New TimeToFirstURLJob* navigation metrics now work with PlzNavigate.

content/browser/frame_host/navigation_request.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_REQUEST_H_
 #define CONTENT_BROWSER_FRAME_HOST_NAVIGATION_REQUEST_H_
 
 #include "base/basictypes.h"
 #include "base/memory/ref_counted.h"
 #include "base/memory/scoped_ptr.h"
@@ skipping 92 matching lines @@
   }
 
  private:
   // NavigationURLLoaderDelegate implementation.
   void OnRequestRedirected(
       const net::RedirectInfo& redirect_info,
       const scoped_refptr<ResourceResponse>& response) override;
   void OnResponseStarted(const scoped_refptr<ResourceResponse>& response,
                          scoped_ptr<StreamHandle> body) override;
   void OnRequestFailed(int net_error) override;
+  void AboutToBeginNavigation(base::TimeTicks timestamp) override;

[clamy, 2015/01/27 12:35:03]

How about NavigationStarted? This is called after the navigation begun so
AboutToBeginNavigation seems a bit weird.

[carlosk, 2015/01/27 17:12:03]

Agreed but I don't think NavigationStarted is much better... 

I based my choice on the method names from ResourceDispatcherHostImpl but
something closer to what we actually intend to get from it should be  better.

So, how about AboutToAttemptFirstNetworkRequest?

[clamy, 2015/01/28 13:44:21]

I don't like AboutToAttemptFirstNetworkRequest much. First the AboutTo part:
this function is called asynchronously, so by the time it executes chances are
that we started doing what we were about to do. Second we are sending the
request to the network stack, but we may not be doing an actual network request
if it is cached. And it's fairly long :).

I suggested NavigationStarted because NavigationRequest goes through the
following states:
- NOT_STARTED
- WAITING_FOR_RENDERER_RESPONSE
- STARTED
- RESPONSE_STARTED.

It goes into state STARTED after BeginNavigation has been called (so calling
BeginNavigation means the navigation started).

So I think that something like NavigationStarted, NavigationBegun or
BeginNavigationComplete would be better, since it is about the RDH informing
NavigationRequest that it did what was asked of it (BeginNavigation).

[carlosk, 2015/01/28 15:21:08]

I agree with your point on the AboutTo not being precise as the navigation is
not waiting on this message handling so in fact it might be already under way
when the message is handled. But I don't agree it should relate ti the named
states of NavigationRequest (NR) as the information it conveys regards solely
the NavigationURLLoader. The NR is merely observing through the
NavigationURLLoaderDelegate interface and its state change to STARTED is has no
relation to this notification.

Having this said, BeginNavigationComplete is a good option but I also suggest
BeginNavigationRequested or NavigationRequested. I'm preemptively changing to
NavigationRequested that I find short and clear but please let me know WDYT.

 
   FrameTreeNode* frame_tree_node_;
 
   // Initialized on creation of the NavigationRequest. Sent to the renderer when
   // the navigation is ready to commit.
   // Note: When the navigation is ready to commit, the url in |common_params|
   // will be set to the final navigation url, obtained after following all
   // redirects.
   CommonNavigationParams common_params_;
   const CommitNavigationParams commit_params_;
@@ skipping 10 matching lines @@
   NavigationEntryImpl::RestoreType restore_type_;
   bool is_view_source_;
   int bindings_;
 
   DISALLOW_COPY_AND_ASSIGN(NavigationRequest);
 };
 
 }  // namespace content
 
 #endif  // CONTENT_BROWSER_FRAME_HOST_NAVIGATION_REQUEST_H_