Implicit conversion from float -30.2295918367347 to int loses precision