Implicit conversion from float -34.79087452471484 to int loses precision