Graphics

Applications can optionally generate graphics, which are displayed by the core client in either its application (window) or screensaver (full-screen) form.

The graphics architecture has not been completed. It may differ between platforms. Some possibilities:

• Applications render their graphics into an offscreen buffer in shared memory. The core client or screensaver BLITs this to the screen. (Problem: no X11 toolkits let you render into memory in the client. Mesa does, but that forces you to use OpenGL).
• Applications are passed a window descriptor by the core client or screensaver. Problem: can you do this in Windows? in X11?
• Applications encode their rendering in a byte stream of some sort (OpenGL had something like this), and send if through a pipe to the core client or screensaver.
• Applications supply their rendering code in a separate shared library which is dynamically linked into the core client or screensaver, and which communicates with the main application through shared memory. This is how United Devices works. Disadvantage: complex.