Implicit conversion from float -256.77966101694915 to int loses precision