PlzNavigate: implement RequestNavigation in the no live renderer case

content/browser/frame_host/navigation_commit_info.h

+// Copyright 2014 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef CONTENT_BROWSER_FRAME_HOST_NAVIGATION_COMMIT_INFO_H_
+#define CONTENT_BROWSER_FRAME_HOST_NAVIGATION_COMMIT_INFO_H_
+
+#include <string>
+
+#include "base/basictypes.h"
+#include "content/public/common/page_transition_types.h"
+#include "content/public/common/referrer.h"
+#include "url/gurl.h"
+
+namespace content {
+
+// Holds the parameters determined during the load of a navigation request

[Charlie Reis, 2014/07/29 17:46:56]

Sounds like this is mainly for the time before commit, so calling it CommitInfo
is a bit confusing to me.

[clamy, 2014/07/30 13:10:24]

I renamed it NavigationBeforeCommitInfo. Is that better?

+// until receiving the response. Initialized on the IO thread, then passed to
+// the UI thread where it should be used to send a FrameMsg_CommitNavigation
+// message to the renderer.
+struct NavigationCommitInfo {
+  NavigationCommitInfo() {};
+
+  // The url that is actually being loaded.
+  GURL navigation_url;
+
+  // A url to be subsequently requested by the renderer as the main resource for
+  // the frame, that will be used to find the associated stream of data.
+  GURL content_url;

[Charlie Reis, 2014/07/29 17:46:56]

I don't understand what this is for.

1) I thought we were going to push the stream to the renderer process, rather
than having the renderer request it.  Is |content_url| a temporary thing?

2) Why is it different from |navigation_url|?  If it's just for uniquely
identifying a stream, why is it a URL and not an identifier?

[clamy, 2014/07/30 13:10:24]

We are using the blob url construct to request the stream (so the parameter was
initially called |blob_url|). Nasko suggested that the fact that we use blob
urls was a detail of implementation, so I switched the parameter name to
content_url. Maybe stream_url would be more explicit?

+
+  // TODO(clamy): Maybe add the redirect chain if needed by the renderer.
+
+  // The navigationStart time to expose to JS for this navigation.
+  // TODO(clamy): Add the other values that matter to the Navigation Timing API.
+  base::TimeTicks browser_navigation_start;
+};
+
+}  // namespace content
+
+#endif  // CONTENT_BROWSER_FRAME_HOST_NAVIGATION_COMMIT_INFO_H_