Table cell background painting

third_party/WebKit/Source/core/layout/LayoutTableSection.h

Index: third_party/WebKit/Source/core/layout/LayoutTableSection.h
diff --git a/third_party/WebKit/Source/core/layout/LayoutTableSection.h b/third_party/WebKit/Source/core/layout/LayoutTableSection.h
index bc28e972d5aba743c58880c43cb1d670e0297e5c..c9ff835a1aa66f4d3c7fd9fb529c929d1673d43c 100644
--- a/third_party/WebKit/Source/core/layout/LayoutTableSection.h
+++ b/third_party/WebKit/Source/core/layout/LayoutTableSection.h
@@ -313,6 +313,8 @@ public:
     bool foregroundIsKnownToBeOpaqueInRect(const LayoutRect&, unsigned) const override { return false; }
     bool backgroundIsKnownToBeOpaqueInRect(const LayoutRect&) const override { return false; }
 
+    LayoutRect positionOfIdealCell(unsigned row, unsigned col) const;

[Xianzhu, 2016/03/09 18:40:56]

s/col/effectiveCol/ so that any one reading this header file will know a
effective column is accepted. Also move document from .cc to here.

I think we can just use 'Cell' instead of 'IdealCell', because
LayoutTableSection is already using the term to indicate the grid cells (may
correspond to a part of real cell). See cellAt(), primaryCellAt(), etc.

The implementation seems to return physical position of the 'ideal cell', so
this function could be named getCellPhysicalPosition(). (Our current style
suggests a verb at the beginning of a function.)

Perhaps you can split this function into two:
1. getCellPosition() which returns logical position;
2. a function to convert logical position to physical position;

Then getCellPhysicalPosition() can be implemented by calling the above two
functions.

[atotic1, 2016/03/15 16:50:48]

Split the function into:

getCellPosition
getCellPhysicalPosition
transformLogicalToPhysicalPosition

+
 protected:
     void styleDidChange(StyleDifference, const ComputedStyle* oldStyle) override;
     bool nodeAtPoint(HitTestResult&, const HitTestLocation& locationInContainer, const LayoutPoint& accumulatedOffset, HitTestAction) override;