Implicit conversion from float -7.211155378486069 to int loses precision