Implicit conversion from float -109.7041420118343 to int loses precision