Implicit conversion from float -115.42082738944367 to int loses precision