Implicit conversion from float -109.86376021798364 to int loses precision