Add Multi-tab System Health Story

tools/perf/page_sets/system_health/multi_tab_stories.py

Index: tools/perf/page_sets/system_health/multi_tab_stories.py
diff --git a/tools/perf/page_sets/system_health/multi_tab_stories.py b/tools/perf/page_sets/system_health/multi_tab_stories.py
new file mode 100644
index 0000000000000000000000000000000000000000..0ec2472f0aaa4a80c3d2ab3994739bb98796291d
--- /dev/null
+++ b/tools/perf/page_sets/system_health/multi_tab_stories.py
@@ -0,0 +1,78 @@
+# Copyright 2017 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.
+
+import py_utils
+import logging
+
+from page_sets.system_health import system_health_story
+from page_sets.system_health import platforms
+
+
+class _MultiTabStory(system_health_story.SystemHealthStory):
+  ABSTRACT_STORY = True
+
+  def RunNavigateSteps(self, action_runner):
+    super(_MultiTabStory, self).RunNavigateSteps(action_runner)
+
+    tabs = action_runner.tab.browser.tabs

[perezju, 2017/03/02 14:33:12]

High level question, do we really want to

1) open all tabs at the same time
2) wait for each page to be ready

Or would it be better:

A) for each tab:
   B) open the tab, wait for its page to be ready

??

[vovoy, 2017/03/02 18:02:29]

I open all tabs and then wait for each tabs because 
I think it's fast to load each tab's content simultaneously.
while it wait for tab 1 to be ready, others tabs are also loading.

I can do some experiment to compare the total time for the 1)2) or A)B)

[vovoy, 2017/03/03 08:30:11]

test command:
tools/perf/run_benchmark system_health.memory_desktop --browser=reference
--device=desktop --story-filter=multitab:typical24:typical24
--also-run-disabled-tests

1)2) takes 52 +- 1 (secs)
A)B) takes 119 +- 2 (secs)

I will keep parallel loading 1)2) unless there are other reasons for serial
loading A)B)

[perezju, 2017/03/03 10:32:42]

Sounds good. Thanks for checking!

+
+    for i in range(1, len(self.URL_LIST)):

[perezju, 2017/03/02 14:33:11]

for url in self.URL_LIST[1:]:
  ...


Also add a small comment explaining why the first url is skipped.

[vovoy, 2017/03/02 18:02:29]

OK, will add comment

+      new_tab = tabs.New()
+      new_tab.action_runner.Navigate(self.URL_LIST[i])
+
+    for i in range(len(self.URL_LIST)):

[perezju, 2017/03/02 14:33:11]

A possibility here could be:

for i, url in enumerate(self.URL_LIST):
  ...

[vovoy, 2017/03/03 08:30:11]

Done.

+      try:
+        py_utils.WaitFor(tabs[i].HasReachedQuiescence, 15)

[perezju, 2017/03/02 14:33:12]

tabs[i].action_runner.WaitForNetworkQuiescence(15)

[vovoy, 2017/03/03 08:30:11]

Done.

+      except py_utils.TimeoutException:
+        logging.warning('HasReachedQuiescence timeout, url[%d]: %s'
+                        % (i, self.URL_LIST[i]))
+
+  def RunPageInteractions(self, action_runner):
+    tabs = action_runner.tab.browser.tabs
+    for i in range(len(tabs)):

[perezju, 2017/03/02 14:33:12]

for tab in action_runner.tab.browser.tabs:
  ...

[vovoy, 2017/03/03 08:30:11]

Done.

+      tabs[i].Activate()
+      tabs[i].WaitForFrameToBeDisplayed()
+
+
+class MultiTabTypical24Story(_MultiTabStory):
+  NAME = 'multitab:typical24:typical24'

[nednguyen, 2017/03/02 13:26:00]

can you add:
TAGS = [story_tags.TABS_SWITCHING]?

[perezju, 2017/03/02 14:33:12]

nit: I think 'multitab:misc:typical24' might be a better name. /bike-shedding

[vovoy, 2017/03/02 18:02:29]

OK

[vovoy, 2017/03/03 08:30:11]

Done.

+  URL_LIST = [

[perezju, 2017/03/02 14:33:11]

Is this list coming from somewhere else? Could we import it from there rather
than duplicate?

[vovoy, 2017/03/02 18:02:29]

the list is coming from page_sets/typical_25.py
maybe I could modify typical_25.py to export the list

[nednguyen, 2017/03/02 19:08:19]

I think there is no need to share code here. As long as the list of sites here
seems reasonable, we're good to go.

+    # Why: Alexa games #48
+    'http://www.nick.com/games',
+    # Why: Alexa sports #45
+    'http://www.rei.com/',
+    # Why: Alexa sports #50
+    'http://www.fifa.com/',
+    # Why: Alexa shopping #41
+    'http://www.gamestop.com/ps3',
+    # Why: Alexa news #55
+    ('http://www.economist.com/news/science-and-technology/21573529-small-'
+     'models-cosmic-phenomena-are-shedding-light-real-thing-how-build'),
+    # Why: Alexa news #67
+    'http://www.theonion.com',
+    'http://arstechnica.com/',
+    # Why: Alexa home #10
+    'http://allrecipes.com/Recipe/Pull-Apart-Hot-Cross-Buns/Detail.aspx',
+    'http://www.html5rocks.com/en/',
+    'http://www.mlb.com/',
+    'http://gawker.com/5939683/based-on-a-true-story-is-a-rotten-lie-i-hope-you-never-believe',
+    'http://www.imdb.com/title/tt0910970/',
+    'http://www.flickr.com/search/?q=monkeys&f=hp',
+    'http://money.cnn.com/',
+    'http://www.nationalgeographic.com/',
+    'http://premierleague.com',
+    'http://www.osubeavers.com/',
+    'http://walgreens.com',
+    'http://colorado.edu',
+    ('http://www.ticketmaster.com/JAY-Z-and-Justin-Timberlake-tickets/artist/'
+     '1837448?brand=none&tm_link=tm_homeA_rc_name2'),
+    # pylint: disable=line-too-long
+    'http://www.theverge.com/2013/3/5/4061684/inside-ted-the-smartest-bubble-in-the-world',
+    'http://www.airbnb.com/',
+    'http://www.ign.com/',
+    # Why: Alexa health #25
+    'http://www.fda.gov',
+  ]
+  URL = URL_LIST[0]
+  SUPPORTED_PLATFORMS = platforms.DESKTOP_ONLY