[
Lists Home |
Date Index |
Thread Index
]
Jeff Rafter wrote:
> I think that there is a middle ground-- or more importantly, that item
> (2) can be used in conjunction with item (1). Instead of considering
> single languages within a single validation why not consider multiples?
> Then as the validator proceeds it will switch between the languages it
> knows (most likely based on namespace). This way it can switch
> validation modes internally as it encounters languages.
Agreed.
> Currently the only way that this is possible is by having a bunch of
> <any> elements in XML Schema. I wonder if there will be a grammar that
> allows you to define a compound document on top of a (name your
> favorite) schema. So you could override a <p> (or flow.mix) and say that
> it allows <any> element from MathML.
This is one possibility. It gives more control over the shape of the
document and could easily be built on top of XML Schemas -- just say
where an <xs:any> element is allowed without having to modify the
underlying (possibly standard) schema.
The other possibility is to rewrite validators / redefine validation so
that you don't have to define the overall shape of the document.
Instead, the shapes of each fragment are defined by a schema and the
validator validates each fragment, regardless of where it occurs. This
allows lots more flexibility.
In addition to validating compound documents constructed from an
arbitrary set of schemas, it would be useful when you're interested in
only a subset of a document and don't care about anything except that
subset.
-- Ron
|
