "Unicode".

Hmm, I just saw Cedric's note (1.4 switched to 3.0, but not 3.1,
from 2.0) ... maybe "downgrade to 3.0" then!

I think this particular software is atypical in its requirements, since
it wants to achieve conformance against a function of the 2.0 tables,
while most applications will probably be content to leave all such
details toUnicode libraries.

- Dave

> On the other hand, that logic would say that applications can't ever
> rely on java.lang.Character to baseline their own Unicode conformance,
> since they couldn't expect that behavior to be stable.  I don't expect
> many would want their own copies of the Unicode tables, or get them
> very consistently right/fast if they did choose to incorporate them.