Implicit conversion from float -142.73542600896863 to int loses precision