Implicit conversion from float -85.93320235756383 to int loses precision