Re: Beware the count method with Muenchian grouping (w

Hi Daniel,

> I did some time runs with MSXML 3.0 and Saxon 6.0.2 (with Sun JDK
> 1.3) on Windows NT 4.0 Workstation with Sp5.
[snip]

Wow - thanks very much for such a detailed test and report. It's very
interesting to see the differences between the processors (BTW Saxon
6.1's out now - I don't know how much difference that would have
made), which does seem to rest on the generate-id() implementation.
Mike's mentioned on the Saxon list that he's planning to address that,
so perhaps things will change there.

The other interesting thing was about defining keys with predicates in
the match as opposed to filtering them before or after finding the
relevant ones. I imagine that the differences here are highly
dependent on the source XML: if there were a higher proportion of
'secondary' (i.e. filtered out) nodes than 'primary' nodes, for
example. Whether the filter is best placed before or after identifying
the unique nodes probably also depends on how many nodes share the
same key value.

Thanks again,

Jeni

---
Jeni Tennison
http://www.jenitennison.com/



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