Implicit conversion from float -23.23722149410223 to int loses precision