Implicit conversion from float -7.369942196531781 to int loses precision