RE: xinclude fails to use URI syntax for referring to resources?

Perhaps my second draft is better?

<note>
	<p>A key feature of XInclude is that it allows a resource to be
cast to a user-specifed type for inclusion (XML or text).  The returned
media type is therefore essentially ignored for the purposes of
inclusion processing, and the syntax of the fragment identifier of the
returned media type will generally not be applicable to the
user-specified type.  For <att>parse="xml"</att> inclusions,
sub-resources are identified by a separate <att>xpointer</att>
attribute, which is applied after the casting takes place.  While this
does not prevent subresources of XML documents to be identified by URI
(See <loc
href="http://www.w3.org/TR/webarch/#identification">Architecture of the
World Wide Web [Identification]</loc>), it does limit the use of these
identifiers directly within XInclude.</p>
</note>

When I describe the problem that way, there is an improvement we could
make.  We could say that if the returned media type is actually
application/xml or text/xml, the fragment syntax is interpreted as an
xpointer, and a separate xpointer attribute is not necessary.  Resources
returned with other media types rely on the xpointer attribute.

> -----Original Message-----
> From: Dan Connolly [mailto:connolly@w3.org]
> Sent: Wednesday, January 07, 2004 10:35 AM
> To: Jonathan Marsh
> Cc: www-xml-xinclude-comments@w3.org; Ian Jacobs
> Subject: RE: xinclude fails to use URI syntax for referring to
resources?
> 
> On Wed, 2004-01-07 at 12:23, Jonathan Marsh wrote:
> > The Core WG decided to add some explanation about why the split was
> > made.
> 
> Good...
> 
> >   Here is my first draft of this text (suggestions welcome):
> >
> > <note>
> > 	<p>For interoperability, fragment identifiers <termref
> > def="dt-must">should not</termref> be used; the interpretation and
> > application of media-specific fragment identifiers in creating
> > info-items is not guaranteed to be supported across implementations.
> > For <att>parse="xml"</att> inclusions, sub-resources are identified
by a
> > separate <att>xpointer</att> attribute instead of by a fragment
> > identifier.  A separate attribute is necessary because a fragment
> > identifier is interpreted according to the media type of the
returned
> > resource, while the <att>xpointer</att> attribute is applied after
the
> > resource has been cast to application/xml.</p>
> > </note>
> 
> Hmm... no pointer to nor excerpt from
> http://www.w3.org/TR/webarch/#identification.
> 
> Can I please have one? Perhaps I'll find time to craft a
> specfici textual selection presently.
> 
> 
> > Sorry for the delay in responding
> 
> Understandable; no harm done.
> 
> 
> > > On Mon, 2003-11-10 at 17:16, Dan Connolly wrote:
> > > > Regarding...
> > > >   http://www.w3.org/TR/2003/WD-xinclude-20031110/
> 
> > > After discussion with the TAG
> > >   http://www.w3.org/2003/11/15-tag-summary.html#xincl
> > > I've changed my mind...
> 
> > > > "A URI SHOULD be assigned to each resource that is intended to
be
> > > > identified, shared, or described by reference."
> > > >
> > > >   -- http://www.w3.org/TR/webarch/#identification
> > > >
> > > > If a URI has been assigned to a resource, Web agents SHOULD
refer to
> > the
> > > > resource using the same URI, character for character.
> > > >   -- http://www.w3.org/TR/webarch/#identifiers-comparison
> > > >
> > > > which is basically: to refer from one thing to
> > > > another in the Web, use URI reference syntax.
> > > >
> --
> Dan Connolly, W3C http://www.w3.org/People/Connolly/
> 
>

Received on Wednesday, 7 January 2004 14:18:57 UTC