Yes. Definitely. The more of this that can be generalised, the better.
Essentially quite a lot of JUMBO is involved in this sort of processing
and I'd be more than happy to try to migrate JUMBO's ideas towards an API.
My Node class (== Element, more or less) has nearly 100 member functions,
and I'll try to post them as a javadoc API (just needs locating on the WWW).
>
> Now the question remains if this approach should substitute
> the event base stream that built the bottom layer of
> our XAPI-J discussion. I think the event based approach
> should still form the base. Many people, I believe,
> feel still very comfortable with it.
I feel very comfortable with the event stream API and I would certainly
not substitute it. Essentially JUMBO can consume Elements from either an
NXP-like event stream, or a Lark-like tree structure. It may be useful to
identify those objects such as Element and Attribute that are relevant to both
environments, and there could be an Element/Attribute Factory sitting on top
of both (but leaving them exposed as well for those who need the lower level).
P.
-- Peter Murray-Rust, domestic net connection Virtual School of Molecular Sciences http://www.vsms.nottingham.ac.uk/