Implicit conversion from float -9.238578680203048 to int loses precision