Compare two methods of computing partitions using Python's "yield" construct

partitions.py

#!/usr/bin/env python

def partitions(n, curtot = 0, cur = []) :
    if curtot == n :
        yield cur

    t = cur[-1] if cur else 1

    for i in xrange(t, n - curtot + 1) :
        for j in partitions(n, curtot + i, cur + [i]) :
            yield j

def part(n):
    # base case of recursion: zero is the sum of the empty list
    if n == 0:
        yield []
        return
        
    # modify partitions of n-1 to form partitions of n
    for p in partitions(n-1):
        yield [1] + p
        if p and (len(p) < 2 or p[1] > p[0]):
            yield [p[0] + 1] + p[1:]

if __name__ == '__main__':
    import timeit
    t1 = timeit.Timer('for i in partitions(30) : pass', 'from __main__ import partitions')
    t2 = timeit.Timer('for i in part(30) : pass', 'from __main__ import part')
    print t1.timeit(number=10), t2.timeit(number=10)