Componentize chrome/browser/rlz

Componentize chrome/browser/rlz

Add new class RLZTrackerDelegate that abstract access to embedder
specific singletons and informations and provide an implementation for
Chrome based on the previous implementation.

Split rlz_unittest.cc in tests of the RLZTracker and Chrome
implementation of the RLZTrackerDelegate interface and move the
RLZTracker tests into the component.

Add a new gyp/gn variable "enable_rlz_support". This variable is true on
the platforms that support RLZ (currently Windows, Mac, iOS and
ChromeOS). Use it to build library and unit tests even when the RLZ
support is not enabled in the Chrome binary (this is still controlled by
"enable_rlz" and depends on the branding).

Enable the tests on iOS and convert rlz_tracker_ios.mm to a C++ file.

TEST=Run unit_tests and components_unittests on a platform that supports
RLZ (Windows, Mac, iOS or ChromeOS) and check that they pass. Then build
Chrome with "enable_rlz" and manually checks that RLZ are sent with
searches as expected.

BUG=504841, 508148
Committed: https://crrev.com/d967d55cf7a0f78e6ace9ec681d2d082e631810b
Cr-Commit-Position: refs/heads/master@{#339002}

[sdefresne, 2015-07-01 15:59:40]

sdefresne@chromium.org changed reviewers:
+ rogerta@chromium.org, thakis@chromium.org

[sdefresne, 2015-07-01 15:59:40]

rogerta: please take a look
 1. as OWNERS of chrome/browser/rlz & components/rlz
 2. for new DEPS on rlz from components/rlz

thakis: please take a look
 1. as OWNERS of //chrome/browser for changes after RLZTracker API change
 2. for new DEPS on base, net, ui/base from components/rlz

Note that patchset #1 set "enable_rlz=1" when building Chromium and patchset #2
revert this change (the goal being to use the bots to test the architecture I
don't have access to).

[Roger Tawa OOO till Jul 10th, 2015-07-02 18:22:28]

lgtm

Thanks for the cleanup Sylvain.  The one thing I wonder is if the tests should
always be compiled and run, even if enable_rlz is false.  This flag should
control whether rlz is baked into the chrome/chromium binary or not.  But it
would be much convenient for developers if we always built the rlz component and
ran its tests.  What do you think?

https://codereview.chromium.org/1212163011/diff/40001/chrome/browser/rlz/chro...
File chrome/browser/rlz/chrome_rlz_tracker_delegate_unittest.cc (right):

https://codereview.chromium.org/1212163011/diff/40001/chrome/browser/rlz/chro...
chrome/browser/rlz/chrome_rlz_tracker_delegate_unittest.cc:18: class
ChromeRLZTrackerDelegateTest : public testing::Test {
I'm surprised this test case does not need to derive from
ChromeRenderViewHostTestHarness.  All the better.  Do you know why it was
necessary before?

https://codereview.chromium.org/1212163011/diff/40001/components/BUILD.gn
File components/BUILD.gn (right):

https://codereview.chromium.org/1212163011/diff/40001/components/BUILD.gn#new...
components/BUILD.gn:312: }
I wonder if the tests should always be compiled and run.  Is there any advantage
to no doing so?

https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
File components/rlz/rlz_tracker.cc (right):

https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
components/rlz/rlz_tracker.cc:179: if (!delegate_) {
DCHECK(!delegate_);  ?  When would we want to change the delegate?

https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
File components/rlz/rlz_tracker_unittest.cc (right):

https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
components/rlz/rlz_tracker_unittest.cc:173: // using RLZTracker::Observe;
Remove line.

[sdefresne, 2015-07-03 16:16:37]

Please take another look.

https://codereview.chromium.org/1212163011/diff/40001/chrome/browser/rlz/chro...
File chrome/browser/rlz/chrome_rlz_tracker_delegate_unittest.cc (right):

https://codereview.chromium.org/1212163011/diff/40001/chrome/browser/rlz/chro...
chrome/browser/rlz/chrome_rlz_tracker_delegate_unittest.cc:18: class
ChromeRLZTrackerDelegateTest : public testing::Test {
On 2015/07/02 at 18:22:28, Roger Tawa wrote:
> I'm surprised this test case does not need to derive from
ChromeRenderViewHostTestHarness.  All the better.  Do you know why it was
necessary before?

