Implicit conversion from float -109.66555183946488 to int loses precision