[
Lists Home |
Date Index |
Thread Index
]
> From: DuCharme, Bob (LNG) [mailto:bob.ducharme@lexisnexis.com]
<snip/>
> But I understand if you want a more helpful answer than that,
> and what Tim
> Bray said at
> http://lists.xml.org/archives/xml-dev/200201/msg01403.html will
> take you far. If you know the name of the root element, and
> you know the
> namespace that it's from, then you know a hell of a lot about
> the potential
> processing that you can do with that document.
I would think this to be true in the vast majority of cases. But if you take
XSLT as an example, the literal result as stylesheet approach really breaks
the mold on this one. I would think this to be a rather rare exception, but
it's one to consider nonetheless.
I would think that in this case, you'd have to fallback onto some sort of PI
if you needed to identify the "document type" (using this term very
loosely).
My preference is for a PI that associates a set of links to resources with
the document instance. This set of links could be thought of as the
document's "nature", which could be orthogonal to the DOCTYPE or root
namespace, or it could be explicitly tied to the root namespace or DOCTYPE
and resolved via RDDL or a linkbase. Use of a PI would enable an instance to
override the default resolution based on DOCTYPE or root namespace, and
associate a different nature with the document that matches its intended use
rather than its declared DOCTYPE or root namespace.
|
