Alexander Blinne <n...@blinne.net> writes: >> def gen_s(): >> s = [1] >> m = skipdups(heapq.merge(*[(lambda j: (k*j for k in s))(n) for n in >> [2,3,5]])) >> yield s[0] >> while True: >> k = m.next() >> s.append(k) >> yield k
Nice. I wouldn't have been sure that "for k in s" worked properly when s was changing like that. There is a space complexity problem compared to the Scheme or Haskell version: all the s[i]'s are saved in the s array, instead of being discarded once they are yielded. That means generating n elements needs O(n) space instead of O(n**0.7) or something like that. I guess you can get around it with collections.deque instead of a list. -- http://mail.python.org/mailman/listinfo/python-list
