Implicit conversion from float -5.990180032733221 to int loses precision