Implicit conversion from float -23.317757009345783 to int loses precision