Implicit conversion from float -10.090293453724598 to int loses precision