Implicit conversion from float -8.841059602649011 to int loses precision