Implicit conversion from float -42.1232876712329 to int loses precision