Supporting these is somewhat easier for <browser remote> because there's a choke point in RenderFrameParent (through nsDisplayRemote) that's a clear place to set the compensating transform on the shadow root that these APIs entail. For <browser remote>, it's not clear to me where this should be done, though that may just be because I don't know the in-process path as well. I suspect that we can simulate these APIs pretty well by setting a CSS transform on a "convenient element" (not sure exactly which). We're also not going to have the layer FrameMetrics immediately available for the <browser remote=false> container layer that we get for <browser remote>. We very much want to support these so that the fennec frontend code can be the same for in- and out-of-process <browser> content. This makes it possible to create a nice little reftest-like framework as well.

Chris - are you the guy to work on this?

One of; I, tn, roc, and stechz now know what's needed for this I think.

b- on bug 597081 means that this bug is test-only for the time being, so it doesn't block. Will look for approval on patches as they arrive.

Metro Firefox for Windows 8 is using in-process content (at least for now) and we would very much like to have this feature or some equivalent (see bug 766142, bug 789906). Can we talk to someone on the graphics team about when this might be feasible?

I think you want AsyncPanZoomController

This code isn't maintained anymore.