Implicit conversion from float -10.176678445229683 to int loses precision