No longer can they assume one processing model per call - they must be able to switch between the two based on what they get back from the server.

• 它们不再能够假定每次调用一个处理模型它们必须能够基于从服务器取回的信息来在两者之间转换。