On Wed, 2006-05-31 at 12:15 +0100, David Carlisle wrote:
> > I'm pretty sure that this is some kind of grouping problem,
> 
> it is, which means that the answer looks rather different for xslt2 or
> xslt1 (conceptually the same but looks nicer in 2) is an xslt2 answer OK
> or do you want a muenchian grouping xslt1 version?

Hmm, good question :)

I'm using xsltproc, so I'm thinking XSLT 1.0.

I did look at the Muenchian grouping stuff, but to be honest I couldn't
make head or tail of it :D   I think I conceptually get for-each-group,
and know roughly what it is I'm trying to do, but there must be some
paradigm shift my brain hasn't achieved yet :)

Thanks,

Alex.