[
Lists Home |
Date Index |
Thread Index
]
Tim Bray wrote:
>...
>
> Paul, you're about the fifth smart person to fall into this trip which
> means it's a nasty one and RDDLers have to be careful. There's no such
> thing as a RDDL for a document. There's a RDDL for a *namespace*.
Actually I'm not convinced that the RDDL specification is technically
limited that way**, but I understand that namespace explanation is the
central point of RDDL. Let me rephrase my point:
* if I encounter a document "out of the blue" with only one namespace,
I would confidently look up the RDDL and use the schemas, stylesheets,
classes, etc. specified there.
* if I encounter a document o-o-t-b with only one namespace
represented on the root element, but others inline, I would proceed to
the RDDL with about 80% certainty (not sure what to do in code about
this confusion, but...). I would have NO IDEA what to do about the
schemas, etc. for the inner namespaces.
* if I encounter a document o-o-t-b with a namespace on the root and
another namespace on an attribute then my certainty about the
applicability of the RDDL has to drop to about 60% because of the XSLT
precedent.
I find this personally an uncomfortable situation but I think that RDDL
only exposes it, not causes it.
Paul Prescod
** on thinking about it, I kind of wonder whether documents should be
able to suggest RDDLs as they can suggest schemas and stylesheets.
|