Implicit conversion from float -88.83369330453564 to int loses precision