It was required to send / receive notifications. Since I directly call the
observer Observe() method and do not use the notification service, I don't have
to derive from ChromeRenderViewHostTestHarness.

https://codereview.chromium.org/1212163011/diff/40001/components/BUILD.gn
File components/BUILD.gn (right):

https://codereview.chromium.org/1212163011/diff/40001/components/BUILD.gn#new...
components/BUILD.gn:312: }
On 2015/07/02 at 18:22:28, Roger Tawa wrote:
> I wonder if the tests should always be compiled and run.  Is there any
advantage to no doing so?

Make sense. This will require changing client code to #include rlz_tracker.h
conditionally depending on ENABLE_RLZ macro instead of checking the macro in the
header though.

https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
File components/rlz/rlz_tracker.cc (right):

https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
components/rlz/rlz_tracker.cc:179: if (!delegate_) {
On 2015/07/02 at 18:22:28, Roger Tawa wrote:
> DCHECK(!delegate_);  ?  When would we want to change the delegate?

RLZTracker is a singleton that is not tied to the Profile, however it is
initialized upon Profile creation (init was previously called
InitRlzFromProfileDelayed). I've put the calls to SetRlzDelegate() just before
the call to InitRlzDelayed. For desktop it means it is called during
PreMainMessageLoopRunImpl, but on ChromeOs it is called from
chrome/browser/chromeos/login/session/user_session_manager.cc and I think this
method may be called multiple time.

So the check is to avoid breaking the ChromeOS build.

[Roger Tawa OOO till Jul 10th, 2015-07-03 17:25:20]

still lgtm

Thanks for building and running the tests all the time.  In fact, I think this
has uncovered some problems with the tests themselves.  For example, the red
bots are rlz test breaks.  I think the problem is that some tests hard code
string like kHomepageFirstSeach when they should use HomepageFirstSeach(). 
Yes/no?

https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
File components/rlz/rlz_tracker.cc (right):

https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
components/rlz/rlz_tracker.cc:179: if (!delegate_) {
On 2015/07/03 16:16:37, sdefresne wrote:
> On 2015/07/02 at 18:22:28, Roger Tawa wrote:
> > DCHECK(!delegate_);  ?  When would we want to change the delegate?
> 
> RLZTracker is a singleton that is not tied to the Profile, however it is
> initialized upon Profile creation (init was previously called
> InitRlzFromProfileDelayed). I've put the calls to SetRlzDelegate() just before
> the call to InitRlzDelayed. For desktop it means it is called during
> PreMainMessageLoopRunImpl, but on ChromeOs it is called from
> chrome/browser/chromeos/login/session/user_session_manager.cc and I think this
> method may be called multiple time.
> 
> So the check is to avoid breaking the ChromeOS build.

Acknowledged.

[sdefresne, 2015-07-06 14:06:46]

On 2015/07/03 at 17:25:20, rogerta wrote:
> still lgtm
> 
> Thanks for building and running the tests all the time.  In fact, I think this
has uncovered some problems with the tests themselves.  For example, the red
bots are rlz test breaks.  I think the problem is that some tests hard code
string like kHomepageFirstSeach when they should use HomepageFirstSeach(). 
Yes/no?

No, the error was due to me calling the wrong method when registering the
callback (SetOmniboxSearchCallback was called twice instead of
SetOmniboxSearchCallback and SetHomepageSearchCallback being both called). The
test was only failing on non-iOS platforms (since iOS does not register that
second callback), so it's good that the tests were enabled on all platform
always.

> 
>
https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
> File components/rlz/rlz_tracker.cc (right):
> 
>
https://codereview.chromium.org/1212163011/diff/40001/components/rlz/rlz_trac...
> components/rlz/rlz_tracker.cc:179: if (!delegate_) {
> On 2015/07/03 16:16:37, sdefresne wrote:
> > On 2015/07/02 at 18:22:28, Roger Tawa wrote:
> > > DCHECK(!delegate_);  ?  When would we want to change the delegate?
> > 
> > RLZTracker is a singleton that is not tied to the Profile, however it is
> > initialized upon Profile creation (init was previously called
> > InitRlzFromProfileDelayed). I've put the calls to SetRlzDelegate() just
before
> > the call to InitRlzDelayed. For desktop it means it is called during
> > PreMainMessageLoopRunImpl, but on ChromeOs it is called from
> > chrome/browser/chromeos/login/session/user_session_manager.cc and I think
this
> > method may be called multiple time.
> > 
> > So the check is to avoid breaking the ChromeOS build.
> 
> Acknowledged.

