- From: Jonathan Robie <jonathan.robie@softwareag.com>
- Date: Tue, 08 Jan 2002 15:25:52 -0500
- To: "Vun Kannon, David" <dvunkannon@kpmg.com>, "'Kay, Michael'" <Michael.Kay@softwareag.com>, www-xpath-comments@w3.org
At 12:55 PM 1/7/2002 -0500, Vun Kannon, David wrote: >After rereading the WD, I see that "element of type $foo" is not correct. >However, issue 197 uses this kind of syntax, so I may have been misled >there. I agree that it would be better to have a data model in which types >and functions were first class objects. > >I agree that evaluate() may be more appropriate in the function set of XSLT >than core XPath, but won't XQuery need it also? Xlink semantics could also >be a separate function library. It is certainly useful in queries, but there are difficulties in supporting it with a static type system. The question, for me, is whether it belongs to the minimum required to declare victory. I suspect that it does not. >Here is a use case for testing facets. Find all the elements whose type >looks like a US Social Security number (123-45-6789). This is in distinction >to any text in the document which happens to look like an SS#. Searching >over a heterogenous collection of documents, what the schema author called >it is less important than the 'shape'. Do you think that belongs to the minimum required to declare victory for XPath 2.0? Jonathan
Received on Tuesday, 8 January 2002 15:26:07 UTC