Implicit conversion from float -8.982035928143716 to int loses precision