The snapshot file reader uses system encoding. This causes failing tests on different machines. UTF-8 encoding should be ok.