On Tue, 21 Sep 2004 17:59:21 +0100, David Carlisle wrote
> //item
> 
> are you sure that you want to use // here (or anywhere) it is very
> expensive causing a full search of the entire document for item
> elements. If you know that the item elements are at a particular 
> depth using an explict path (or an xsl:key) is likely  to be quicker.

I probably don't want //item.  What I want is all elements named item which
are descendants of the current element - but I am not sure what the syntax
would be (still looking).  I don't know xsl:key or how it could help me - do
you know of a good example/URL?

> I didn't fully follow your program logic but if this comment means that
> you want to process all but the last $remaining nodes in the 
> selection then you can do
> 
> <xsl:for-each select="(...... whatever you had before)
>            [position() &lt; last()-$remaining]">
> 

I have a question about position().  If this xsl:for-each contained an
xsl:sort, the position() values in the select of the for-each would not be the
sorted positions, right?  

<xsl:for-each select="whatever[position() &lt; last()-$remaining]">
<xsl:sort select="somethign">
<!-- would the node position() values here be the same as those in the
for-each? -->

Hopefully I am making some sense and not just confusing myself.