Code review comment for lp:~alan-griffiths/mir/fix-1535894

> Fixing the problem in the compositor is ideal, because that's the only place
> where you might know about the additional rendering that's outside the
> expected confines of the surface rectangle.

I assume you mean "Fixing the problem in the DisplayBufferCompositor" (not the Compositor).

Shifting the logic from this MP and from -c 3274 to the DefaultDisplayBufferCompositor (with appropriate wiring) would address your concerns with both?

I recall that we had similar logic once in an earlier version, with the DisplayBufferCompositor returning a measure of uncomposited buffers. Can you recall why that was removed - we don't want to repeat the mistakes of the past.