Implicit conversion from float -78.40336134453781 to int loses precision