GEOS SDK TechDocs
|
| 
3.3 Thread Scheduling
| 
4 Using Multiple Threads
There are two standard architectures for GEOS applications: single-thread and dual-thread. While the single-thread option is somewhat easier to program, there are distinct advantages to the dual-thread method.
In the dual-thread architecture, one thread manages the application's user interface while the other manages the rest of the application's functionality. Since both threads are event-driven, each has an event queue. Messages that are sent to user interface objects (resulting from mouse clicks, keyboard input, etc.) can be handled without waiting for other tasks in the application to be completed. This allows the application to respond to user input (by putting up menus, moving windows, and so on) without first completing the current non-user-interface task (which may involve a lot of computation).
The dual-thread architecture, however, poses a problem of synchronization: One thread can get ahead of the other. Threads that count on each other must keep track of each other's progress in order to avoid this; when potential problems are identified, use semaphores to keep the threads in line (see Synchronizing Threads ).
GEOS SDK TechDocs
|
| 3.3 Thread Scheduling
| 4 Using Multiple Threads