Implicit conversion from float -23.588390501319253 to int loses precision