> Actually, I meant:
> 
>     - Extend the definition of the concat() function, so that *each*
> of its (one or more)
> arguments is of type non-empty sequence of xs:string or non-empty
> sequence of text
> nodes.

We didn't do that because it would introduce backwards compatibility
problems.

concat(author, title, subject) in 1.0 gives you the first author and ignores
the rest.

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