I was assuming a deep-equal comparison applied to the Target element - but
perhaps I didn't examine the problem carefully enough.

Michael Kay
http://www.saxonica.com/ 

> -----Original Message-----
> From: Owen Rees [mailto:owen.rees@xxxxxx] 
> Sent: 15 July 2007 22:01
> To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
> Subject: RE:  comparing a part of the XML tree
> 
> --On 15 July 2007 18:15 +0100 Michael Kay wrote:
> 
> > Calling deep-equal on every element would do the trick. It could 
> > potentially be quite expensive, but with luck deep-equal 
> doesn't take 
> > very long when the arguments are obviously not equal, for 
> example when 
> > they are elements with different names.
> 
> For the example given, deep-equal would not find any pair of 
> items that match. The 'selection' element has two children 
> but does not itself appear in the 'original' so there is no 
> item in the selection that can be deep-equal to any item in 
> the 'original'.
> 
> The first child of 'selection' does seem to be deep-equal to 
> an item in 'original' but the 'selection' item's following 
> sibling - a text node - is not equal to the following sibling 
> of the item in the 'original'. If spaces were to be 
> normalised then the content of the 'selection' text node 
> would be a prefix of the content of the 'original' text node.
> 
> -- 
> Owen Rees
> =======================================================
> Hewlett-Packard Limited
> Registered Office: Cain Road, Bracknell, Berks RG12 1HN
> Registered No: 690597 England