What are your particular requirements? Using XSLT for this sort of thing
is >really a pretty good idea...
That would be my own recommendation; write an XSLT stylesheet which uses
the document() function to access the documents to be merged (perhaps
getting the URIs from a top-level document, if they might change) and
appropriately recombines the data from those.
If you're only using the XPath API, you'll have to hand-code the second
half of that problem. In my experience, the inefficiencies are actually in
the repeated XPath lookups (and in the possibility that the XPath might
reference ancestor/preceeding nodes) more than in the generation side of
things, so I _think_ you'll find that coding your own generator isn't as
much faster as you might expect.
But it'd certainly be interesting to try both approaches and compare the
results...
______________________________________
Joe Kesselman / IBM Research
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list