PlzNavigate: identify same-page browser-initiated navigation.

components/subresource_filter/content/browser/content_subresource_filter_driver_factory.cc

Index: components/subresource_filter/content/browser/content_subresource_filter_driver_factory.cc
diff --git a/components/subresource_filter/content/browser/content_subresource_filter_driver_factory.cc b/components/subresource_filter/content/browser/content_subresource_filter_driver_factory.cc
index 1f53fd3d92feca16fd1a9692146b69372fa14cd4..7f854997eab09beefe183a4c8c40bae37da8bf93 100644
--- a/components/subresource_filter/content/browser/content_subresource_filter_driver_factory.cc
+++ b/components/subresource_filter/content/browser/content_subresource_filter_driver_factory.cc
@@ -209,7 +209,8 @@ void ContentSubresourceFilterDriverFactory::RenderFrameDeleted(
 
 void ContentSubresourceFilterDriverFactory::ReadyToCommitNavigation(
     content::NavigationHandle* navigation_handle) {
-  DCHECK(!navigation_handle->IsSamePage());
+  if (navigation_handle->IsSamePage())

[clamy, 2017/01/09 16:58:48]

I don't think we want to call ReadyToCommitNavigation for a same page
navigation.

[arthursonzogni, 2017/01/12 17:32:29]

I don't know. I will filter the call to ReadyToCommitNavigation for the moment
and we will see later what should be done.

[nasko, 2017/01/13 02:51:09]

This means that users of WebContentsObserver will have to account for this
difference in behavior, which might make their code more complex. I wonder what
would be the issue with dispatching ReadyToCommitNavigation for same-page cases?

+    return;
   content::RenderFrameHost* render_frame_host =
       navigation_handle->GetRenderFrameHost();
   GURL url = navigation_handle->GetURL();