Implicit conversion from float -30.37366548042705 to int loses precision