Implicit conversion from float -100.23605150214593 to int loses precision