Implement "create an element" when sync for Custom Element V1

third_party/WebKit/Source/core/dom/Document.h

 /*
  * 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, 2010, 2012 Apple Inc. All rights reserved.
  * Copyright (C) 2008, 2009 Torch Mobile Inc. All rights reserved. (http://www.torchmobile.com/)
  * Copyright (C) 2010 Nokia Corporation and/or its subsidiary(-ies)
  * Copyright (C) 2011 Google Inc. All rights reserved.
  *
@@ skipping 291 matching lines @@
     DocumentFragment* createDocumentFragment();
     Text* createTextNode(const String& data);
     Comment* createComment(const String& data);
     CDATASection* createCDATASection(const String& data, ExceptionState&);
     ProcessingInstruction* createProcessingInstruction(const String& target, const String& data, ExceptionState&);
     Attr* createAttribute(const AtomicString& name, ExceptionState&);
     Attr* createAttributeNS(const AtomicString& namespaceURI, const AtomicString& qualifiedName, ExceptionState&, bool shouldIgnoreNamespaceChecks = false);
     Node* importNode(Node* importedNode, bool deep, ExceptionState&);
     Element* createElementNS(const AtomicString& namespaceURI, const AtomicString& qualifiedName, ExceptionState&);
     Element* createElement(const QualifiedName&, CreateElementFlags);
+    Element* createElement(const QualifiedName&, CreateElementFlags, ExceptionState&);

[dominicc (has gone to gerrit), 2016/06/10 06:13:28]

OK now I'm really confused. Who uses *this*?

I think we're adding too many methods called createElement, and we don't have
good layering. Instead of having everything calling everything and using flags
to control where the recursion bottoms out, we need a each method to have a
distinct responsibility.

So, what do we absolutely need:

Document::createElement(NS) for the DOM callbacks. These can call
CustomElement::createElementSync.
Document::createElement for cloning, etc. This calls HTMLElementFactory.
CustomElement::createElementSync which swallows (and reports) exceptions. This
is called by the parser. Calls HTMLElementFactory.
HTMLElementFactory::create. This actually creates elements.

The only question is where create-and-queue-upgrade happens. That could be
HTMLElementFactory, but it needs an option to turn it off (for
createElementSync.) Or it could be Document::createElement.

is="" means probably HTMLElementFactory should do *less* and *not* schedule
upgrade. The reason is HTMLElementFactory::create does NOT get the attribute
list, so it CAN NOT KNOW which definition to queue upgrade with.

So, again, I really think we need fewer createElement methods and fewer flags to
HTMLElementFactory.

I am fine to go in and do this refactoring. But it would be get this patch
committed. So let's not make the clean up more difficult by adding more
createElement methods, more use of flags in this patch.

[kojii, 2016/06/10 19:01:45]

Done. took some refactoring, but not all yet.

 
     Element* elementFromPoint(int x, int y) const;
     HeapVector<Member<Element>> elementsFromPoint(int x, int y) const;
     Range* caretRangeFromPoint(int x, int y);
     Element* scrollingElement();
 
     String readyState() const;
 
     AtomicString characterSet() const { return Document::encodingName(); }
 
@@ skipping 1130 matching lines @@
 DEFINE_TYPE_CASTS(TreeScope, Document, document, true, true);
 
 } // namespace blink
 
 #ifndef NDEBUG
 // Outside the WebCore namespace for ease of invocation from gdb.
 CORE_EXPORT void showLiveDocumentInstances();
 #endif
 
 #endif // Document_h