Re: References in XPath that can be representedbyastringandstill can be garbage collected

Vladimir Nesterovsky wrote:
>> I don't think so. For node identity, there is no issue with order of
>> evaluation, the only constraint is some strictness about evaluating an
>> expression the right number of times.
> 
> I think these are very close issues:
> 
> for
>  $i in (1 to 10),
>  $v in f:function-depending-on-node-id()
> return
>  ($i, $v)
> 
> You cannot rewrite foregoing like:
> 
> for
>  $v in f:function-depending-on-node-id(),
>  $i in (1 to 10)
> return
>  ($i, $v)
> 
> as when you change the order, the function will be called different 
> number of times.

Sure, they're related issues, but they aren't the same. As an 
implementer there's a big difference between the two constraints - the 
latter leaves many more optimisation approaches available. From a 
specification perspective, the latter does not require the definition of 
a strict execution order - therefore XPath 2.0 and XQuery do not have one.

John

-- 
John Snelson, Oracle Corporation            http://snelson.org.uk/john
Berkeley DB XML:            http://oracle.com/database/berkeley-db/xml
XQilla:                                  http://xqilla.sourceforge.net