Implicit conversion from float -5.977443609022544 to int loses precision