Re: Near final draft of TAG finding on the Self-Describing Web

Thanks Noah, that addresses my specific concern about application/xml,
but alas, not the broader implications.  As I believe the two examples
I offered demonstrated, at least indirectly, one cannot assume that a
rddl:resource element inside any XML content declares a RDDL resource.
 Yes, the element exists in the DOM, but the containing language may
do any number of things to nullify its declaration.  For example, the
rddl:resource element might be inside the semantic equivalent of an
html:pre, or an atom:content/@type="text".  Or, also using Atom, a
feed may contain two or more items, each with its own rddl:resource,
and an Atom-unaware XML-consuming agent cannot know which of those -
if any - to use when processing a representation.  I'm sure there are
several more examples of commonly used language design constructs
which would similarly break such assumptions.

As the Authoritative Metadata finding says, external metadata trumps
internal metadata, and so the media type must first be consulted
before any meaning can be extracted from an XML document.

The rest of 4.2.3 concerning the self-descriptive value of giving
elements and attributes URIs is terrific.

Mark.

Received on Wednesday, 21 January 2009 06:56:28 UTC