Implicit conversion from float -23.465346534653463 to int loses precision