Telemetry: Create new profile creator large_profile_creator.

Telemetry: Create new profile creator large_profile_creator.

This CL creates two subclasses of FastNavigationProfileExtender:
CookieProfileExtender and HistoryProfilerExtender. The former performs up to
500 navigations, with the goal of filling up the Cookie Database but not
overfilling it (which is possible, unfortunately). The latter performs a large
number ~20,000 navigations to URIs pointing at the local file system to fill up
the History Database.

The run these profile extenders, this CL adds the class LargeProfileCreator.
This new class intentionally contains minimal logic, since the existing
profile_creator.py file is pretty hacky, and the goal is to eventually move
away from it entirely.

BUG=442546
Committed: https://crrev.com/40e8ee4f57bdb5d0c6fcc27f8a551a9ea2420029
Cr-Commit-Position: refs/heads/master@{#317112}

[erikchen, 2015-02-12 02:02:00]

erikchen@chromium.org changed reviewers:
+ nednguyen@google.com

[erikchen, 2015-02-12 02:02:01]

nednguyen: Please review.

This CL builds on top of https://codereview.chromium.org/907503002/. If/when we
land that CL, I will rebase against that.

In additional to the new classes mentioned in the CL description, this CL
includes several changes to FastNavigationProfileExtender which I thought I
would point out:

The class now dynamically generates each batch of URLs. This allows
HistoryProfileExtender to create/destroy a small number of temporary files at a
time, rather than generating 20,000+ plus all at once.

The class has a new hook for subclasses to perform clean-up after each batch of
navigations.

There are some smaller changes, but none of them significantly alter the
functionality. (e.g. The class now chooses the number of tabs to use based on
the number of logical cores in the CPU).

[erikchen, 2015-02-13 04:02:12]

nednguyen: Ping?

[nednguyen, 2015-02-13 04:04:05]

On 2015/02/13 04:02:12, erikchen wrote:
> nednguyen: Ping?

I want to wait for the other patches to land & this gets rebased before
reviewing.

[erikchen, 2015-02-13 04:05:37]

On 2015/02/13 04:04:05, nednguyen wrote:
> On 2015/02/13 04:02:12, erikchen wrote:
> > nednguyen: Ping?
> 
> I want to wait for the other patches to land & this gets rebased before
> reviewing.

Okay.

[erikchen, 2015-02-18 00:18:15]

Patchset #4 (id:60001) has been deleted

[erikchen, 2015-02-18 00:18:20]

Patchset #4 (id:80001) has been deleted

[erikchen, 2015-02-18 00:24:37]

nednguyen: PTAL

If you diff patch set 4 against patch set 3, you will get an exact diff against
the CL that is in the commit queue.

[erikchen, 2015-02-18 02:26:49]

nednguyen: PTAL

The dependency CL has landed, and patch set 5 has rebased against top of tree.

[nednguyen, 2015-02-18 02:31:26]

nednguyen@google.com changed reviewers:
+ sullivan@chromium.org

[sullivan, 2015-02-18 21:18:06]

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
File tools/perf/profile_creators/fast_navigation_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:28: # The path
of the profile that the browser will use while its running.
nit: it's

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:98:
self._browser = possible_browser.Create(finder_options)
I think there are some browser types this won't work on: android devices, cros
devices, trybots, single tab browsers. Should assert that this is a browser the
code will work on here.

[erikchen, 2015-02-18 22:04:01]

sullivan: PTAL

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
File tools/perf/profile_creators/fast_navigation_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:28: # The path
of the profile that the browser will use while its running.
On 2015/02/18 21:18:05, sullivan wrote:
> nit: it's

Done.

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:98:
self._browser = possible_browser.Create(finder_options)
On 2015/02/18 21:18:05, sullivan wrote:
> I think there are some browser types this won't work on: android devices, cros
> devices, trybots, single tab browsers. Should assert that this is a browser
the
> code will work on here.

I added asserts for "supports_tab_control" and os is mac, windows or linux. I
didn't see a reason to prevent this from running on try-bots.

[nednguyen, 2015-02-18 22:43:14]

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
File tools/perf/profile_creators/cookie_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/cookie_profile_extender.py:41: def
_IsCookieDBFull(self):
Can you add unittest coverage for this?

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
File tools/perf/profile_creators/fast_navigation_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:39:
self._NUM_TABS = None
Why not make this a parameter in the constructor?

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:86: return
multiprocessing.cpu_count()
Why do you think hat number of tabs should equal number of cpus?

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:114: def
FinishedNavigationOfBatch(self):
CleanUpAfterBatchNavigation

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:236: if
self.ShouldExit():
Just looking at this method name, it's unclear to me when it's called and what
is the state of the world when it's called.

maybe ShouldExitAfterBatchNavigation()?

