Implicit conversion from float -2.7792915531335325 to int loses precision