Implicit conversion from float -23.91417425227567 to int loses precision