Isn't that testing the *implementation* rather than the intended
effect? I think it'd be better to test the memoization by ensuring
the method doesn't calculate the value twice, rather than asserting
(shoulding?) the thing is memoized?
OK, you've convinced me. Send a patch and I'll look it over . I
still think that you should include the implementation test, but I can
see where you're coming from and piggy backing on the AS tests is a