RE: treating a CDATA node as Elements nodes

> CDATA sections are viewed as a single node that can't
> be split up even more in template rules right?

Wrong. CDATA sections don't result in nodes in the tree. CDATA is simply a
way of escaping special characters, so that & means ampersand rather than
introducing an entity reference, and < means less-than rather than
introducing a start tag. The content of the CDATA section is concatenated
with adjacent text to produce a normal text node in the tree. CDATA section
boundaries are not considered to carry any information.

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