Jesus Cea said: > Current durus btree implementation is indexed by key. So you can > get the > smaller or the bigger key in the btree, or iterate the btree, just > fine. > > For an application I'm designing just now, I would need a "counted" > btree, just like explained in > http://www.chiark.greenend.org.uk/~sgtatham/algorithms/cbtree.html > > This structure can retrieve the i-th element in the btree very easily, > without knowing its key. Also, "len()" would be very fast. I just stumbled across this same need and option. Did you ever implement this? Dave