follow your nose from XML documents to namespace documents? xmlFunctions-34, nsMediaType-3, RDFinXHTML-35

Consider this document, with an application/xml representation
that bears the RDF namespace name on its root element:

$ curl -I
HTTP/1.1 200 OK
Last-Modified: Wed, 01 Nov 2006 07:01:41 GMT
Content-Type: application/xml; qs=0.9

$ curl -q
<!-- ... -->

That example comes from the GRDDL test colletion,
but suppose the RDF statements weren't just
about GRDDL dataview:namespaceTransformation stuff, but were
statements about drug interactions, or privacy policies,
or mobileOK labels, or whatnot; i.e. something with non-trivial
impact. Clearly, if we had a representation with 
  Content-Type: application/rdf+xml
we could follow our nose thru the IANA registry to the RDF spec,
which would make it clear that we can attribute the RDF statements
about privacy policies and mobileOK claims to the
author/publisher/owner of the document.

But in the case of application/xml, has the author issued
a privacy policy, or just said "look at these tags"?

Analagously, suppose the namespace were ;
can I attribute a picture with circles and lines to the author,
or just a bunch of elements and attributes?
(Hm... fragmentInXML-28 is also sort of relevant...)

The answer is not entirely clear from the ratified specifications.
so the GRDDL WG has asked me to take this up with the TAG.

This is clearly relevant to issue nsMediaType-3; our issues
list says that was subsumed by mixedNamespaceMeaning-13 which
was then split, but it's not clear whether xmlFunctions-34
covers this question; if not, we should please re-open nsMediaType-3. 
The GRDDL WG does not plan to wait until the TAG resolves these
issues before making its decision, but welcomes any input
(to as soon as it is available.

Here's one way to phrase the question:

Which law makes for a better society?

  (1) if you use application/xml and the RDF (or SVG) namespace name,
  you're accountable for the RDF statements (or circles
  and lines...) therein.

  (2) if you want to hold somebody accountable for RDF statements
  (or attribute a graphic to them),
  you have to find a representation labelled application/rdf+xml
  (or image/svg+xml)

I suggest (1), partly because it's more practical; there is quite
a bit of RDF and OWL with representations labelled application/xml
but no representation labelled application/rdf+xml, and I in most
of the cases I have seen, the authors meant it. (I haven't tested
to see what happens in the case of the SVG namespace in the XML
media type.) But also because the XML media type specification says:

   An XML document labeled as text/xml or application/xml might contain
   namespace declarations, stylesheet-linking processing instructions
   (PIs), schema information, or other declarations that might be used
   to suggest how the document is to be processed.  For example, a
   document might have the XHTML namespace and a reference to a CSS
   stylesheet.  Such a document might be handled by applications that
   would use this information to dispatch the document for appropriate

Now "might" and "suggest" are not very strong words, but this
paragraph seems to be enough notice to those choosing the
application/xml media type that it is somewhat common practice
for applications to do this sort of dispatching, and that
the resulting processing is "appropriate". Since they
have the option to choose a more opaque media type (e.g. text/plain
or application/octet-stream) and they chose this media type,
they accepted the consequences of such dispatching.

As a result, it is best for the GRDDL WG to specify that
the GRDDL result of
is not empty but has some RDF statements in it.

Dan Connolly, W3C
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Tuesday, 14 November 2006 17:37:18 UTC