Implicit conversion from float -6.008119079837613 to int loses precision