Implicit conversion from float -42.60504201680672 to int loses precision