Yes, although the context node upon which the XPath is evaluated is in another
XML document altogether. Upon first glance, dyn:evaluate seem to use the 
current context,
and thus the same XML document, when evaluating the XPath.

Otoh, I might very well be missing something. I'm quite new to XSLT... ;)

/Rob

> -----Original Message-----
> From: Mark Weaver [mailto:[EMAIL PROTECTED]
> Sent: Monday, April 28, 2003 12:21 PM
> To: [email protected]
> Subject: RE: Best way to create a XalanNode inside an 
> extension function
> 
> 
> Sounds a bit like 
> http://exslt.org/dyn/functions/evaluate/index.html?           
> 
> > -----Original Message-----
> > From: Bielik, Robert [mailto:[EMAIL PROTECTED]
> > Sent: 28 April 2003 06:32
> > To: [email protected]
> > Subject: RE: Best way to create a XalanNode inside an 
> extension function
> > 
> > 
> > I don't know if this is pertinent to the original question, but I 
> > have scenario where I have
> > an XML with data to be transformed by XSL. However, in order to 
> > do this the XSL may need extra
> > information which I didn't want to import into the main XML due 
> > to having several XSL:s each needing
> > its own extra info.
> > 
> > So I solved it this way: For each XSL I create a new 
> > XalanDocument that contains the info that the
> > particular XSL need. Then I provide an extension function 
> > "myInfo(...)" that takes a string as argument.
> > This argument is then evaluated as an XPath in the new 
> > XalanDocument (external to transform) and returns a result tree 
> > fragment. And it seems to work nicely, not having to copy the XSL 
> > specific data into the main XML.
> > 
> > I've only tested returning nodesets, and that seems to be working 
> > nicely. However, I see no reason (yet)
> > why the function couldn't return an entire tree fragment that in 
> > turn would be XPath-able.
> > 
> > If this is different from creating XalanNodes from scratch, I 
> > apologize for the digression.
> > 
> > Note: The XObjectPtr returned from the XPathEvaluator must be 
> > cloned upon returning it from the 
> > extension function (but you all know this kind of stuff 
> already... :)
> > 
> > Regards
> > /Rob

Reply via email to