Implicit conversion from float -34.72178060413354 to int loses precision