RE: issue 168 proposal: xsi:type of external references in Enco ding

Noah,

> Does this make sense?

What you suggest make sense to me... I think it also highlights a need to be
(very) sure what base URI is in scope. 

It's not clear to me how to establish a reference to a SOAP envelope as a
base URI, particularly if some outer scope (in the XML document or the
protocol binding) has established a base URI that is not a reference to the
SOAP envelope.

Regards,

Stuart
BTW I also found Andrew's analysis really helful.

> -----Original Message-----
> From: noah_mendelsohn@us.ibm.com [mailto:noah_mendelsohn@us.ibm.com]
> Sent: 15 December 2001 03:15
> To: andrewl@microsoft.com
> Cc: xml-dist-app@w3.org
> Subject: RE: issue 168 proposal: xsi:type of external references in
> Encoding
> 
> 
> 
> As several have noted, your analysis is indeed right on the 
> mark, and very
> helpful.
> 
> >> 1a leads naturally to the idea that external and internal links should
> be syntactically distinguished,
> 
> The tension seems to be this:  from a pure web architecture point of view,
> efficiency and middleware architecture is not an issue, therefore do not
> distinguish one link from another.  In practice, it's often nice to deal
> efficiently with the special case of the closed graph that originated with
> the sender, was intended to be "part of the message", and always travels
> with the message.  Certainly, one will often wish to create middleware
that
> recreates the closed subgraph either in advance of invoking a receiving
> application, or in some optimized manner.
> 
> Which in turn leads to the further question, if these links are
> distinguished, distinguished how?  The IDREF suggestion essentially says:
> only the potentially open graph references are true web-references; the
> others use mechanisms private to XML.  I wonder whether there might be
> another heuristic that might be applied, one which would differentiate
> local references while unifying all references into a single URI-based
> graph?   The rough idea would be:  "all graph references are represented
as
> href attributes which are URI references;  references to multiref targets
> within the envelope SHOULD be relative URI's using fragment identifier
> syntax and starting with the "#" (I.e. #someid).  These envelope-relative
> references thus form a subgraph which is always known to be carried with
> the message, and which in many cases can be decoded without further
network
> traffic upon receipt."
> 
> Does this make sense?
> 
> ------------------------------------------------------------------------
> Noah Mendelsohn                                    Voice: 
> 1-617-693-4036
> Lotus Development Corp.                            Fax: 1-617-693-8676
> One Rogers Street
> Cambridge, MA 02142
> ------------------------------------------------------------------------
> 
> 
> 
>

Received on Monday, 17 December 2001 08:38:51 UTC