Implicit conversion from float -90.23255813953489 to int loses precision