Implicit conversion from float -9.876237623762378 to int loses precision