Implicit conversion from float -5.983146067415731 to int loses precision