> Would an attribute or element be preferred over a PI or easier to
implement than a PI? If so, why?
See my message at:
http://lists.oasis-open.org/archives/relax-ng/200109/msg00110.html
I've been thinking about a catalog-based approach, and I think that would be
more powerful than a PI although significantly more complicated. The
catalog could contain rules like:
- if the root element has namespace URI X and local name Y, use schema S
- if the root element has an attribute with namespace URI X, local name Y,
value Z, use schema S
- if the root element has an attribute with namespace URI X, local name Y,
value Z, use schema Z
- if the prolog has a processing instruction with target X and data Y, use
schema S
- if the prolog has a processing instruction with target X and data Y, use
schema Y
- if the prolog has a processing instruction with target X, pseudo attribute
Y with value Z, use schema Z
- if the document has a DOCTYPE with public ID X, use schema S
etc
James
