an integer to avoid an unhelpful deprecation warning.  (Before Python
2.3, range() and many other built-in functions accepted floats as ints
"by accident"; those cases now generate warnings when passes a float.)