Implicit conversion from float -5.8239277652370305 to int loses precision