Implicit conversion from float -72.65060240963854 to int loses precision