[erikchen, 2015-02-18 23:32:30]

nednguyen: PTAL

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
File tools/perf/profile_creators/cookie_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/cookie_profile_extender.py:41: def
_IsCookieDBFull(self):
On 2015/02/18 22:43:14, nednguyen wrote:
> Can you add unittest coverage for this?

I moved the meat of the logic into a new method: _CookieCountInDB and added a
unit test for that.

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
File tools/perf/profile_creators/fast_navigation_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:39:
self._NUM_TABS = None
On 2015/02/18 22:43:14, nednguyen wrote:
> Why not make this a parameter in the constructor?

Done.

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:86: return
multiprocessing.cpu_count()
On 2015/02/18 22:43:14, nednguyen wrote:
> Why do you think hat number of tabs should equal number of cpus?

I moved this logic into the specific subclasses, and added comments. In short: A
lot of these operations are CPU bound, so having 1 tab per logical core is close
to optimal.

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:114: def
FinishedNavigationOfBatch(self):
On 2015/02/18 22:43:14, nednguyen wrote:
> CleanUpAfterBatchNavigation

Done.

https://codereview.chromium.org/914253005/diff/120001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:236: if
self.ShouldExit():
On 2015/02/18 22:43:14, nednguyen wrote:
> Just looking at this method name, it's unclear to me when it's called and what
> is the state of the world when it's called.
> 
> maybe ShouldExitAfterBatchNavigation()?

Done.

[nednguyen, 2015-02-19 00:33:32]

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/cookie_profile_extender_unittest.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/cookie_profile_extender_unittest.py:31: handle, path
= tempfile.mkstemp()
Can you try tempfile.TemporaryFile() here also?

with tempfile.TemporaryFile() as tmp:
 ...

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/fast_navigation_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:35:
self.profile_path = None
If this doesn't need to be accessed from outside, then make it _profile_path.
If this could be accessed from outside, but shouldn't be mutated, please add
@property def profile_path(self):...

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:96: assert
possible_browser.platform.GetOSName() in ["win", "mac", "linux"]
Is the what we want here "possible_browser.platform is
platform.GetHostPlatform()"?

Do you intend for this to be run on ChromeOs?

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:221: urls =
self.GetUrlsToNavigate(self._NUM_TABS)
I don't quite like the NUM_TABS param because it requires the subclass to handle
the "batch logic". 

I think the code here could be written:

urls_iter = iter(self.GetUrlsToNavigate()) # <- or this can return an iterator
directly

while True:
  for _ in xrange(self._NUM_TABS):
     urls.append(self.urls_iter.next())

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/history_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:42: suffix =
"reallylongsuffixintendedtoartificiallyincreasethelengthoftheurl"
Can you add some comment about why we want a long suffix?

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:66: os.remove(path)
You can move TemporaryFile which is deleted on close() instead.

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:76: max_size = ((10**6 *
HistoryProfileExtender._HISTORY_DB_MAX_SIZE_IN_MB) *
Technically, this should be 2 ** 20?

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/large_profile_creator.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/large_profile_creator.py:9: class
LargeProfileCreator(profile_creator.ProfileCreator):
Do we have a plan to get rid of this in subsequent patches?

[erikchen, 2015-02-19 01:56:03]

nednguyen: PTAL

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/cookie_profile_extender_unittest.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/cookie_profile_extender_unittest.py:31: handle, path
= tempfile.mkstemp()
On 2015/02/19 00:33:31, nednguyen wrote:
> Can you try tempfile.TemporaryFile() here also?
> 
> with tempfile.TemporaryFile() as tmp:
>  ...

The behavior of TemporaryFile() is OS dependent, and does not make any
guarantees about the actual state of the file system. This unit test requires
actual file system modifications, since it passes around a path of a sql db.

Quoting the python documentation:
"""
your code should not rely on a temporary file created using this function having
or not having a visible name in the file system.
"""

NamedTemporaryFile() doesn't work either.
 "Whether the name can be used to open the file a second time, while the named
temporary file is still open, varies across platforms (it can be so used on
Unix; it cannot on Windows NT or later)"
So we'd have to close the file (with delete=False) for sqlite3 to be able to
access it on Windows, which is exactly the same as mkstemp.

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/fast_navigation_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:35:
self.profile_path = None
On 2015/02/19 00:33:32, nednguyen wrote:
> If this doesn't need to be accessed from outside, then make it _profile_path.
> If this could be accessed from outside, but shouldn't be mutated, please add
> @property def profile_path(self):...

It needs to be accessed from subclasses. I added a property, and prefixed the
member name with an underscore.

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:96: assert
possible_browser.platform.GetOSName() in ["win", "mac", "linux"]
On 2015/02/19 00:33:32, nednguyen wrote:
> Is the what we want here "possible_browser.platform is
> platform.GetHostPlatform()"?
> 
> Do you intend for this to be run on ChromeOs?

