Implicit conversion from float -10.241545893719817 to int loses precision