Add convenience methods for PLSA visible client dimensions.

third_party/WebKit/Source/core/paint/PaintLayerScrollableArea.h

Index: third_party/WebKit/Source/core/paint/PaintLayerScrollableArea.h
diff --git a/third_party/WebKit/Source/core/paint/PaintLayerScrollableArea.h b/third_party/WebKit/Source/core/paint/PaintLayerScrollableArea.h
index a00b42bb565f10b645cde6b8b1ffc33bbdfc8a6b..515333af532fad11a147d22687db1399ffa16ccb 100644
--- a/third_party/WebKit/Source/core/paint/PaintLayerScrollableArea.h
+++ b/third_party/WebKit/Source/core/paint/PaintLayerScrollableArea.h
@@ -297,6 +297,9 @@ class CORE_EXPORT PaintLayerScrollableArea final
   CompositorAnimationHost* GetCompositorAnimationHost() const override;
   CompositorAnimationTimeline* GetCompositorAnimationTimeline() const override;
 
+  LayoutSize ClientSize() const;

[skobes, 2017/04/27 23:22:47]

Isn't this the same as VisibleContentRect().Size()?  Should we call it
VisibleContentSize for consistency?

[szager1, 2017/04/28 11:04:53]

VisibleContentRect is based on layer_.Size(), which gets updated at a different
time from Box().ClientWidth().  The methods that use ClientSize() may run after
box layout but before UpdateLayerPositionsAfterLayout().

At the end of this series of patches, the box and layer will get their sizes
simultaneously (from the point of view of PLSA).  When that happens, the code
can (and should) use layer size everywhere.

The name ClientSize is temporary; I chose it because it mostly replaces calls to
Box().[PixelSnapped]Client(Width|Height), and communicates the fact that it's
based on box size, so is safe to call from UpdateAfterLayout.

I don't think it's worth bikeshedding over the name, so I'll do whatever you
prefer.

+  IntSize PixelSnappedClientSize() const;
+
   void VisibleSizeChanged();
 
   // FIXME: We shouldn't allow access to m_overflowRect outside this class.