Make TestRenderWidgetHostView::Show/Hide call through to RWHI

content/browser/renderer_host/render_widget_host_view_base.cc

Index: content/browser/renderer_host/render_widget_host_view_base.cc
diff --git a/content/browser/renderer_host/render_widget_host_view_base.cc b/content/browser/renderer_host/render_widget_host_view_base.cc
index d09da0790d0df887ac8265565c89e265dfe90e46..f1c64550a026fd903b0aa80bb0ce8e3db6cd3d46 100644
--- a/content/browser/renderer_host/render_widget_host_view_base.cc
+++ b/content/browser/renderer_host/render_widget_host_view_base.cc
@@ -103,8 +103,10 @@ bool RenderWidgetHostViewBase::GetBackgroundOpaque() {
 }
 
 gfx::Size RenderWidgetHostViewBase::GetPhysicalBackingSize() const {
-  display::Display display =
-      display::Screen::GetScreen()->GetDisplayNearestWindow(GetNativeView());
+  display::Screen* screen = display::Screen::GetScreen();
+  if (!screen)

[boliu, 2017/02/08 20:13:34]

I'm curious which tests and which platform had had this problem?

Maybe should fix test harness rather than doing this.

If we don't end up doing that, definitely add a comment that the check is for
tests

[sky, 2017/02/08 23:21:36]

+1 to this. I'm mildly worried this is going to cause problems down the line and
it would be better to fix it now.

[johnme, 2017/02/09 18:59:29]

Good news - it turns out this change to GetPhysicalBackingSize is no longer
necessary! It seems some refactoring has happened since I originally wrote this
patch that makes this more robust.

(I can't remember exactly which tests this was causing to fail, but you can see
which platforms were failing by looking at the try results for patch set 2:
Windows, Linux, and Chrome OS).

+    return gfx::Size();
+  display::Display display = screen->GetDisplayNearestWindow(GetNativeView());
   return gfx::ScaleToCeiledSize(GetRequestedRendererSize(),
                                 display.device_scale_factor());
 }