Update an Element's ancestor chain at the end of its activation

Source/core/dom/Document.cpp

 /*
  * Copyright (C) 1999 Lars Knoll (knoll@kde.org)
  *           (C) 1999 Antti Koivisto (koivisto@kde.org)
  *           (C) 2001 Dirk Mueller (mueller@kde.org)
  *           (C) 2006 Alexey Proskuryakov (ap@webkit.org)
  * Copyright (C) 2004, 2005, 2006, 2007, 2008, 2009, 2011, 2012 Apple Inc. All rights reserved.
  * Copyright (C) 2008, 2009 Torch Mobile Inc. All rights reserved. (http://www.torchmobile.com/)
  * Copyright (C) 2008, 2009, 2011, 2012 Google Inc. All rights reserved.
  * Copyright (C) 2010 Nokia Corporation and/or its subsidiary(-ies)
  * Copyright (C) Research In Motion Limited 2010-2011. All rights reserved.
@@ skipping 5437 matching lines @@
     Element* oldActiveElement = activeHoverElement();
     if (oldActiveElement && !request.active()) {
         // We are clearing the :active chain because the mouse has been released.
         for (RenderObject* curr = oldActiveElement->renderer(); curr; curr = curr->parent()) {
             if (curr->node()) {
                 ASSERT(!curr->node()->isTextNode());
                 curr->node()->setActive(false);
                 m_userActionElements.setInActiveChain(curr->node(), false);
             }
         }
+        // The oldActiveElement renderer is null, dropped on :active by setting display: none,
+        // for instance. We still need to clear the ActiveChain as the mouse is released.
+        if (!oldActiveElement->renderer()) {
+            for (Node* node = oldActiveElement; node; node = node->parentNode()) {

[esprehn, 2014/06/17 08:37:20]

This is not correct for Shadow DOM. You need to use
NodeRenderingTraversal::parent(). You should also just be able to modify the
loop above to use that. That also doesn't seem quite right though since you
could change the distribution and break the parentNode() chain. Perhaps
::detach() should clear the bits? What does Firefox do here?

[spartha, 2014/06/17 15:07:00]

I tried this content http://jsbin.com/fasoyuqi on firefox nightly and Chrome
with my patch. As you have correctly pointed out, using parentNode() fails to
restore the style for the shadow DOM elements. The same content on Firefox
nightly is inconclusive as the shadow DOM implementation is still not very
mature. As far as the behaviour without shadow DOM, Firefox restores the non
active state for all the active chain elements at the end of active state.

Thanks for the insight...

[spartha, 2014/06/18 12:40:09]

I have uploaded a patch using NodeRenderingTraversal, PTAL. I quite did not get
the scenario where chaning the distribution could be a problem. 

+                node->setActive(false);
+                m_userActionElements.setInActiveChain(node, false);
+            }
+        }
         setActiveHoverElement(nullptr);
     } else {
         Element* newActiveElement = innerElementInDocument;
         if (!oldActiveElement && newActiveElement && request.active() && !request.touchMove()) {
             // We are setting the :active chain and freezing it. If future moves happen, they
             // will need to reference this chain.
             for (RenderObject* curr = newActiveElement->renderer(); curr; curr = curr->parent()) {
                 if (curr->node() && !curr->isText())
                     m_userActionElements.setInActiveChain(curr->node(), true);
             }
@@ skipping 370 matching lines @@
     visitor->trace(m_compositorPendingAnimations);
     visitor->trace(m_contextDocument);
     visitor->registerWeakMembers<Document, &Document::clearWeakMembers>(this);
     DocumentSupplementable::trace(visitor);
     TreeScope::trace(visitor);
     ContainerNode::trace(visitor);
     ExecutionContext::trace(visitor);
 }
 
 } // namespace WebCore