history.replaceState(..., location.href) clearing page-action icons.

content/browser/web_contents/navigation_controller_impl.h

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #ifndef CONTENT_BROWSER_WEB_CONTENTS_NAVIGATION_CONTROLLER_IMPL_H_
 #define CONTENT_BROWSER_WEB_CONTENTS_NAVIGATION_CONTROLLER_IMPL_H_
 
 #include "build/build_config.h"
 #include "base/compiler_specific.h"
 #include "base/memory/linked_ptr.h"
@@ skipping 138 matching lines @@
   // Notifies us that we just became active. This is used by the WebContentsImpl
   // so that we know to load URLs that were pending as "lazy" loads.
   void SetActive(bool is_active);
 
   // Returns true if the given URL would be an in-page navigation (i.e. only
   // the reference fragment is different) from the "last committed entry". We do
   // not compare it against the "active entry" since the active entry can be
   // pending and in page navigations only happen on committed pages. If there
   // is no last committed entry, then nothing will be in-page.
   //
-  // Special note: if the URLs are the same, it does NOT count as an in-page
-  // navigation. Neither does an input URL that has no ref, even if the rest is
-  // the same. This may seem weird, but when we're considering whether a
-  // navigation happened without loading anything, the same URL would be a
-  // reload, while only a different ref would be in-page (pages can't clear
+  // Special note: if the URLs are the same, it does NOT automatically count as
+  // an in-page navigation. Neither does an input URL that has no ref, even if
+  // the rest is the same. This may seem weird, but when we're considering
+  // whether a navigation happened without loading anything, the same URL could
+  // be a reload, while only a different ref would be in-page (pages can't clear
   // refs without reload, only change to "#" which we don't count as empty).
-  bool IsURLInPageNavigation(const GURL& url) const;
+  bool IsURLInPageNavigation(const GURL& url) const {
+    return IsURLInPageNavigation(url, false);

[Charlie Reis, 2012/08/28 20:06:16]

Isn't overloading usually frowned upon in Chrome?

[Tom Sepez, 2012/08/28 20:17:41]

Yep, but it avoids having to write a comment about "if you don't know what to
pass here because you don't have a suggestion from the renderer, pass false".

+  }
+
+  // The situation is made murkier by history.replaceState(), which could
+  // provide the same URL as part of an in-page navigation, not a reload. So
+  // we need this form which lets this the (untrustworthy) renderer resolve the

[Charlie Reis, 2012/08/28 20:06:16]

nit: which lets the

[Tom Sepez, 2012/08/28 20:17:41]

Done.

+  // ambiguity, but only when the URLs are equal. This should be safe since the

[Charlie Reis, 2012/08/28 20:06:16]

maybe: "URLs are equal (apart from refs)"?

[Tom Sepez, 2012/08/28 20:17:41]

Current impl is strict equality.

+  // origin isn't changing.
+  bool IsURLInPageNavigation(const GURL& url, bool renderer_says_in_page) const;
 
   // Sets the SessionStorageNamespace for the given |partition_id|. This is
   // used during initialization of a new NavigationController to allow
   // pre-population of the SessionStorageNamespace objects. Session restore,
   // prerendering, and the implementaion of window.open() are the primary users
   // of this API.
   //
   // Calling this function when a SessionStorageNamespace has already been
   // associated with a |partition_id| will CHECK() fail.
   void SetSessionStorageNamespace(
@@ skipping 168 matching lines @@
   static size_t max_entry_count_for_testing_;
 
   // If a repost is pending, its type (RELOAD or RELOAD_IGNORING_CACHE),
   // NO_RELOAD otherwise.
   ReloadType pending_reload_;
 
   DISALLOW_COPY_AND_ASSIGN(NavigationControllerImpl);
 };
 
 #endif  // CONTENT_BROWSER_WEB_CONTENTS_NAVIGATION_CONTROLLER_IMPL_H_