Once bug 1101628 is done we should not need the APZCTreeManager::GetAllowedTouchBehavior function any more. Even now it's mostly just a dummy since we had a different idea of how the code would be architected when that function was added. With the new architecture as described in https://github.com/mozilla/gecko-dev/blob/master/gfx/doc/AsyncPanZoom.md#technical-details it is not needed.

... and by that I mean https://treeherder.mozilla.org/#/jobs?repo=try&revision=afe86e170d54

Comment on attachment 8572865 [details] [diff] [review] Patch Review of attachment 8572865 [details] [diff] [review]: ----------------------------------------------------------------- ::: widget/windows/winrt/MetroInput.cpp @@ +516,4 @@ > for (uint32_t i = 0; i < aOutBehaviors.Length(); i++) { > + // performing hit testing fallback: asking content to perform hit testing itself > + // (in spite that this operation has high latency). > + aOutBehaviors.AppendElement(mWidget->ContentGetAllowedTouchBehavior(aTransformedEvent->touches[i]->mRefPoint)); We're looping through aOutBehaviors, and for each one appending a new element to it? Should we be looping through the touches instead?

Eek, good catch! :) There was also a build error in TestAsyncPanZoomController that I missed, will update the patch in a sec.

Comment on attachment 8572869 [details] [diff] [review] Patch Review of attachment 8572869 [details] [diff] [review]: ----------------------------------------------------------------- ::: gfx/tests/gtest/TestAsyncPanZoomController.cpp @@ +132,5 @@ > }; > > class TestAsyncPanZoomController : public AsyncPanZoomController { > public: > + typedef uint32_t TouchBehaviorFlags; I'd rather have this typedef in an APZ header file than have different files (here, InputQueue and TestAsyncPanZoomController) define it independently.

Updated as per review comment. Try push for build is clean: https://treeherder.mozilla.org/#/jobs?repo=try&revision=b725c665bfab