[sdefresne, 2015-07-06 14:09:28]

thakis: ping?

[sdefresne, 2015-07-07 08:09:52]

sdefresne@chromium.org changed reviewers:
+ cpu@chromium.org, phajdan.jr@chromium.org, pkasting@chromium.org,
scottmg@chromium.org, stevenjb@chromium.org
- thakis@chromium.org

[sdefresne, 2015-07-07 08:09:54]

Trying multiple OWNERS as root OWNERS were unresponsive. Can you each review the
files highlighted?

stevenjb:
  chrome/browser/chromeos/chrome_browser_main_chromeos.cc
  chrome/browser/chromeos/login/login_utils_browsertest.cc
  chrome/browser/chromeos/login/session/user_session_manager.cc
  chrome/browser/ui/app_list/app_list_controller_delegate.cc
pkasting:
  chrome/browser/search_engines/template_url_service_factory.cc
  chrome/browser/search_engines/ui_thread_search_terms_data.cc
  chrome/browser/ui/browser_commands.cc
  chrome/browser/ui/startup/startup_browser_creator_impl.cc
cpu:
  chrome/browser/BUILD.gn
  chrome/browser/DEPS
  chrome/browser/browser_shutdown.cc
  chrome/browser/chrome_browser_main.cc
  rlz/test/rlz_test_helpers.h
scottmg:
  build/common.gypi
  build/config/features.gni
phajdan.jr:
  chrome/test/BUILD.gn

[Paweł Hajdan Jr., 2015-07-07 10:49:49]

chrome/test LGTM

[cpu_(ooo_6.6-7.5), 2015-07-07 17:08:04]

I reviewed my files and they look good. Ping me at the end since my stamp is
valid for all src/ .

Please engage with QA (anantha@) to do a manual test. They have a comprehensive
script. This code means $$$ for some people some we need to do manual
validation.

[Peter Kasting, 2015-07-07 18:43:03]

LGTM.  It's also possible to TBR various owners when you're making minor,
purely-mechanical changes to their files.

[sdefresne, 2015-07-08 11:26:48]

sdefresne@chromium.org changed reviewers:
+ jochen@chromium.org, xiyuan@chromium.org
- scottmg@chromium.org, stevenjb@chromium.org

[sdefresne, 2015-07-08 11:26:49]

Removing OWNERS currently in vacation, and picking other OWNERS (xiyuan,
jochen). Can you look at the following files?

xiyuan:
  chrome/browser/chromeos/chrome_browser_main_chromeos.cc
  chrome/browser/chromeos/login/login_utils_browsertest.cc
  chrome/browser/chromeos/login/session/user_session_manager.cc
  chrome/browser/ui/app_list/app_list_controller_delegate.cc
jochen:
  build/common.gypi
  build/config/features.gni

[jochen (gone - plz use gerrit), 2015-07-08 13:24:14]

lgtm

[xiyuan, 2015-07-08 14:39:20]

lgtm

[sdefresne, 2015-07-08 14:42:29]

cpu: all other OWNERS have approved the CL, pinging you as requested.

[cpu_(ooo_6.6-7.5), 2015-07-08 18:21:08]

lgtm

[sdefresne, 2015-07-09 08:20:45]

The CQ bit was checked by sdefresne@chromium.org

[sdefresne, 2015-07-09 08:20:45]

The patchset sent to the CQ was uploaded after l-g-t-m from rogerta@chromium.org
Link to the patchset: https://codereview.chromium.org/1212163011/#ps120001
(title: "Fix "gn gen ..."")

