Implicit conversion from float -88.30802603036878 to int loses precision