Added a timezone policy and pyauto tests for it.

chrome/browser/automation/testing_automation_provider.h

 // Copyright (c) 2012 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 CHROME_BROWSER_AUTOMATION_TESTING_AUTOMATION_PROVIDER_H_
 #define CHROME_BROWSER_AUTOMATION_TESTING_AUTOMATION_PROVIDER_H_
 
 #include <map>
 #include <string>
 #include <vector>
@@ skipping 1583 matching lines @@
 
   void SetMute(base::DictionaryValue* args, IPC::Message* reply_message);
 
   void CaptureProfilePhoto(Browser* browser,
                            DictionaryValue* args,
                            IPC::Message* reply_message);
 
   // Html terminal.
   void OpenCrosh(base::DictionaryValue* args, IPC::Message* reply_message);
 
+  // Set ChromeOS Settings.
+  // Uses the JSON interface for input/output.
+  // Sample json input:
+  // { "command": "SetCrosSetting",
+  //   "path": path,
+  //   "value": value }
+  void SetCrosSetting(base::DictionaryValue* args, IPC::Message* reply_message);

[Joao da Silva, 2012/07/27 13:40:57]

This is a big hammer. We previously had a similar call to SetPolicies() and
ultimately decided to remove it.

Notice that there's already a SetTimezone() automation call, isn't it enough to
cover your test?

[pneubeck (no reviews), 2012/07/31 22:21:05]

The idea of this definition was to change chromeos settings (controlled by
CrosSettings) in the same fashion as normal settings are set using the function
SetPrefs. I see, that this means that policies can also be changed via the owner
of the device because these are also controlled by CrosSettings.
I don't see in which way this is problematic. Surely it is not really necessary
as long as the timezone test is the only use case. But the difference of the
generic function i provided and the a special function for timezone isn't that
huge... so why should I provide the special one?

The major difference between setting the timezone with this function and the
existing SetTimezone is, that this one talks to CrosSettings and the other to
the underlying chromeos::system::TimezoneSettings.

Depending on what we like to have, more a functional, integration or unit test,
both may be more appropriate. For functional tests, which I wanted to implement
and many pyauto test seem to be, the new function should be proper.

Do you have any other suggestions or which of the variants do you think fits
better?

[pneubeck (no reviews), 2012/08/06 14:50:48]

Done.

+
   void AddChromeosObservers();
   void RemoveChromeosObservers();
 
 #endif  // defined(OS_CHROMEOS)
 
   void WaitForTabCountToBecome(int browser_handle,
                                int target_tab_count,
                                IPC::Message* reply_message);
 
   void WaitForInfoBarCount(int tab_handle,
@@ skipping 40 matching lines @@
   scoped_ptr<AutomationEventQueue> automation_event_queue_;
 
   // List of commands which just finish synchronously and don't require
   // setting up an observer.
   static const int kSynchronousCommands[];
 
   DISALLOW_COPY_AND_ASSIGN(TestingAutomationProvider);
 };
 
 #endif  // CHROME_BROWSER_AUTOMATION_TESTING_AUTOMATION_PROVIDER_H_