Actually, a colleague recently convinced me to start using signed ints for, e.g., for loops instead of unsigned ints. His reasoning was that if you're overflowi

Actually, a colleague recently convinced me to start using signed ints for, e.g., for loops instead of unsigned ints. His reasoning was that if you're overflowing with signed integers, you'll probably overflow with unsigned integers too (we work with very large numbers), but it's easier to notice if you have a negative number rather than silently wrapping around to a still-valid value.