> > couldn't one get there with, say,
> > sequence-before(sequence, item+)
> 
> Not really. How would you do "return the subsequence of $S 
> before the first element that has @page='2'"?
> 

Oh, I see now what you're suggesting: something based on comparing node
identity. Well, for a large class of problems that might work:

sequence-before($chapters, $chapters[@page='2'][1])

But it's clumsy, it conceptually involves scanning the sequence twice (and
therefore can't be streamed without some optimization); and it only works
for nodes (or works differently for atomics).

Michael Kay
http://www.saxonica.com/