Implicit conversion from float -30.38216560509555 to int loses precision