Implicit conversion from float -23.978494623655905 to int loses precision