[
Lists Home |
Date Index |
Thread Index
]
> -----Original Message-----
> From: Elliotte Rusty Harold [mailto:elharo@metalab.unc.edu]
> Sent: Friday, December 07, 2001 7:24 PM
> To: xml-dev@lists.xml.org
> Subject: RE: [xml-dev] Attribute Order
>
>
> So far Schematron's the most powerful language, but it still hasn't
> reached Turing completeness. Using actual program code lets your
> schemas verify criteria of types one and two. Type three is still an
> area of active research. But existing schema languages seem mostly
> stuck at Type 0.
Sorry, I'm not familiar with this categorization ... what "criteria of
types" being referred to here?
Anyway, I completely agree -- it's perfectly reasonable for schema designers
to specify constraints that can't be expressed in an existing XML schema
language. We went around on this a couple of months ago, and as I recall we
concluded:
- schemas are certainly useful to define "contracts" constraining the
markup that a receiver is expected to understand;
- schema validators or validating parsers may are likely to be very useful
in debugging or auditing software to ensure that the contract is being
adhered to;
- XML-level validation may, in some (limited?) circumstances, suffice as the
"business logic" to ensure that an XML document or message should be
accepted by some receiving process.
- XML-level validation is more likely to be useful as a cost-effective way
of doing a coarse-grained screening out of blatantly unacceptable documents
or messages so that program code can check the constraints that can't be
expressed in a schema language. Whether it makes sense to do both the
schema validation and the programmatic validation is something that
individual designers must determine for their own environment.
- [this is more my personal understanding than a consensus of the list] Most
"real world, real-time" XML document/message validation today involves
program code or human intelligence rather than XML DTD or schema validation.
|
