Implicit conversion from float -34.80916030534351 to int loses precision