Implicit conversion from float -135.08379888268155 to int loses precision