Implicit conversion from float -125.7190635451505 to int loses precision