> > >        <etp id="200">
> > >            <![CDATA[Trying accessing only this]]>
> > >            <nvgs>
> 
>   Maybe your parser makes a text node for "\n ", between the
> 'etp' element and the CDATA section, and an other text node
> for the cntent of the CDATA section itself.  I think it has
> to merge all the text in 'etp' before the 'nvgs' element in
> one single text node.

I've heard of products that do that - especially if you supply the input as
a DOM - but it's not conformant behaviour. 

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