Right now, our XPath implementation traverses a data structure much like a
syntax tree to evaluate its results.
In practice, an XPath expression can be compiled directly into a set of
function executions (with a few if/else statements mixed in).
Further, those function executions follow the form "get all child nodes with
the name prefix:localname". Those actually requests
(model.getChildElementByName()) can be compiled directly to the underlying
model.
For a bonus addition, the XPath evaluation could be rewritten on top of the
"Cursor" model, at which point it could be readily applied to Java objects....
and the meaning could be compiled down directly to Java operations on Java
objects.
Well, maybe most of the above. We should see what we can push on here.
Original issue reported on code.google.com by eric%tib...@gtempaccount.com on 7 May 2012 at 2:15
Original issue reported on code.google.com by
eric%tib...@gtempaccount.com
on 7 May 2012 at 2:15