Lists Home |
Date Index |
Thomas B. Passin wrote:
> If the default types are xs:anyType, etc., I wonder how XPATH can be
> extended in the future to allow other schema systems, like RELAX NG.
RELAX NG has the good sense not to augment the infoset;
it's used for validation (and documentation) only.
> it won't be so hard to do if the PSVI info includes just the schema
> namespace and type names.
It seems to me that a good strategy for infoset-augmenting processes
would be to simply add attributes with a distinguished namespace.
This keeps things entirely within the XML 1.0 data model.
For example, instead of adding a [type definition name] information
set item to an element, it could add an "xs:type" attribute.
(This is a gross simplification of what W3C XML Schemas actually
add to the InfoSet to describe type assessments, but hopefully
you get the idea.)
Infoset augmentors that need to annotate individual bits of text
(e.g., Regular Fragmentations) can wrap the text up in new subelements.
Those that need to augment attributes should really think twice
about wanting to do that :-), but HyTime-style paired lists would