Implicit conversion from float -12.188782489740092 to int loses precision