Implicit conversion from float -109.3165467625899 to int loses precision