Replace cursor operations

The X Link Kit supports three types of cursor implementations for either the NFB or the X Consortium porting layer. All cursor movement comes in through the SCO system layer, so it must know how to move the cursor. The three types of cursors are:

complete software cursor

used by default and works with any graphics adapter. It is implemented by the call to scoSWCursorInitialize(D3nfb) in the xxxInit.c file. Internally, this routine calls an mi routine called PushPixels( ) that breaks down to ReadImage(D3nfb) and DrawImage(D3nfb) for NFB drivers or SetSpans( ) and GetSpans( ) for the X Consortium layers.

sprite cursor

improves cursor performance for adapters that have a BitBlt routine but do not have a hardware cursor. This is the most difficult level to write; seven routines must be written and then passed to scoSpriteInitialize( ) in a structure. See the ddx/mi/miDispCur.c file for an example.

pointer level

provides full hardware cursor control. The DIX layer assumes that cursor drawing does not affect drawing to the screen. Four routines must be written, for creating, destroying, changing, and moving the cursor.