Web Automation: add command to asynchronously load the Web Inspector frontend in the background

[BJ Burg]

In case you want to use `debugger;` or hit a breakpoint whilst running other actions.

[Radar WebKit Bug Importer]

<rdar://problem/26190121>

[BJ Burg]

Created attachment 278481 [details]
Proposed Fix

[Timothy Hatcher]

Comment on attachment 278481 [details]
Proposed Fix

View in context: https://bugs.webkit.org/attachment.cgi?id=278481&action=review

> Source/WebKit2/ChangeLog:36
> +        * WebProcess/Automation/WebAutomationSessionProxy.js:
> +        (let.AutomationSessionProxy.prototype.evaluateJavaScriptFunction):
> +        Rename clearTimeout to window.clearTimeout to be less ambiguous.

I'm not sure why this is needed. We hardly ever prefix the timeout functions.

> Source/WebKit2/UIProcess/Automation/Automation.json:303
> +            },

Stray?

[Joseph Pecoraro]

Comment on attachment 278481 [details]
Proposed Fix

View in context: https://bugs.webkit.org/attachment.cgi?id=278481&action=review

> Source/WebKit2/UIProcess/Automation/Automation.json:298
> +            "description": "Inspect the specified browsing context using the Web Inspector.",

We normally drop 'this' in the phrase "the Web Inspector". You wouldn't say "the Xcode" or "the Safari".

> Source/WebKit2/UIProcess/Automation/WebAutomationSession.cpp:448
> +    page->inspector()->connect();

Most code that uses page->inspector() includes a nullptr check. We should do that here as well. If the WebPage is closed / invalid then inspector() would be nullptr. Unless there is some other way that this would be avoided.

[Joseph Pecoraro]

> We normally drop 'this'

Err, 'the'

[BJ Burg]

Committed r200702: <http://trac.webkit.org/changeset/200702>