XOM API for Processing XML with Java

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

Wolfgang Hoschek wrote:

> Everybody already understands XPath's adjacent text semantics since the 
> beginning of this thread.
> But that's not the question. The question is why would output 1 *always* 
> be the "right" thing, as opposed to output 2 or 3, or even output 4 
> (which is returning the first Text node as DOM XPath seems to do, as 
> noted by Mike). In other words how to map XPath results on output to XOM 
> nodes.

Because an XPath query has a clear meaning when operating over an XML 
document. There's no ambiguity as to how contiguous sequences of 
characters in PCDATA should be treated in a statement like 
//data/text()[2] or //node()[17]. Either the expression matches all of 
the characters as a group or it matches none of them. The incidental 
fact that these sequences of characters may currently be represented in 
1, 2, or more XOM Text objects doesn't mean anything more than that they 
may be represented in 1, 2, or more bytes. The XOM object model is an 
implementation detail on which the XPath engine sits. The object model 
boundaries should not affect the results of an XPath query.

--
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