Implicit conversion from float -7.349177330895799 to int loses precision