On Fri, 14 Mar 2003, David Carlisle wrote:

> The chances are that your XSLT system includes rewrite rules
> that transform these kinds of things from one style to the other,
> depending on which is the best for the implementation so in practive it
> may make no difference whatsoever, however if the predicate is expensive
> to evaluate and you assume wandering around the tree is less so, and
> you assume the system does no rewrites or optimisations then
> 
> //part[--complicated-test--involving .]/ancestor::*
> 
> is on the face of it more efficient than
> 
> //*[--complicated-test---involving descendant::part] 
> 
> as in the former the complicated test will be done just once for each
> part element, but in the latter it will be executed on each part element
> as many times as that element has ancestors.

yup, i see your point.  and i suspect this isn't the first time
someone has asked this question, i just wanted to be sure there
wasn't something subtle about the above that i didn't notice at
first glance.

thanks.

rday


 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list