Bug 176436 – Add Live PLT implemented with WebDriver

• Component: New Bugs
• Status: RESOLVED FIXED

[Matthew Stewart]

Add Live PLT implemented with WebDriver

[Matthew Stewart]

Created attachment 319968 [details]
Patch

[Build Bot]

Attachment 319968 [details] did not pass style-queue:


ERROR: Tools/Scripts/run-webdriver-plt.py:2:  No name 'webdriver_plt' in module 'webkitpy'  [pylint/E0611] [5]
ERROR: Tools/Scripts/run-webdriver-plt.py:3:  No name 'webdriver_plt' in module 'webkitpy'  [pylint/E0611] [5]
Total errors found: 2 in 12 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Stephanie Lewis]

Comment on attachment 319968 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=319968&action=review

> Tools/Scripts/run-webdriver-plt.py:13
> +    parser.add_argument('-n', '--instances', dest='instances', type=int, default=3, help='Test the suite n times in the same browser session')

This backwards from the other plt.  Instances is number of times safari is launched

> Tools/Scripts/webkitpy/webdriver_plt/liveplt.py:23
> +        self.size = size

I'd create a log message dumping the parameters here.  It's useful to go back and look at the defaults when trying to compare two logs.

> Tools/Scripts/webkitpy/webdriver_plt/liveplt.py:90
> +        sleep(1)

Ideally you'd use web driver to figure out when the window is maximized.  Blanket sleep statements often behave badly when things aren't going as planned.

> Tools/Scripts/webkitpy/webdriver_plt/liveplt.py:116
> +                    pass  # opening new tabs not currently possible

please print a help message here

> Tools/Scripts/webkitpy/webdriver_plt/liveplt.py:133
> +        NT_results = driver.execute_script('return performance.getEntriesByType("navigation")[0]')  # navigation timing 2

nit probably should have used lower case nt.

> Tools/Scripts/webkitpy/webdriver_plt/suites/social.suite:13
> +    ]

pleased add a media suite.  ie. youtube flickr

Comment on attachment 319968 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=319968&action=review

> Tools/ChangeLog:36
> +        * Scripts/webkitpy/webdriver_plt/suites/__init__.py: Added.

This file should not be empty, otherwise, it will be ignored by svn.

> Tools/Scripts/webkitpy/webdriver_plt/liveplt.py:12
> +    'safari': 76,
> +    'chrome': 0,
> +    'firefox': 37,
> +    'stp': 76,
> +}

Does those number ever change?

> Tools/Scripts/webkitpy/webdriver_plt/pltresults.py:28
> +        self.product *= time

I would prefer using log function to calculate geomean.
geomean = (x * y)^(1/2)
-> log(geomean) = log((x * y)^(1/2))
-> log(geomean) = (1/2) * (log(x) + log(y))
So we can accumulate log(time) here, and get geomean by apply pow(e, accumulated_value / n)

[Matthew Stewart]

Created attachment 320164 [details]
Patch

[Matthew Stewart]

Created attachment 320224 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 320224 [details]
Patch

Clearing flags on attachment: 320224

Committed r221777: <http://trac.webkit.org/changeset/221777>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/34693639>