Implicit conversion from float -88.18574514038878 to int loses precision