The content api for shouldoverrideurlloading

content/public/renderer/content_renderer_client.h

Index: content/public/renderer/content_renderer_client.h
diff --git a/content/public/renderer/content_renderer_client.h b/content/public/renderer/content_renderer_client.h
index b5b11cd11beb9e47e1ac44254c3525d37d88da2b..904242331a140ee26984bc7cbf0a05b8a78a52e1 100644
--- a/content/public/renderer/content_renderer_client.h
+++ b/content/public/renderer/content_renderer_client.h
@@ -49,6 +49,7 @@ struct WebURLError;
 
 namespace content {
 
+class DocumentState;
 class RenderView;
 class SynchronousCompositor;
 struct KeySystemInfo;
@@ -168,8 +169,11 @@ class CONTENT_EXPORT ContentRendererClient {
   virtual bool AllowPopup();
 
   // Returns true if the navigation was handled by the embedder and should be
-  // ignored by WebKit. This method is used by CEF.
-  virtual bool HandleNavigation(WebKit::WebFrame* frame,
+  // ignored by WebKit. This method is used by CEF and android_webview.
+  virtual bool HandleNavigation(RenderView* view,

[darin (slow to review), 2013/10/24 03:59:34]

Our intent is to delete this method. It is better to use the browser-side
ResourceThrottle hook. Note also, that the OOPIF infrastructure is moving way
more of the navigation system to the browser process. You are much better off
sitting on top of the ResourceThrottle infrastructure. What does that not
provide that you need?

[joth, 2013/11/22 07:03:20]

As a compromise to get this moving forward, how about we land this change but
wrap the whole function in ifdef OS_ANDROID.
- we'll open a bug to move to the ResourceThrottle if at all possible.
- Moving to throttle will be much more practical to do once we have unforked and
got automated tests covering this very bug-sensitive piece (not to mention we
can switch our canary builds back to master branch then, to do get manual QA and
dogfood coverage)
- If the resource throttle does prove unworkable, then I'm almost certain the
OOPIF changes to move nav to browser will completely remove our need for this
hook anyway (we'd much much rather get navigation layer hooks right there in the
browser than have to do our own render->browser hops and jumps)

+                                DocumentState* document_state,
+                                int opener_id,
+                                WebKit::WebFrame* frame,
                                 const WebKit::WebURLRequest& request,
                                 WebKit::WebNavigationType type,
                                 WebKit::WebNavigationPolicy default_policy,