topmost windows

[Deron Johnson]

I'm going to try to summarize the current state of the
discussion. Please let me know if you disagree with any of these
points, or if there was a candidate that I've missed.

First of all, it appears that the IncludeInferiors-to-the-root
approach is not feasible because it requires significant changes to
OpenGL. The grab/raise/ungrab approach is a non-starter because of the
attendant visual artifacts. And the approach I've been investigating
of reparenting windows to a pseudo-root window is kludge city.

Whatever approach we adopt it must solve at least these two
issues:

1. Parent clipping issue - children punch holes in the parent clip

2. DID rendering issue - Composited windows still have DIDs rendering to
   the screen.

Here is my understanding of the proposals which have not yet been
completely ruled out. Are there any others that I have missed?

1. Topmost windows
    Pros:
	Easy to implement.
	Precedent in other window systems.
    Cons:
	Not clear how general we need them to be.
	Must be careful about how we define semantics in order
	    to minimize changes to WMs

2. Composite (or XFIXES) priority window
    Pros:
	Avoids some of the WM issues surrounding topmost windows
    Cons:
	May not be general enough.

3. Pseudo-root window
    Pros:
        Specification already exists.
    Cons:
	Doesn't solve the DID rendering problem