rdb

For my own benefit, I tried to extract the bits from this that have actually changed from the bits that are cleaned up, and if I'm not mistaken, this change...

It would help to understand the original bug (perhaps with a repro case) since—following the code in my mind—the original behaviour seems more correct than the fix.

Ah, that's great—I had unfortunately simply not noticed that you had split them up.

Is there an issue for the bug report or a repro case so that I can validate the solution against the problem?

I tested this with your updated Roaming Ralph sample and it does look like it may be the right fix for how CollisionHandlerFloor was written, namely with the assumption that...

I don't think it's possible to set a non-screen-aligned scissor region in OpenGL. We'd have to switch to clipping planes.

Yes, it would perform more poorly. Perhaps we should add a rotatable flag that causes it to be done with clip planes.

I appreciate the contribution, but I'm inclined to reject this for the reasons described in #1491. As-is there is no unit test, it increases the surface area of the API...

Yes, and this should probably be done on top of the work that's been done in the multitouch branch, though it hasn't been touched in a while.

@drewc5131 I agree that DirectGUI is not a great GUI system in many respects, but this is not the right thread to be having that discussion. For what it's worth,...