Implicit conversion from float -88.76288659793815 to int loses precision