Dealing with time out failures

It is possible that a data interface command, especially a GET command, can take a significant amount of time. One effect of this is that, by the time the response is available, the HTTP socket connection between the client application and Web User Interface server could have expired. To avoid this situation, the Web User Interface server can respond early with a status of TIMEOUT, while continuing to process the original request.

To retrieve the results of the original request, the client application should reply to a TIMEOUT response by sending another request to the server. On this new request, the client application specifies the TIMEOUTTOKEN query string and value taken from the TIMEOUT response. If the server receives a request with a TIMEOUTTOKEN, it either sends the response built for the original request or, if it is still not ready after a further period of time, sends another TIMEOUT response.

If a client application fails to reply to a TIMEOUT response, the results of the original request become inaccessible, but continue to use the resources of the user session. It is important, therefore, for the client application to reply to all TIMEOUT responses.
Note: The TIMEOUTTOKEN value is a "black box" to the client application, which should return the TIMEOUTTOKEN to the server unaltered. The client application should also assume that the TIMEOUTTOKEN value is variable length to protect against possible change in a future release of the data interface.


eyuadar.html | Timestamp icon Last updated: Monday, 5 June 2017