[Telemetry] Googler Mobile First benchmark

[Telemetry] Googler Mobile First benchmark

A benchmark to simulate the behaviour of a user interacting with
Gmail and Chrome for code search/review/bugs on an Android tablet.

BUG=567696, 557641
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_10_10_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect

[perezju, 2015-12-08 11:37:51]

Description was changed from

==========
[Telemetry] Googler Mobile First benchmark

A benchmark to simulate the behaviour of a user interacting with
Gmail and Chrome for code search/review/bugs on an Android tabled.

BUG=557641
==========

to

==========
[Telemetry] Googler Mobile First benchmark

A benchmark to simulate the behaviour of a user interacting with
Gmail and Chrome for code search/review/bugs on an Android tabled.

BUG=557641
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_10_10_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
==========

[perezju, 2015-12-08 11:47:08]

perezju@chromium.org changed reviewers:
+ jbudorick@chromium.org, nednguyen@google.com, picksi@google.com

[perezju, 2015-12-08 11:47:09]

Ned: this is pretty much ready, but still depends on
https://codereview.chromium.org/1491183003/ for WPR to work.

John: just a tiny addition on devil.android.sdk.keyevent

https://codereview.chromium.org/1510833002/diff/1/tools/perf/benchmarks/memor...
File tools/perf/benchmarks/memory_infra.py (right):

https://codereview.chromium.org/1510833002/diff/1/tools/perf/benchmarks/memor...
tools/perf/benchmarks/memory_infra.py:96: return 'nexus' not in
possible_browser.platform.GetDeviceTypeName().lower()
I'm thinking instead maybe we could also land this disabled for all platforms
and pass --also-run-disabled on a bot to try it out for a while before running
it more widely on other android bots.

https://codereview.chromium.org/1510833002/diff/1/tools/telemetry/telemetry/i...
File tools/telemetry/telemetry/internal/backends/android_app_backend.py (right):

https://codereview.chromium.org/1510833002/diff/1/tools/telemetry/telemetry/i...
tools/telemetry/telemetry/internal/backends/android_app_backend.py:94:
'^%s(:|$)' % re.escape(self._default_process_name), n)
Otherwise it would also catch "gms" processes when trying to find "gm" (Gmail)
processes.

[perezju, 2015-12-08 11:50:06]

Description was changed from

==========
[Telemetry] Googler Mobile First benchmark

A benchmark to simulate the behaviour of a user interacting with
Gmail and Chrome for code search/review/bugs on an Android tabled.

BUG=557641
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_10_10_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
==========

to

==========
[Telemetry] Googler Mobile First benchmark

A benchmark to simulate the behaviour of a user interacting with
Gmail and Chrome for code search/review/bugs on an Android tabled.

BUG=567696,557641
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_10_10_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
==========

[jbudorick, 2015-12-08 15:22:35]

devil/ lgtm

[nednguyen, 2015-12-08 19:08:41]

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

[nednguyen, 2015-12-08 19:08:42]

Kari: can you take a pass at this?

https://codereview.chromium.org/1510833002/diff/1/tools/perf/page_sets/mobile...
File tools/perf/page_sets/mobile_first_story.py (right):

https://codereview.chromium.org/1510833002/diff/1/tools/perf/page_sets/mobile...
tools/perf/page_sets/mobile_first_story.py:15: # pylint: disable=import-error
Can you update PRESUBMIT of tools/perf to include devil/?

[perezju, 2015-12-09 12:30:18]

https://codereview.chromium.org/1510833002/diff/1/tools/perf/page_sets/mobile...
File tools/perf/page_sets/mobile_first_story.py (right):

https://codereview.chromium.org/1510833002/diff/1/tools/perf/page_sets/mobile...
tools/perf/page_sets/mobile_first_story.py:15: # pylint: disable=import-error
On 2015/12/08 19:08:42, nednguyen wrote:
> Can you update PRESUBMIT of tools/perf to include devil/?

Done.

[perezju, 2015-12-10 13:56:12]

Kari, hope you can have a look.

[aiolos (Not reviewing), 2015-12-10 20:37:05]

https://codereview.chromium.org/1510833002/diff/20001/tools/perf/page_sets/mo...
File tools/perf/page_sets/mobile_first_story.py (right):

https://codereview.chromium.org/1510833002/diff/20001/tools/perf/page_sets/mo...
tools/perf/page_sets/mobile_first_story.py:3: # found in the LICENSE file.
Nit: rename this file to mobile_first.py instead of mobile_first_story.py

https://codereview.chromium.org/1510833002/diff/20001/tools/perf/page_sets/mo...
tools/perf/page_sets/mobile_first_story.py:10: from telemetry.internal.platform
import android_device
You shouldn't use anything in telemetry/internal. It's not part of the telemetry
public API. The Telemetry team may change it at anytime, and possibly break your
test, with no warning.

I'm going to talk offline with Ned about the best fix for this to avoid a bunch
of codereview churn.

[perezju, 2015-12-11 09:30:18]

Description was changed from

==========
[Telemetry] Googler Mobile First benchmark

A benchmark to simulate the behaviour of a user interacting with
Gmail and Chrome for code search/review/bugs on an Android tabled.

BUG=567696,557641
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_10_10_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
==========

to

==========
[Telemetry] Googler Mobile First benchmark

A benchmark to simulate the behaviour of a user interacting with
Gmail and Chrome for code search/review/bugs on an Android tablet.

BUG=567696,557641
CQ_EXTRA_TRYBOTS=tryserver.chromium.perf:linux_perf_bisect;tryserver.chromium.perf:mac_10_10_perf_bisect;tryserver.chromium.perf:win_perf_bisect;tryserver.chromium.perf:android_nexus5_perf_bisect
==========

[perezju, 2015-12-11 09:41:38]

https://codereview.chromium.org/1510833002/diff/20001/tools/perf/page_sets/mo...
File tools/perf/page_sets/mobile_first_story.py (right):

https://codereview.chromium.org/1510833002/diff/20001/tools/perf/page_sets/mo...
tools/perf/page_sets/mobile_first_story.py:3: # found in the LICENSE file.
On 2015/12/10 20:37:05, aiolos wrote:
> Nit: rename this file to mobile_first.py instead of mobile_first_story.py 

Acknowledged.

https://codereview.chromium.org/1510833002/diff/20001/tools/perf/page_sets/mo...
tools/perf/page_sets/mobile_first_story.py:10: from telemetry.internal.platform
import android_device
On 2015/12/10 20:37:05, aiolos wrote:
> You shouldn't use anything in telemetry/internal. It's not part of the
telemetry
> public API. The Telemetry team may change it at anytime, and possibly break
your
> test, with no warning.
> 
> I'm going to talk offline with Ned about the best fix for this to avoid a
bunch
> of codereview churn.

I see. That makes sense, and I understand the rationale behind that.

I basically need these modules to (1) get hold of an android_platform instance,
and (2) get hold of a browser instance.

Just throwing some ideas now, not sure if it makes sense, but maybe something
like my methods GetAndroidPlatform and CreateBrowser below could be moved to the
SharedState base class? This could make it easier for developers to write their
own shared state subclasses (like I need to do now).

[aiolos (Not reviewing), 2016-01-20 23:39:40]

We will be adding functionality to the API, but have been distracted by other
work.