[commit-bot: I haz the power, 2015-07-09 08:20:50]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1212163011/120001

[commit-bot: I haz the power, 2015-07-09 08:23:08]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-07-09 08:23:10]

Try jobs failed on following builders:
  linux_chromium_compile_dbg_32_ng on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)
  ios_dbg_simulator_ninja on tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/ios_dbg_simulator...)
  mac_chromium_compile_dbg_ng on tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_comp...)
  mac_chromium_rel_ng on tryserver.chromium.mac (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.mac/builders/mac_chromium_rel_...)

(exceeded global retry quota)

[sdefresne, 2015-07-09 09:13:10]

The CQ bit was checked by sdefresne@chromium.org

[sdefresne, 2015-07-09 09:13:11]

The patchset sent to the CQ was uploaded after l-g-t-m from jochen@chromium.org,
xiyuan@chromium.org, cpu@chromium.org, rogerta@chromium.org,
phajdan.jr@chromium.org, pkasting@chromium.org
Link to the patchset: https://codereview.chromium.org/1212163011/#ps140001
(title: "Rebase")

[commit-bot: I haz the power, 2015-07-09 09:13:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1212163011/140001

[commit-bot: I haz the power, 2015-07-09 10:22:53]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-07-09 10:22:55]

Failed to apply patch for components/rlz/rlz_tracker.cc:
While running git apply --index -3 -p1;
  error: patch failed: chrome/browser/rlz/rlz.cc:6
  Falling back to three-way merge...
  Applied patch to 'components/rlz/rlz_tracker.cc' with conflicts.
  U components/rlz/rlz_tracker.cc

Patch:   NR  chrome/browser/rlz/rlz.cc->components/rlz/rlz_tracker.cc
Index: components/rlz/rlz_tracker.cc
diff --git a/chrome/browser/rlz/rlz.cc b/components/rlz/rlz_tracker.cc
similarity index 65%
rename from chrome/browser/rlz/rlz.cc
rename to components/rlz/rlz_tracker.cc
index
71d459b17942948d2f4d3fa5e1f079c218a0a83d..0f32d00d7e86b5c456bed179c6b6d837cd7b8f97
100644
--- a/chrome/browser/rlz/rlz.cc
+++ b/components/rlz/rlz_tracker.cc
@@ -6,61 +6,20 @@
 // with or without the DLL being present. If the DLL is not present the
 // functions do nothing and just return false.
 
-#include "chrome/browser/rlz/rlz.h"
+#include "components/rlz/rlz_tracker.h"
 
 #include <algorithm>
 
 #include "base/bind.h"
-#include "base/command_line.h"
 #include "base/message_loop/message_loop.h"
-#include "base/prefs/pref_service.h"
 #include "base/strings/string_util.h"
 #include "base/strings/utf_string_conversions.h"
 #include "base/trace_event/trace_event.h"
-#include "chrome/browser/browser_process.h"
-#include "chrome/browser/chrome_notification_types.h"
-#include "chrome/browser/google/google_brand.h"
-#include "chrome/browser/prefs/session_startup_pref.h"
-#include "chrome/browser/search_engines/template_url_service_factory.h"
-#include "chrome/browser/ui/startup/startup_browser_creator.h"
-#include "chrome/common/chrome_switches.h"
-#include "chrome/common/pref_names.h"
 #include "components/google/core/browser/google_util.h"
-#include "components/omnibox/omnibox_log.h"
-#include "components/search_engines/template_url.h"
-#include "components/search_engines/template_url_service.h"
-#include "content/public/browser/browser_thread.h"
-#include "content/public/browser/navigation_controller.h"
-#include "content/public/browser/navigation_details.h"
-#include "content/public/browser/navigation_entry.h"
-#include "content/public/browser/notification_service.h"
+#include "components/rlz/rlz_tracker_delegate.h"
 #include "net/http/http_util.h"
 
-#if defined(OS_WIN)
-#include "chrome/installer/util/google_update_settings.h"
-#else
-namespace GoogleUpdateSettings {
-static bool GetLanguage(base::string16* language) {
-  // TODO(thakis): Implement.
-  NOTIMPLEMENTED();
-  return false;
-}
-
-// The referral program is defunct and not used. No need to implement these
-// functions on non-Win platforms.
-static bool GetReferral(base::string16* referral) {
-  return true;
-}
-static bool ClearReferral() {
-  return true;
-}
-}  // namespace GoogleUpdateSettings
-#endif
-
-using content::BrowserThread;
-using content::NavigationEntry;
-using content::NavigationController;
-
+namespace rlz {
 namespace {
 
