Implicit conversion from float -58.654708520179355 to int loses precision