Implicit conversion from float -58.574144486692006 to int loses precision