Implicit conversion from float -41.29175946547886 to int loses precision