compositor

That doesn't mean it can't be forced to work through huge render farms, hacks, compositor painting and who knows what else.

However, if it's in CSS, it can be proxied out to the compositor thread, which runs very little.

Blender uses DAGs for a number of things - eg an image compositor, and for shading networks.

Or perhaps the presence of a background image on the li tags forces the compositor to behave?

Syncing the compositor's page flips to vsync is already possible if you have hw layers on.

In composited environments, tearing would be inevitable because the compositor would use a nonsynchronized back-to-front-buffer copy.

GetPipelineTitleMsg(PipelineId), /// Requests that the constellation inform the compositor of the a cursor change.

Just changing the scroll offset on a native scrolling root should cause the compositor's tiled rendering infrastructure to automatically do the right thing.

As of early 2013, Høgsberg is experimenting with network transparency using a proxy Wayland server which sends compressed images to the real compositor.

There's a big difference now in that a hardware-accelerated compositor is used.

There is active work on allowing WebGL and canvas in workers, with the worker talking directly (not via the main thread) to the compositor.

Arch + xmonad/i3/awesome/whatever no-desktop-compositor-included-DE looks like that out of the box.

What has been added to the spec is being able to run one "compositor" in another.

These get "direct" output (they skip the compositor) but don't use a hardware overlay, and therefore tear.

This is actually how Desktop Window Manager (aka "Aero") on Windows and Quartz Compositor on OS X actually work.

I think that you're suggesting to "stack" compositor to avoid this issue (not a new idea), I hope that this is what will happen eventually..

It happens in design patterns like Observer, Delegator, or Compositor.