xkcd

Welp, guess it's time for:

for(float i = 0.0; ....) {...}

but therewill be occurences when you add/subtract the same number when writing and reading an int, so it is not entirely eliminated

wait, but what about off-by-39-or-51 errors?

In this language you would account for this behavior so all comparisons and lookups would be designed to be valid for the increased range. To do this you'd have to record integers in increments of 10, so there would be no (or very few) off-by-40-to-50 errors, or off-by-1 errors, but there would be off-by-10 errors.