Implicit conversion from float -7.257383966244717 to int loses precision