I replaced possible_browser.platform with platform.GetHostPlatform(). 

I'm not familiar with ChromeOS, and I haven't tried running this script there.
The goal is to be able to run on all platforms, eventually.

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/fast_navigation_profile_extender.py:221: urls =
self.GetUrlsToNavigate(self._NUM_TABS)
On 2015/02/19 00:33:32, nednguyen wrote:
> I don't quite like the NUM_TABS param because it requires the subclass to
handle
> the "batch logic". 
> 
> I think the code here could be written:
> 
> urls_iter = iter(self.GetUrlsToNavigate()) # <- or this can return an iterator
> directly
> 
> while True:
>   for _ in xrange(self._NUM_TABS):
>      urls.append(self.urls_iter.next())

I like your suggestion. The result is cleaner. Done.

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/history_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:42: suffix =
"reallylongsuffixintendedtoartificiallyincreasethelengthoftheurl"
On 2015/02/19 00:33:32, nednguyen wrote:
> Can you add some comment about why we want a long suffix?

Done.

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:66: os.remove(path)
On 2015/02/19 00:33:32, nednguyen wrote:
> You can move TemporaryFile which is deleted on close() instead.

For a full explanation, see my earlier response. In short:

TemporaryFile() does not guarantee the creation of a file-system visible file. 
NamedTemporaryFile() is no different from mkstemp(), since we still need to
close the file handle on Windows, and set the flag Delete=False.

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:76: max_size = ((10**6 *
HistoryProfileExtender._HISTORY_DB_MAX_SIZE_IN_MB) *
On 2015/02/19 00:33:32, nednguyen wrote:
> Technically, this should be 2 ** 20?

Done.

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/large_profile_creator.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/large_profile_creator.py:9: class
LargeProfileCreator(profile_creator.ProfileCreator):
On 2015/02/19 00:33:32, nednguyen wrote:
> Do we have a plan to get rid of this in subsequent patches? 

Yup.

Most of the logic in profile_generator.py is reasonable. Its biggest problem is
that its location makes no sense. It should be in this directory. Also, there's
not much point in having a separate class ProfileCreator. Everything should be a
profile_extender, although that class also doesn't exist yet.

[nednguyen, 2015-02-19 17:07:37]

lgtm with nits. Please wait for Annie's approval also.

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/cookie_profile_extender_unittest.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/cookie_profile_extender_unittest.py:31: handle, path
= tempfile.mkstemp()
On 2015/02/19 01:56:03, erikchen wrote:
> On 2015/02/19 00:33:31, nednguyen wrote:
> > Can you try tempfile.TemporaryFile() here also?
> > 
> > with tempfile.TemporaryFile() as tmp:
> >  ...
> 
> The behavior of TemporaryFile() is OS dependent, and does not make any
> guarantees about the actual state of the file system. This unit test requires
> actual file system modifications, since it passes around a path of a sql db.
> 
> Quoting the python documentation:
> """
> your code should not rely on a temporary file created using this function
having
> or not having a visible name in the file system.
> """
> 
> NamedTemporaryFile() doesn't work either.
>  "Whether the name can be used to open the file a second time, while the named
> temporary file is still open, varies across platforms (it can be so used on
> Unix; it cannot on Windows NT or later)"
> So we'd have to close the file (with delete=False) for sqlite3 to be able to
> access it on Windows, which is exactly the same as mkstemp.

Thanks for clear explanation. :-)

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/large_profile_creator.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/large_profile_creator.py:9: class
LargeProfileCreator(profile_creator.ProfileCreator):
On 2015/02/19 01:56:03, erikchen wrote:
> On 2015/02/19 00:33:32, nednguyen wrote:
> > Do we have a plan to get rid of this in subsequent patches? 
> 
> Yup.
> 
> Most of the logic in profile_generator.py is reasonable. Its biggest problem
is
> that its location makes no sense. It should be in this directory. Also,
there's
> not much point in having a separate class ProfileCreator. Everything should be
a
> profile_extender, although that class also doesn't exist yet.

