Removing ContentViewCore dependencies from few functions which acts as direct wrapper to WebContents

content/public/browser/web_contents.h

Index: content/public/browser/web_contents.h
diff --git a/content/public/browser/web_contents.h b/content/public/browser/web_contents.h
index b16330626f7ce925b91c97466f6b85b02f14ef71..46c8dc4837923771f1004f22cb78659904cd38d9 100644
--- a/content/public/browser/web_contents.h
+++ b/content/public/browser/web_contents.h
@@ -53,6 +53,7 @@ class RenderProcessHost;
 class RenderViewHost;
 class RenderWidgetHostView;
 class SiteInstance;
+class WebContentsAndroid;
 class WebContentsDelegate;
 struct CustomContextMenuContext;
 struct DropData;
@@ -579,6 +580,7 @@ class WebContents : public PageNavigator,
   CONTENT_EXPORT static WebContents* FromJavaWebContents(
       jobject jweb_contents_android);
   virtual base::android::ScopedJavaLocalRef<jobject> GetJavaWebContents() = 0;
+  virtual WebContentsAndroid* GetWebContentsAndroid() = 0;

[Avi (use Gerrit), 2014/08/08 15:23:25]

I don't get the relationship here.

From line 582 I see that there's something called a "JavaWebContents" that's a
jobject. From line 581 I see the opposite of that function, but that calls the
jobject a "web_contents_android". And from line 583 I see that a
"WebContentsAndroid" is an entirely different thing, a C++ class.

Can you please clarify your naming? Can we have only one special Android class
in correspondence here?

[AKVT, 2014/08/08 16:15:51]

Thanks, I think GetNativeWebContents() suits correctly. Please share your
opinion ?

[Avi (use Gerrit), 2014/08/08 16:26:04]

The name GetWebContentsAndroid is fine if you rename the parameter on line 580
(to java_web_contents or something like that). That doesn't solve the real
problem.

Before, it was clear that there was an Android corresponding object. There were
two functions, one to go WebContents → Android object (GetJavaWebContents), and
one to go Android object → WebContents (FromJavaWebContents).

OK, I can deal with that.

Now there are three functions?

You're proposing that there are two different objects that WebContents is in 1:1
relationship with. It's not immediately obvious via naming what the difference
is, and it's weird to me that one of them has two-way conversion while the other
has only one-way conversion (WebContents → WebContentsAndroid).

 #elif defined(OS_MACOSX)
   // The web contents view assumes that its view will never be overlapped by
   // another view (either partially or fully). This allows it to perform