Implicit conversion from float -7.883211678832126 to int loses precision