RE: XSLT2: Clustering, or Grouping the groups

> Thanks, I'll give these a spin. Do you have any comment on which
> one would be the better solution in the general case, or what
> criteria I'd use to decide? Does it matter at all?

In principle I would probably go for a variable containing a sequence of
parentless element nodes. However, in the current Saxon implementation a
document node holding these elements is much more efficient.

Michael Kay