I've been implementing support for none Node XPath results, which I have working to an extent.
My sticking point is this. If I query 'count(//foo)' on a collection with say 5 documents should there be 5 result nodes or should there be one aggregated node? Unfortunatly the spec is a little unclear: "Run an XPath query against the Collection. The XPath will be applied to all XML resources stored in the Collection. The result is a ResourceSet containing the results of the query. Any namespaces used in the query string will be evaluated using the mappings setup using setNamespace." Thoughts? -k.
