Implicit conversion from float -42.28486646884272 to int loses precision