Re: XPath API start

jeroen@tcf.nl wrote:
> Below you can find an XPath API as we've been implementing

Cool.  This isn't so far from my thoughts, and much closer than having
methods directly on the Node object.

> - the interfaces should be simple and straight forward

+1.

> - it doesn't make too much sense to offer a complete API for parsed XPath

> queries although this is possible

What might you add if it was a "complete API"?  Interfaces for the
structure so that tools could manipulate it?

> - it should cover the complete XPath implementation

Should this include function extensions?  (I would think yes).

> - every query object has an authority object which organizes authority on

> queries. We've created it to serve our repository but it might easily be
skipped

Yeah, I would think this should belong to a full query API.

> - our product is a Java product so the API's are in java which is alright
for
> discussion purposes.

I think Java is fine.  But we should make sure a C++ API of some sort is
defined before we freeze, IMHO.

> - XPath queries should support the context supported by Xpath (the
context node,
> a set of variable bindings, a set of namespace declarations)

+1, adding extension bindings.

> and a clear module (something like
> org.w3c.dom.xpath) should be created.

Why org.w3c.dom.xpath and not org.w3c.xpath??  (as per my previous mails).

> XPathQuery

I don't think this object can be multi-threaded without locking, is that
correct?  Might it be better to have a simpler XPath object to which a
XPathContext is handed in?

> XPathResultIf

I don't understand the "If" postfix in the names.  I assume it has some
significance?

-scott