Yeah, I think you have sound vision here.

https://codereview.chromium.org/914253005/diff/200001/tools/perf/profile_crea...
File tools/perf/profile_creators/history_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/200001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:41: self._file_paths =
[]
Can you change this to self._generated_temp_files?

https://codereview.chromium.org/914253005/diff/200001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:65: return
_HistoryUrlIterator(self)
How about using yield?

while True:
   yield self.MakeTemporaryFile()

[erikchen, 2015-02-19 18:24:01]

New patchsets have been uploaded after l-g-t-m from nednguyen@google.com

[sullivan, 2015-02-19 18:24:42]

lgtm

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
File tools/perf/profile_creators/cookie_profile_extender_unittest.py (right):

https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
tools/perf/profile_creators/cookie_profile_extender_unittest.py:31: handle, path
= tempfile.mkstemp()
On 2015/02/19 17:07:37, nednguyen wrote:
> On 2015/02/19 01:56:03, erikchen wrote:
> > On 2015/02/19 00:33:31, nednguyen wrote:
> > > Can you try tempfile.TemporaryFile() here also?
> > > 
> > > with tempfile.TemporaryFile() as tmp:
> > >  ...
> > 
> > The behavior of TemporaryFile() is OS dependent, and does not make any
> > guarantees about the actual state of the file system. This unit test
requires
> > actual file system modifications, since it passes around a path of a sql db.
> > 
> > Quoting the python documentation:
> > """
> > your code should not rely on a temporary file created using this function
> having
> > or not having a visible name in the file system.
> > """
> > 
> > NamedTemporaryFile() doesn't work either.
> >  "Whether the name can be used to open the file a second time, while the
named
> > temporary file is still open, varies across platforms (it can be so used on
> > Unix; it cannot on Windows NT or later)"
> > So we'd have to close the file (with delete=False) for sqlite3 to be able to
> > access it on Windows, which is exactly the same as mkstemp.
> 
> Thanks for clear explanation. :-)

We've had test breakages before due to platform-specific usage of tempfile APIs.
Maybe it's worth adding a comment why we are using mkstemp?

[erikchen, 2015-02-19 18:24:49]

sullivan: PTAL

https://codereview.chromium.org/914253005/diff/200001/tools/perf/profile_crea...
File tools/perf/profile_creators/history_profile_extender.py (right):

https://codereview.chromium.org/914253005/diff/200001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:41: self._file_paths =
[]
On 2015/02/19 17:07:37, nednguyen wrote:
> Can you change this to self._generated_temp_files? 

Done.

https://codereview.chromium.org/914253005/diff/200001/tools/perf/profile_crea...
tools/perf/profile_creators/history_profile_extender.py:65: return
_HistoryUrlIterator(self)
On 2015/02/19 17:07:37, nednguyen wrote:
> How about using yield?
> 
> while True:
>    yield self.MakeTemporaryFile()

Ah yes, saves an entire class. Thanks.

[erikchen, 2015-02-19 18:30:40]

New patchsets have been uploaded after l-g-t-m from sullivan@chromium.org

[erikchen, 2015-02-19 18:31:15]

On 2015/02/19 18:24:42, sullivan wrote:
> lgtm
> 
>
https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
> File tools/perf/profile_creators/cookie_profile_extender_unittest.py (right):
> 
>
https://codereview.chromium.org/914253005/diff/160001/tools/perf/profile_crea...
> tools/perf/profile_creators/cookie_profile_extender_unittest.py:31: handle,
path
> = tempfile.mkstemp()
> On 2015/02/19 17:07:37, nednguyen wrote:
> > On 2015/02/19 01:56:03, erikchen wrote:
> > > On 2015/02/19 00:33:31, nednguyen wrote:
> > > > Can you try tempfile.TemporaryFile() here also?
> > > > 
> > > > with tempfile.TemporaryFile() as tmp:
> > > >  ...
> > > 
> > > The behavior of TemporaryFile() is OS dependent, and does not make any
> > > guarantees about the actual state of the file system. This unit test
> requires
> > > actual file system modifications, since it passes around a path of a sql
db.
> > > 
> > > Quoting the python documentation:
> > > """
> > > your code should not rely on a temporary file created using this function
> > having
> > > or not having a visible name in the file system.
> > > """
> > > 
> > > NamedTemporaryFile() doesn't work either.
> > >  "Whether the name can be used to open the file a second time, while the
> named
> > > temporary file is still open, varies across platforms (it can be so used
on
> > > Unix; it cannot on Windows NT or later)"
> > > So we'd have to close the file (with delete=False) for sqlite3 to be able
to
> > > access it on Windows, which is exactly the same as mkstemp.
> > 
> > Thanks for clear explanation. :-)
> 
> We've had test breakages before due to platform-specific usage of tempfile
APIs.
> Maybe it's worth adding a comment why we are using mkstemp?

I added comments to both locations where I use mkstemp.

[erikchen, 2015-02-19 18:31:22]

The CQ bit was checked by erikchen@chromium.org

[commit-bot: I haz the power, 2015-02-19 18:32:34]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/914253005/240001

[commit-bot: I haz the power, 2015-02-19 19:53:13]

Committed patchset #11 (id:240001)

[commit-bot: I haz the power, 2015-02-19 19:53:39]

Patchset 11 (id:??) landed as
https://crrev.com/40e8ee4f57bdb5d0c6fcc27f8a551a9ea2420029
Cr-Commit-Position: refs/heads/master@{#317112}