[WIP][NotForLand] IDL dictionary support

Source/core/dom/SomeDictionaryUser.cpp

+// NOT FOR LAND
+#include "config.h"
+#include "SomeDictionaryUser.h"
+
+namespace WebCore {
+
+SomeDictionaryUser::SomeDictionaryUser()
+    : m_a(0)
+    , m_hasA(false)
+    , m_b(0)

[haraken, 2014/07/21 16:11:59]

This should be m_b(2). The default value of m_b is 2.

[bashi, 2014/07/22 02:33:58]

Yes. But actually, there is no need for the default value of
SomeDictionaryUser::m_b is the same as the default value of SomeDictionary::m_b.
SomeDictionaryUser is just an user of SomeDictionary and there is no restriction
between the user (SomeDictionaryUser) and a dictionary(SomeDictionary).

The purpose of these fields is just showing how to get/put values from/to
dictionary classes.

+{
+    ScriptWrappable::init(this);
+}
+
+SomeDictionaryUser::~SomeDictionaryUser()
+{
+}
+
+void SomeDictionaryUser::set(const SomeDictionary* someDictionary)
+{
+    if (someDictionary->hasA()) {
+        m_a = someDictionary->a();
+        m_hasA = true;
+    }
+    m_b = someDictionary->b();
+    m_stringMember = someDictionary->stringMember();
+    m_element = someDictionary->element();
+}
+
+SomeDictionary* SomeDictionaryUser::get()
+{
+    SomeDictionary* result = SomeDictionary::create();
+    if (m_hasA)
+        result->setA(m_a);
+    result->setB(m_b);

[haraken, 2014/07/21 16:11:59]

If introducing m_hasB makes things more consistent, I'm fine with introducing
m_hasB, which always returns true.

[bashi, 2014/07/22 02:33:58]

Yes. I should have introduced m_hasB for consistency. I was cutting corners here
because SomeDictionaryUser is never going to land..

+    result->setStringMember(m_stringMember);
+    result->setElement(m_element);
+    return result;
+}
+
+}