Implicit conversion from float -32.89156626506025 to int loses precision