Implicit conversion from float -7.43801652892563 to int loses precision