Implicit conversion from float -0.27829313543600165 to int loses precision