Re: Media typs and XPointer, XLink, XPath, and XSLT

 From: Paul Grosso <pgrosso@arbortext.com>
 > Asks Rick Jelliffe:

> >I am confused about a related question: I have heard somewhere that the
> >XPointer group was going to make XPointers only valid for pointing into
XML
> >documents. So, for example, it would not be legitimate for an XPointer to
point
> >into a WebCGM document or other structured data.  Is this true?
>
> No.

Now I really am confused!  The current TR for XPointer says in its abstract:
"This document specifies a language that builds upon the XML Path Language
(XPath), to support addressing into the internal structures of XML
documents."

That seems pretty clear that XPointers are only for XML documents. Do you
mean "No: Xpointers are not just for XML documents" or "No: XPointers are
not just for text/xml and application/xml documents (as labelled at the HTTP
client
side)" or "No: XPointers are not just for text/xml and application/xml
documents
(as labelled in the HTTP header of the response)"?

> >Also, there seems to be paradigm difference between the conventional idea
> >of a URL (which fetches some resource) and an XPointer (which points to
some
> >span in some resource but doesn't necessarily return anything).
>
> XPointer doesn't necessarily point to a span.  It addresses things
> (where things might be many of the objects in an XML tree or possibly
> a "span" of such objects).

(I was using span treating WFness are a degenerate case of spanning.
And, to emphasize, I am trying to figure out how XLink and XPointer
fit wrt the "conventional idea"; I am well aware that the analytical idea
of XPointer and perhaps XLink in W3C/ISO/hypertext circles is
different--Eliot Kimber was always very strong on  this from the
earliest days.)

> >I had a bit of difficulty with this at first, and I suspect that others
with HTML
> >expectations will be the same. Murata-san's question relates to this too:
> >does an XLink always return a resource.
>
> "Return" is a semantic, and XLink's semantics are not finalized.
> I don't really know how else to answer your question, as it doesn't
> really make sense to me.  XPointer addresses, and XLink (which may
> use XPointer) links addressed things.  What is means to link addressed
> things is yet a different question.

The question comes down to whether it makes sense to think of making
a link without also having some traversal resource for it. If not, then
there
is always a "return" or "fetch".

I know XPointers can be used in non-traversal purposes (e.g., naming
comparison, queries, I suppose even namespace URIs): "whether user
'traversal' is the purpose of an XPointer at all, is application-dependent."

In XLink current public draft, traversal is "The action of using a link;
that is, of accessing a resource."

My current answer to my question ("Does XLink always return a
resource?") is that traversing a link always involves accessing a
resource, and that accessing implies some kind of fetching; however
there are non-traversing uses of links (which are really uses of the URI
or of the attributes of the link.)  Taking the pointer and sending it
in a query is a use of the link, but it is not a traversal of the link.
But I am not satisifed that "traversal/access" does mean HTTP GET,
and that other uses are non-traversal uses.

 Rick Jelliffe

Received on Wednesday, 14 July 1999 00:38:19 UTC