Live views can be used in different kinds of applications. They can be used in GUI, interactive applications and in non-GUI, batch processing-style applications (see How to use live views in non-GUI code). Live views are optimized for both modes, GUI (interactive) and non-GUI (batch). They distinguish between these modes and operate accordingly. In GUI, live views react to a change immediately when the change happens. They do it fast, using incremental algorithms without re-populating themselves, (see Maintaining the view: Incremental View Maintenance), but still it is not always suitable for batch, non-interactive processing. In GUI, interactive programs, with data binding, immediate reaction to change is what is usually needed because the user needs to see the change on the screen. In batch processing, on the other hand, a view may be accessed long after the change occurred or even not at all. So updating that view before it is actually accessed by the program may be an unnecessary drain on resources. By default, live views distinguish between these two modes automatically, but the programmer has an option to control that using the MaintenanceMode|keyword=MaintenanceMode Property property. In Immediate mode, which is the default for GUI, the view is maintained immediately on every change. In Deferred mode, which is the default in a non-GUI case (when there are no listeners attached to the view), the view is maintained on demand. It is not kept in sync with base data until it is needed or until there is a request for data from that view. When such a request arrives, the view sees that it is in a "dirty" unsynchronized state, so it automatically updates (maintains) itself to come in sync with the changed base data.