>This prompts another question. Is it only the order-dependence of
>combine="group" that you dislike, or is there something else?
I dislike the order-dependence. I also think that use of combine="group"
for refinement is entirely different from what users would expect about
refinement.
>For example,
>do you dislike the fact that forests that match the original pattern don't
>match the new, post-combine pattern?
No, I don't.
>If so, what about attributes?
Addition of attributes by combine="group" is fine to me.
Cheers,
Makoto
