Remove the is_loading_ field from WebContentsImpl

content/test/web_contents_observer_sanity_checker.cc

Index: content/test/web_contents_observer_sanity_checker.cc
diff --git a/content/test/web_contents_observer_sanity_checker.cc b/content/test/web_contents_observer_sanity_checker.cc
index 8ac2dd0f199b84158997e211d665083696d95ce3..170e5888daecc9b207c3518881b8e5f63b761fe1 100644
--- a/content/test/web_contents_observer_sanity_checker.cc
+++ b/content/test/web_contents_observer_sanity_checker.cc
@@ -332,10 +332,21 @@ void WebContentsObserverSanityChecker::WebContentsDestroyed() {
   CHECK(active_media_players_.empty());
 }
 
+void WebContentsObserverSanityChecker::DidStartLoading() {
+  CHECK(!is_loading_);
+  is_loading_ = true;
+}
+
+void WebContentsObserverSanityChecker::DidStopLoading() {
+  CHECK(is_loading_);
+  is_loading_ = false;
+}
+
 WebContentsObserverSanityChecker::WebContentsObserverSanityChecker(
     WebContents* web_contents)
-    : WebContentsObserver(web_contents), web_contents_destroyed_(false) {
-}
+    : WebContentsObserver(web_contents),
+      is_loading_(false),

[nasko, 2016/01/15 01:19:16]

What about window.open() with an URL? Does it also start from non-loading state?

[clamy, 2016/01/15 16:47:44]

I think so. You first create the WebContents then ask it to navigate. In any
case, the WebContentsObserverSanityChecker was not triggered in any of our
browser tests, so I think it should be fine.

+      web_contents_destroyed_(false) {}
 
 WebContentsObserverSanityChecker::~WebContentsObserverSanityChecker() {
   CHECK(web_contents_destroyed_);