RE: [XQuery] Computed CDATA constructor

> We could, but it seems overzealous for something that seems like it 
> belongs in the spec. It is frustrating to see function names and 
> parameters be continuously changed but useful, low 
> overhead/implementation features not adopted.

If the XSLT experience with disable-output-escaping is anything to go by,
anything that distorts the data model to provide hints to the serializer is
going to cause no end of anguish to both users and implementors.

I would personally prefer to see CDATA "constructors" dropped completely.

Michael Kay