> I use our selectNodes method > extensively - more often than virtually any other DOM call. In fact, > selectNodes makes a number of DOM calls redundant, results in more robust > code, and performs better in general. > I agree and I think most MSXML users fit into this category (I bet Oracle users will follow suit). Ever since selectNodes was introduced, I hardly ever use any of the other hierarchical members (of Node) for traversal. > I observe that one person's feature is another person's bloat, and that in > my experience the standardization process leads to a union of everyone's > features rather than the intersection. > Agreed. > There are several directions for XPath APIs to take: > - minimal, convenience-oriented (e.g. Microsoft/Oracle selectNodes) > - complete (e.g. full context initialization, matching, etc.) > - performant (complete + compiled queries, streaming support, etc.) > I'm starting to believe that a combination of *minimal* and *performant* should be the scope of this work. I think a *minimal* solution that covers 90+% of the use cases is a good first step, while looking forward to a more long-term, complete solution that ties in with the work of XML Query. -aaronReceived on Friday, 5 May 2000 05:29:15 GMT
This archive was generated by hypermail 2.2.0 + w3c-0.30 : Friday, 25 March 2005 11:19:25 GMT