Implicit conversion from float -30.35175879396985 to int loses precision