RE: ORA-FO-UNORDERED 15.1.15 fn:unordered

> It is a function! It returns a permutation of its input. 
> It's only difference from other functions is in its 
> motivation: it exists in order to be optimized. 
> Michael Kay 
> 
> Which implies that it does nothing for the end user?

It might deliver an order-of-magnitude increase in performance.

> What a curious beast.

Yes, perhaps!
> 
> If its for query, why not put it in the query spec, not in 
> the xslt arena as well?
> 

While XQuery implementors are definitely pursuing the need to run queries
against terabyte-sized XML databases, where optimizing by selecting the
right indexes is of paramount importance, there are also people who want to
run XSLT stylesheets directly against such databases. Since it's no burden
on the implementor to provide this function (it can be implemented as a
no-op) there would be no advantage in making it the only function in the
whole library that XSLT doesn't support.

Michael Kay

Received on Monday, 29 September 2003 12:48:59 UTC