 // Maximum and minimum delay for financial ping we would allow to be set
through
@@ -68,10 +27,6 @@ namespace {
 const base::TimeDelta kMaxInitDelay = base::TimeDelta::FromSeconds(200);
 const base::TimeDelta kMinInitDelay = base::TimeDelta::FromSeconds(20);
 
-bool IsBrandOrganic(const std::string& brand) {
-  return brand.empty() || google_brand::IsOrganic(brand);
-}
-
 void RecordProductEvents(bool first_run,
                          bool is_google_default_search,
                          bool is_google_homepage,
@@ -192,7 +147,7 @@ bool SendFinancialPing(const std::string& brand,
 
 }  // namespace
 
-RLZTracker* RLZTracker::tracker_ = NULL;
+RLZTracker* RLZTracker::tracker_ = nullptr;
 
 // static
 RLZTracker* RLZTracker::GetInstance() {
@@ -205,7 +160,6 @@ RLZTracker::RLZTracker()
       is_google_default_search_(false),
       is_google_homepage_(false),
       is_google_in_startpages_(false),
-      worker_pool_token_(BrowserThread::GetBlockingPool()->GetSequenceToken()),
       already_ran_(false),
       omnibox_used_(false),
       homepage_used_(false),
@@ -217,6 +171,19 @@ RLZTracker::~RLZTracker() {
 }
 
 // static
+void RLZTracker::SetRlzDelegate(scoped_ptr<RLZTrackerDelegate> delegate) {
+  GetInstance()->SetDelegate(delegate.Pass());
+}
+
+void RLZTracker::SetDelegate(scoped_ptr<RLZTrackerDelegate> delegate) {
+  DCHECK(delegate);
+  if (!delegate_) {
+    delegate_ = delegate.Pass();
+    worker_pool_token_ = delegate_->GetBlockingPool()->GetSequenceToken();
+  }
+}
+
+// static
 bool RLZTracker::InitRlzDelayed(bool first_run,
                                 bool send_ping_immediately,
                                 base::TimeDelta delay,
@@ -228,56 +195,6 @@ bool RLZTracker::InitRlzDelayed(bool first_run,
                              is_google_in_startpages);
 }
 
-// static
-bool RLZTracker::InitRlzFromProfileDelayed(Profile* profile,
-                                           bool first_run,
-                                           bool send_ping_immediately,
-                                           base::TimeDelta delay) {
-  bool is_google_default_search = false;
-  TemplateURLService* template_url_service =
-      TemplateURLServiceFactory::GetForProfile(profile);
-  if (template_url_service) {
-    const TemplateURL* url_template =
-        template_url_service->GetDefaultSearchProvider();
-    is_google_default_search =
-        url_template && url_template->url_ref().HasGoogleBaseURLs(
-            template_url_service->search_terms_data());
-  }
-
-  PrefService* pref_service = profile->GetPrefs();
-  bool is_google_homepage = google_util::IsGoogleHomePageUrl(
-      GURL(pref_service->GetString(prefs::kHomePage)));
-
-  bool is_google_in_startpages = false;
-#if !defined(OS_IOS)
-  // iOS does not have a notion of startpages.
-  SessionStartupPref session_startup_prefs =
-      StartupBrowserCreator::GetSessionStartupPref(
-          *base::CommandLine::ForCurrentProcess(), profile);
-  if (session_startup_prefs.type == SessionStartupPref::URLS) {
-    is_google_in_startpages =
-        std::count_if(session_startup_prefs.urls.begin(),
-                      session_startup_prefs.urls.end(),
-                      google_util::IsGoogleHomePageUrl) > 0;
-  }
-#endif
-
-  if (!InitRlzDelayed(first_run, send_ping_immediately, delay,
-                      is_google_default_search, is_google_homepage,
-                      is_google_in_startpages)) {
-    return false;
-  }
-
-#if !defined(OS_IOS)
-  // Prime the RLZ cache for the home page access point so that its avaiable
-  // for the startup page if needed (i.e., when the startup page is set to
-  // the home page).
-  GetAccessPointRlz(ChromeHomePage(), NULL);
-#endif  // !defined(OS_IOS)
-
-  return true;
-}
-
 bool RLZTracker::Init(bool first_run,
                       bool send_ping_immediately,
                       base::TimeDelta delay,
@@ -291,46 +208,57 @@ bool RLZTracker::Init(bool first_run,
   send_ping_immediately_ = send_ping_immediately;
 
