cursors and input

[Jim Gettys]

On Tue, 2005-03-15 at 10:52 -0500, Michel Danzer wrote:
> On Tue, 2005-03-15 at 10:38 -0500, Jim Gettys wrote:
> > I'm beginning to wonder if cursors should be present at all, or left to
> > the X server to composite on its own....
> 
> I think the latter would give a more consistent look'n'feel (the cursor
> moves in sync with a window instead of slightly ahead of it, ...), we
> just have to make sure it doesn't lag too much...
> 
> 

Experience shows the cursor hasn't lagged seriously, in the software
implementations (since Keith fixed X's scheduler a couple years back).
Whether this would be true given OpenGL programs running in the graphics
engine is related to the general problem we have to solve in the first
place for this approach to be viable (retaining interactivity of the
window system in the face of abusive OpenGL apps).

I have no problems seeing an interface defined: I'm just pointing out
that having a single cursor is a bug; if we define such an interface, it
should support many pointers.

I'll also point out that these may want to be distinguishable visually
from each other; I wonder if in fact the interfaces should allow two
images to be composited as a result, rather than just one image. (one
would be used to allow the user to be visually distinguished from the
base cursor shape).

Or maybe this is gilding a Lilly.
				- Jim