[
Lists Home |
Date Index |
Thread Index
]
At 03:48 PM 6/7/2002 +0100, Henry S. Thompson wrote:
>Two points:
>
>1) There is no way to say in RNG "this schema must be
>deterministic", so there's no way for a normative requirement that you
>use schemas capable of reliable type assertions to be stated.
Determining determinism isn't exactly rocket science. It can, of course,
be less than obvious, as W3C XML Schema designers find out regularly.
>2) There is no notion in RNG of identifying the types that were
>assigned, reliably or not -- indeed the notion of type assignment
>doesn't occur in RNG. As you and many others so eloquently and
>repeatedly point out, RNG validation delivers a 1-bit result: valid or
>not, nothing else.
Sure thing, and that fits very well with my suggestions for painting types
as a separate process. RELAX NG schemas that contain type information
could be used as a basis for that, but I'm happy to see that RELAX NG
didn't fall into the "type systems everywhere" approach that seems so
painfully infectious at the W3C.
> > I've argued for a long while that type information should be 'painted'
> > as a separate process from validation.
>
>Certainly a possiblity -- should SOAP wait until you or someone else
>works that out?
I'm not sure the current "we must have everything NOW" approach is exactly
brilliant. There seems to be an awful lot of dreck coming out of that
system. Lack of interest in competing approaches could well be one reason...
Simon St.Laurent
"Every day in every way I'm getting better and better." - Emile Coue
|