Implicit conversion from float -109.66386554621849 to int loses precision