[
Lists Home |
Date Index |
Thread Index
]
Dare wrote:
> The advice on avoiding chameleon schemas could be fleshed out.
I don't wish to enter the debate about the man vs the matter, but
on matters of fact, the example given of chameleon schemas is
factually incorrect, I believe.
Error 1: Reference to person type in main schema.
The master schema makes reference to
<xs:element ref="person" maxOccurs="unbounded" />
refs are treated as QNames, and without a default namespace,
"person" is a reference to {None}person.
Having no namespace, the definition of person in the included
schema is coerced into the http://example.com namespace. Thus
the validator should complain about the reference to person
in the including schema and XSV and MSXML do.
Error 2: Assertion that ref="familyName" should explictly placed in
the http://example.com namespace.
Any ref with a {None} namespace in a chameleon schema will be coerced
into the namespace of the includer. Thus there is NO need to explicitly
place familyName in the target namespace. Indeed, if this assertion
were correct, chameleon schemas would be pretty useless. See
http://lists.w3.org/Archives/Public/xmlschema-dev/2001Oct/0123.html
With error one above corrected, both XSV and MSXML validate the included
schema without change to the included schema.
The author's other points may or may not be valid, and the use of chameleon
schemas may or may not be bad, but factual errors tend to cloud the issue
and lead to suspicion about the validity of the other arguments put forward
by the author(as distinct from suspicions about the author themselves).
Regards
Michael
|
