Application based Error Handling

• The application can easily implement “at least once” reliability by using idempotent operations (i.e. operation that can be executed multiple times without side effects) and using a simple timeout mechanism. Figure 6.10 shows the state machines for “at least once” reliability using implicit acknowledgements. When the client sends out the request it starts a timer with the timeout specified for the specific method. If no response is received before the timer expires (round transition at the top), the client will retry the operation. A Typical number of retries would be 2, so that 3 requests are sent.
  應用程序可以通過使用冪等操作（即可以多次執行且無副作用的操作）和使用簡單的超時機制輕松實現“至少一次”可靠性。 圖6.10顯示了使用隱式確認“至少一次”可靠性的狀態機。 當客戶端發出請求時，它會啟動一個定時器，并為特定方法指定timeout。 如果在計時器到期之前沒有收到響應（頂部的圓形轉換），則客戶端將重試該操作。 典型的重試次數為2，因此發送3個請求。

• The number of retries, the timeout values, and the timeout behavior (constant or exponential back off) are outside of the SOME/IP specification and may be added to the interface definition.
  retry次數，timeout值和timeout行為（常量或指數退避）不在SOME / IP規范范圍之內，可以添加到接口定義中。