[Chromoting] Improve DirectX capturer to support multiple outputs

webrtc/modules/desktop_capture/desktop_geometry.h

Index: webrtc/modules/desktop_capture/desktop_geometry.h
diff --git a/webrtc/modules/desktop_capture/desktop_geometry.h b/webrtc/modules/desktop_capture/desktop_geometry.h
index 047eeec3d9c8fd89c13f29347a0a67b77b2096da..9108db5e82e997c6a3f68671a0bab0ab0aace87c 100644
--- a/webrtc/modules/desktop_capture/desktop_geometry.h
+++ b/webrtc/modules/desktop_capture/desktop_geometry.h
@@ -124,6 +124,9 @@ class DesktopRect {
   // Finds intersection with |rect|.
   void IntersectWith(const DesktopRect& rect);
 
+  // Finds union with |rect|.

[Sergey Ulanov, 2016/07/08 22:36:53]

This actually returns a rect that covers both rectangles. It shouldn't be called
it "union". I think it is useful only rarely, so I don't think we want it as a
method of DesktopRect. Most code that needs to combine rects should be using
DesktopRegion

[Hzj_jie, 2016/07/11 00:54:59]

Done.

+  void JoinWith(const DesktopRect& rect);
+
   // Adds (dx, dy) to the position of the rectangle.
   void Translate(int32_t dx, int32_t dy);
   void Translate(DesktopVector d) { Translate(d.x(), d.y()); };