Reducing CPU cycles of hidden/minimized browser windows

content/browser/web_contents/web_contents_view_aura_browsertest.cc

Index: content/browser/web_contents/web_contents_view_aura_browsertest.cc
diff --git a/content/browser/web_contents/web_contents_view_aura_browsertest.cc b/content/browser/web_contents/web_contents_view_aura_browsertest.cc
index 330f624685b00f9f3ec4a9546c72edd1784e7983..ca2362f051deb2984da0e82ef92f7ecae6053aa3 100644
--- a/content/browser/web_contents/web_contents_view_aura_browsertest.cc
+++ b/content/browser/web_contents/web_contents_view_aura_browsertest.cc
@@ -594,8 +594,7 @@ IN_PROC_BROWSER_TEST_F(WebContentsViewAuraTest,
   window->AddChild(shell()->web_contents()->GetContentNativeView());
 }
 
-IN_PROC_BROWSER_TEST_F(WebContentsViewAuraTest,
-                       ContentWindowClose) {
+IN_PROC_BROWSER_TEST_F(WebContentsViewAuraTest, ContentWindowClose) {
   ASSERT_NO_FATAL_FAILURE(
       StartTestWithPage("files/overscroll_navigation.html"));
 
@@ -684,4 +683,18 @@ IN_PROC_BROWSER_TEST_F(WebContentsViewAuraTest,
   EXPECT_FALSE(controller.CanGoForward());
 }
 
+// Verify that hiding a parent of the renderer will hide the content too.
+IN_PROC_BROWSER_TEST_F(WebContentsViewAuraTest, HideContentOnParenHide) {
+  ASSERT_NO_FATAL_FAILURE(StartTestWithPage("files/title1.html"));
+  WebContentsImpl* web_contents =
+      static_cast<WebContentsImpl*>(shell()->web_contents());
+  aura::Window* content = web_contents->GetContentNativeView();

[sky, 2014/05/20 23:33:51]

I'm surprised this works. web_contents->GetContentNativeView returns a
descendant of the window, not an ancestor. You want an ancestor.

[Mr4D (OOO till 08-26), 2014/05/21 14:25:03]

I am not using this window - I am using it's parent. For the sake of argument I
should however even be able to go higher. Done.

+  aura::Window* parent = content->parent();
+  EXPECT_TRUE(web_contents->should_normally_be_visible());
+  parent->Hide();

[ncarter (slow), 2014/05/20 21:42:26]

Does this actually minimize the parent? Wasn't the original bug that Hide() was
never being called on the parent?

[Mr4D (OOO till 08-26), 2014/05/20 21:48:56]

Well - If I understand sky's request correctly, we have a test that a minimize
triggers a hide. As such only a test is required which proofs that the hide
triggers a content hide as well.

[sky, 2014/05/20 23:33:51]

Exactly.

+  EXPECT_FALSE(web_contents->should_normally_be_visible());
+  parent->Show();
+  EXPECT_TRUE(web_contents->should_normally_be_visible());
+}
+
 }  // namespace content