[
Lists Home |
Date Index |
Thread Index
]
Oops pressed wrong button...
To continue: perhaps, like RDF, interoperability plays no part in XQuery.
It may be a technology for centralized databases where the query can be
quite happily tailored to cope with the idiosyncracies of the particular
software. (I am not sure whether XQuery still has that rule that allows an
implementation to fail in the name of optimization, but that rule alone
suggests that interoperability is not a central concern: no reflection on
XQuery b.t.w.)
But XML Schemas are being promoted so heavily and used by organizations
who requir interoperability, that there is really strong need for
implementation issues and best practise knowledge to escape xmlschema-dev.
For example, this dangling type issue. That is definitely one of these
issues which would prevent interoperability. But where is there any
website or articles saying "If you follow XML Schemas and allow dangling
type references, your schemas are ticking bombs w.r.t. interop"? (The W3C
forum Henry mentioned before sounds like a prudent idea.)
XML Schemas implementations should be reasonable mature by now, in 2005.
If there are major conformances issues (not in some silly percentage, but
in their impact ) then surely it is evidence that the technology is just
too damn complicated. I don't see why the normal approach would not
warranted: make a profile or conformance level and have a large push to
make that the bottom level supported everywhere: no xsi:type, no
wildcards, no recursion,
and no dangling references would seem to be the kind profile that
databinding tools would work well with. That work could be done
independently of XML Schemas development: profiling is easier than
invention.
Cheers
Rick Jelliffe
|