   // Enable zero delays for testing.
-  if (base::CommandLine::ForCurrentProcess()->HasSwitch(::switches::kTestType))
+  if (delegate_->ShouldEnableZeroDelayForTesting())
     EnableZeroDelayForTesting();
 
   delay = std::min(kMaxInitDelay, std::max(min_init_delay_, delay));
 
-  if (google_brand::GetBrand(&brand_) && !IsBrandOrganic(brand_)) {
+  if (delegate_->GetBrand(&brand_) && !delegate_->IsBrandOrganic(brand_)) {
     // Register for notifications from the omnibox so that we can record when
     // the user performs a first search.
-    registrar_.Add(this, chrome::NOTIFICATION_OMNIBOX_OPENED_URL,
-                   content::NotificationService::AllSources());
+    delegate_->SetOmniboxSearchCallback(
+        base::Bind(&RLZTracker::RecordFirstSearch, base::Unretained(this),
+                   ChromeOmnibox()));
 
 #if !defined(OS_IOS)
     // Register for notifications from navigations, to see if the user has used
     // the home page.
-    registrar_.Add(this, content::NOTIFICATION_NAV_ENTRY_COMMITTED,
-                   content::NotificationService::AllSources());
-#endif  // !defined(OS_IOS)
+    delegate_->SetHomepageSearchCallback(
+        base::Bind(&RLZTracker::RecordFirstSearch, base::Unretained(this),
+                   ChromeHomePage()));
+#endif
   }
-  google_brand::GetReactivationBrand(&reactivation_brand_);
-
-  net::URLRequestContextGetter* context_getter =
-      g_browser_process->system_request_context();
+  delegate_->GetReactivationBrand(&reactivation_brand_);
 
-  // Could be NULL; don't run if so.  RLZ will try again next restart.
+  // Could be null; don't run if so.  RLZ will try again next restart.
+  net::URLRequestContextGetter* context_getter =
delegate_->GetRequestContext();
   if (context_getter) {
     rlz_lib::SetURLRequestContext(context_getter);
     ScheduleDelayedInit(delay);
   }
 
+#if !defined(OS_IOS)
+  // Prime the RLZ cache for the home page access point so that its avaiable
+  // for the startup page if needed (i.e., when the startup page is set to
+  // the home page).
+  GetAccessPointRlz(ChromeHomePage(), nullptr);
+#endif  // !defined(OS_IOS)
+
   return true;
 }
 
+void RLZTracker::Cleanup() {
+  rlz_cache_.clear();
+  if (delegate_)
+    delegate_->Cleanup();
+}
+
 void RLZTracker::ScheduleDelayedInit(base::TimeDelta delay) {
   // The RLZTracker is a singleton object that outlives any runnable tasks
   // that will be queued up.
-  BrowserThread::GetBlockingPool()->PostDelayedSequencedWorkerTask(
-      worker_pool_token_,
-      FROM_HERE,
-      base::Bind(&RLZTracker::DelayedInit, base::Unretained(this)),
-      delay);
+  delegate_->GetBlockingPool()->PostDelayedSequencedWorkerTask(
+      worker_pool_token_, FROM_HERE,
+      base::Bind(&RLZTracker::DelayedInit, base::Unretained(this)), delay);
 }
 
 void RLZTracker::DelayedInit() {
@@ -338,7 +266,7 @@ void RLZTracker::DelayedInit() {
 
   // For organic brandcodes do not use rlz at all. Empty brandcode usually
   // means a chromium install. This is ok.
-  if (!I…
(message too large)

[commit-bot: I haz the power, 2015-07-09 10:22:57]

The CQ bit was unchecked by commit-bot@chromium.org

[sdefresne, 2015-07-09 12:52:05]

The CQ bit was checked by sdefresne@chromium.org

[sdefresne, 2015-07-09 12:52:06]

