An example of #6 is a situation where you need an element that manually invalidates the session is when you involve CTI.

Lets say youur CTI system is higer up in the network than your voice browser. The idea is that you have an action element that will communicate with a CTI system to determine if it should route the call to an agent. You can use backend calculations to make this determination. If you want to do this, you send a message to the CTI system to take the call away from the voice browser-Audium combination and send it to an agent. In this situation, since the call simply dissapears, the voice browser does not throw any special events indicating the call has ended, it basicaly does nothing. As a result, the Audium port would remain open until it eventually timed out. So you make that action element that communicated with the CTI system manually invalidate the session because you know that after it does that, there will not be another request coming into Audium from the voice browser for this call.

Hope this explains things!

Elam