Implicit conversion from float -135.140712945591 to int loses precision