Wait for front-end ping timeout

An end user may leave its computer for a while without using it. The network policy (firewall) might force a close of the TCP connection after a given period of inactivity. To avoid such connection shutdown when there is no GUI exchange, the front-end sends successive ping events after a timeout period of inactivity, to keep the TCP connection alive. The front-end ping is a normal behavior and part of the GUI client/server protocol, this is a built-in "keep alive" solution provided by Genero.

Note:

This feature is not supported when running on mobile devices, or when displaying applications on mobile devices.

The inactivity timeout to send a ping event can be configured in front-ends. On the runtime system side, it is also possible to configure the wait-for-ping timeout with the gui.protocol.pingTimeout FGLPROFILE setting.

With this keep alive technique, a front-end connection always remains open, even if the user leaves the workstation for several hours. If your network connection comes at a cost to support this type of connection, it is possible to configure the front-end and turn off the ping event, when the front-end allows it. Check the front-end configuration documentation for more details.

The wait-for-ping timeout period of the runtime system can be configured with the gui.protocol.pingTimeout fglprofile entry. By default, the runtime system waits for 600 seconds (10 minutes):

gui.protocol.pingTimeout = 800

If you set the wait-for-ping timeout to a value lower than the ping delay of the front-end, the program will stop with the -8063 error after that timeout, even if the TCP connection is still alive. For example, when a front-end has a ping delay of 5 minutes, the recommended minimum value for this parameter is about 330 seconds (5 minutes + 30 seconds to make sure the client ping arrives).

If the front-end program is not stopped properly as happens when, for example, it is killed by a system reboot, or when there is network failure, the TCP connection is lost and the runtime system no longer receives ping events. In this case, the runtime system waits for a specified time, before it raises the error -8063. By default (when no exception handler is defined), the runtime system just stops. However, the error -8063 can be trapped with a WHENEVER ERROR or TRY/CATCH block, to perform some finalization tasks. The program should then immediately stop with EXIT PROGRAM n (where n > 0).