RE: Comments on XPTR Framework

> From: Keith W. Boone [mailto:keith@woc.org] 
> Sent: Friday, October 18, 2002 6:48 AM
> To: Ron Daniel
> Cc: www-xml-linking-comments@w3.org; w3c-xml-linking-wg@w3.org
> Subject: RE: Comments on XPTR Framework
> 
> Sorry, you missed my intent on that one, which was to have 
> the spec clarify
> that the variable bindings and function library should be 
> considered as part
> of the context, without necessarily specifying how they get 
> updated.

Thanks for the clarification. The problem is that variable
bindings and function library are specific to XPath, and schemes
closely related to it like the xpointer() scheme.
Other schemes that may come along in the future will need
different information in the context.

The group talked about this, and decided the best course of
action was for the FRAMEWORK spec to not say anything that
ruled out the XPath variable bindings and function library,
but not to say anything that priviledged XPath-related schemes
over any other schemes which will need different information
in the context.

The XPOINTER() scheme draft, on the other hand, would say
more about those specific items in the context (although we
had decided to rule out the use of variables, and particularly
extension functions, in the interests of interoperability).
We would be more inclined to make normative statements about
handling XPath evalution context items in that spec than in
the framework spec.

> Motivation for the comment was that I plan on having a number 
> of canned
> XPath queries, where parts of the query are supplied by 
> variables stored in
> the execution context, and I'd like to see the spec at least 
> nod in the
> direction that XPath implementations have already headed.

I can understand that, but this seems more appropriate
in documents other than the framework spec, which needs
to apply to all pointer schemes.

Best regards,
Ron Daniel
Acting chair, XML Linking WG

Received on Tuesday, 22 October 2002 11:49:11 UTC