Allow RemoteFrames to be navigated through targeted navigation.

Source/core/frame/LocalFrame.cpp

 /*
  * Copyright (C) 1998, 1999 Torben Weis <weis@kde.org>
  *                     1999 Lars Knoll <knoll@kde.org>
  *                     1999 Antti Koivisto <koivisto@kde.org>
  *                     2000 Simon Hausmann <hausmann@kde.org>
  *                     2000 Stefan Schimanski <1Stein@gmx.de>
  *                     2001 George Staikos <staikos@kde.org>
  * Copyright (C) 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011 Apple Inc. All rights reserved.
  * Copyright (C) 2005 Alexey Proskuryakov <ap@nypop.com>
  * Copyright (C) 2008 Nokia Corporation and/or its subsidiary(-ies)
@@ skipping 233 matching lines @@
     // client redirect, which is observable via the webNavigation extension api.
     if (isMainFrame() && !m_loader.stateMachine()->committedFirstRealDocumentLoad()) {
         FrameLoadRequest request(&originDocument, url);
         request.resourceRequest().setHasUserGesture(userGestureStatus == UserGestureStatus::Active);
         m_loader.load(request);
     } else {
         m_navigationScheduler.scheduleLocationChange(&originDocument, url.string(), lockBackForwardList);
     }
 }
 
+void LocalFrame::navigate(const FrameLoadRequest& request)
+{
+    // TODO(dcheng): Special case for window.open("about:blank") to ensure it loads synchronously into

[Nate Chapin, 2015/06/08 16:56:22]

I don't love the duplication here. Perhaps we should specify that this version
of navigate() assumes you don't need to start the navigation on a timer, and
just call m_loader.load() unconditionally? We don't need a timer for the sole
callsite you're adding, and we've long had this distinction in blink (i.e.,
things that call FrameLoader directly vs. things that call NavigationScheduler
to start a load).

[nasko, 2015/06/08 17:27:04]

Yeah, not a big fan either. I've modified it as suggested.

+    // a new window. This is our historical behavior, and it's consistent with the creation of
+    // a new iframe with src="about:blank". Perhaps we could get rid of this if we started reporting
+    // the initial empty document's url as about:blank? See crbug.com/471239.
+    // TODO(japhet): This special case is also necessary for behavior asserted by some extensions tests.
+    // Using NavigationScheduler::scheduleNavigationChange causes the navigation to be flagged as a
+    // client redirect, which is observable via the webNavigation extension api.
+    if (isMainFrame() && !m_loader.stateMachine()->committedFirstRealDocumentLoad()) {
+        m_loader.load(request);
+    } else {
+        m_navigationScheduler.scheduleLocationChange(request.originDocument(), request.resourceRequest().url(), request.lockBackForwardList());
+    }
+}
+
 void LocalFrame::reload(ReloadPolicy reloadPolicy, ClientRedirectPolicy clientRedirectPolicy)
 {
     ASSERT(clientRedirectPolicy == NotClientRedirect || reloadPolicy == NormalReload);
     if (clientRedirectPolicy == NotClientRedirect)
         m_loader.reload(reloadPolicy);
     else
         m_navigationScheduler.scheduleReload();
 }
 
 void LocalFrame::detach()
@@ skipping 548 matching lines @@
     , m_textZoomFactor(parentTextZoomFactor(this))
     , m_inViewSourceMode(false)
 {
     if (isLocalRoot())
         m_instrumentingAgents = InstrumentingAgents::create();
     else
         m_instrumentingAgents = localFrameRoot()->m_instrumentingAgents;
 }
 
 } // namespace blink