- From: Daniel Veillard <daniel@veillard.com>
- Date: Wed, 10 Apr 2002 10:40:09 +0200
- To: Jeni Tennison <jeni@jenitennison.com>
- Cc: Rahul Srivastava <Rahul.Srivastava@Sun.COM>, xmlschema-dev@w3.org
On Wed, Apr 10, 2002 at 07:42:08AM +0100, Jeni Tennison wrote: > > 2 Each node in the ·target node set· is an element node among the descendants of the > > context node. > > > > .... > > ]] > > > > Looking at point 2 above, it reflects that, we cannot use '.' as the > > XPath expr. for a selector. Does *descendants of the context node* > > includes the context node? > > Good point. I'd missed that. Usually, "descendants of the context > node" doesn't include the context node, so I guess that would be the > intention here. Right, otherwise I would have expected them to use the XPath name "descendant-or-self" in some way. > > On the other hand the production rule for the XPath expr. for the > > selector allows '.' as a valid XPath expr. > > http://www.w3.org/TR/xmlschema-1/#coss-identity-constraint > > Yes. This leaves us in a position where it's valid for a schema to > contain a selector that has the xpath ., but it's not valid for an > instance to contain an element that uses that identity constraint. The XPath expresssion in the SchForSch is actually very broken, like in accepting "a:1", it seems it wasn't defined and tested with the idea that the expression would really be the normative definition. http://lists.w3.org/Archives/Public/www-xml-schema-comments/2002AprJun/0005.html I would give more trust to the written section over the expression itself. > I'm not sure whether the intention of the WG was to allow identity > constraints to select the context node or to disallow it. None of the > implementations that I tested (XSV, MSXML, Xerces-C++, Xerces-J) > complained about using <xs:selector xpath="." />, either within the > schema or when there was an element in the instance document using the > identity constraint. > > You really need confirmation from W3C. Messages to > www-xml-schema-comments@w3.org might elicit an official response. Right, maybe Henry Thomson could hint here on the expected behaviour, but 2/ seems to very clearly disallow '.' Daniel -- Daniel Veillard | Red Hat Network http://redhat.com/products/network/ veillard@redhat.com | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/
Received on Wednesday, 10 April 2002 04:40:18 UTC