Implicit conversion from float -58.08362369337979 to int loses precision