XOM API for Processing XML with Java

[Elliotte Harold <elharo AT metalab.unc.edu>]

Michael Kay wrote:

> I've raised a note with the WGs about whether we should relax this
> constraint. It's appropriate that a data model constructed by parsing raw
> XML, and a data model constructed by XSLT or XQuery should have conjoined
> text nodes, but I can't see a strong reason why a data model constructed
> programmatically should have to obey this constraint.

Is there a reference so I can respond?

I'm not convinced XPath is wrong here. I'm not doing this solely out of 
conformance with the XPath spec. I'm about 70% convinced that XPath made 
the right decision to disallow adjacent text nodes.

The fact is that the XML structures in the document are the only 
reliable ones XPath can ever see. Object boundaries are just a little 
too contingent for my comfort. For instance, would you want an 
expression like count(div//node()) to return different values depending 
solely on how many Text objects made up the exact same content? Treating 
contiguous Text objects as part of the same text node just seems more 
reproducible and reliable.

--
Elliotte Rusty Harold  elharo AT metalab.unc.edu
XML in a Nutshell 3rd Edition Just Published!
http://www.cafeconleche.org/books/xian3/
http://www.amazon.com/exec/obidos/ISBN=0596007647/cafeaulaitA/ref=nosim