The patchset sent to the CQ was uploaded after l-g-t-m from jochen@chromium.org,
xiyuan@chromium.org, cpu@chromium.org, rogerta@chromium.org,
phajdan.jr@chromium.org, pkasting@chromium.org
Link to the patchset: https://codereview.chromium.org/1212163011/#ps160001
(title: "Rebase, round 2")

[commit-bot: I haz the power, 2015-07-09 12:52:35]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1212163011/160001

[commit-bot: I haz the power, 2015-07-09 13:02:54]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2015-07-09 13:02:56]

Try jobs failed on following builders:
  linux_chromium_gn_chromeos_rel on tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[sdefresne, 2015-07-09 13:56:00]

The CQ bit was checked by sdefresne@chromium.org

[sdefresne, 2015-07-09 13:56:00]

The patchset sent to the CQ was uploaded after l-g-t-m from jochen@chromium.org,
xiyuan@chromium.org, cpu@chromium.org, rogerta@chromium.org,
phajdan.jr@chromium.org, pkasting@chromium.org
Link to the patchset: https://codereview.chromium.org/1212163011/#ps180001
(title: "Rebase, round 3 (fixing gn build)")

[commit-bot: I haz the power, 2015-07-09 13:57:20]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1212163011/180001

[commit-bot: I haz the power, 2015-07-09 14:56:30]

Committed patchset #10 (id:180001)

[commit-bot: I haz the power, 2015-07-09 14:57:13]

Patchset 10 (id:??) landed as
https://crrev.com/311a46585ee5b77486ef42d947bc7cc7566558b1
Cr-Commit-Position: refs/heads/master@{#338040}

[Nico, 2015-07-09 17:10:20]

thakis@chromium.org changed reviewers:
+ thakis@chromium.org

[Nico, 2015-07-09 17:10:22]

This broke official builds like so:

..\..\chrome\browser\chrome_browser_main.cc(1461,33) :  error(clang): no member
named 'IsGoogleHomePage' in 'ChromeRLZTrackerDelegate'; did you mean
'IsGoogleHomepage'?
      ChromeRLZTrackerDelegate::IsGoogleHomePage(profile_),
      ~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
                                IsGoogleHomepage
..\..\chrome/browser/rlz/chrome_rlz_tracker_delegate.h(25,15) :  note(clang):
'IsGoogleHomepage' declared here
  static bool IsGoogleHomepage(Profile* profile);
              ^

I'll fix, but please make it so that we don't have IsGoogleHomepage and
google_util::IsGoogleHomePageUrl with different homepage capitalization.

[Nico, 2015-07-09 21:00:24]

It also broke tests in official builds. I'll revert. (The revert button doesn't
work for this patch, so I'll do it manually)

[sdefresne, 2015-07-10 10:08:00]

rogerta: there were failure when running unit tests with RLZ enable after my
changes. It turned out that unit tests do call to RLZTracker but never install
an RLZTrackerDelegate. I've modified all public method to do early return if
there is no delegate installed which makes the tests pass for me.

Please take another look (you can just look at patchset #11 that fix the test
issue).

[Roger Tawa OOO till Jul 10th, 2015-07-14 12:32:06]

patchset #11 lgtm

[sdefresne, 2015-07-16 07:58:09]

The CQ bit was checked by sdefresne@chromium.org

[sdefresne, 2015-07-16 07:58:09]

The patchset sent to the CQ was uploaded after l-g-t-m from jochen@chromium.org,
pkasting@chromium.org, xiyuan@chromium.org, cpu@chromium.org,
phajdan.jr@chromium.org, rogerta@chromium.org
Link to the patchset: https://codereview.chromium.org/1212163011/#ps220001
(title: "Rebase")

[commit-bot: I haz the power, 2015-07-16 07:58:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1212163011/220001

[commit-bot: I haz the power, 2015-07-16 08:34:47]

Committed patchset #12 (id:220001)

[commit-bot: I haz the power, 2015-07-16 08:35:47]

Patchset 12 (id:??) landed as
https://crrev.com/d967d55cf7a0f78e6ace9ec681d2d082e631810b
Cr-Commit-Position: refs/heads/master@{#339002}