Implicit conversion from float -7.113402061855666 to int loses precision