Implicit conversion from float -89.23728813559322 to int loses precision