Sieve of Eratosthenes in Python

primes.py

def primes(maximum):
    discarded = set()
    was_discarded = discarded.__contains__
    discard_all = discarded.update
    limit = maximum + 1
    yield 2
    for m in xrange(3, limit, 2):
        if not was_discarded(m):
            yield m
            discard_all(xrange(m * 2, limit, m))

How does it compare with mongodb?

Let's say this is not exactly web scale. :c)

BTW, "web scale" this: https://gist.github.com/1256699 :c)