Dimitre Novatchev [dnovatchev@xxxxxxxxx] wrote:
> I have both bad and good news for you.
> 
> The Bad News:
>   The recursive algorithm isn't the fastest.
> The Good News:
>   I am enclosing here the code implementing the  fastest (that I
>   know of) algorithm, anf it is non-recursive.

Wow! Its all good news the way I see it... My testing of your 
stylesheet has confirmed your results. Personally, I think your
algorithm should be immortalized in Dave Pawson's FAQ :-)

> The non-recursive algorithm exhibits linear behaviour with 
> MSXML4 and Saxon and sub-linear! one with JD, XalanJ and 
> XalanC.

Sweet!

> I could try still speeding up the non-recursive algorithm, by 
> using a faster search than linear to find the position of a 
> record node in the string with positions -- this will require 
> that all positions must have the same (some maximum) length. 
> Or I could record the positions in a node-set, for which
> binary search is straight-forward.
> 
> In case you are still not satisfied with the speed of the 
> non-recursive algorithm, just let me know :o)

Well, I'm more than satisfied with it, but if you want to make 
it even faster, don't let me stop you :-)

> Hope this helped.

It certainly did! I'm going to apply your algorithm to my real
problem straight away. Thanks.

Cheers,
Raman Gupta

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