Bug 210010 – RequestTextInputContext.Simple iframe sub-tests may sometimes fail

• Status: RESOLVED
• Resolution: FIXED
• Component: Tools / Tests

[Daniel Bates]

The iframe sub-tests in RequestTextInputContext.Simple are non-deterministic. They may fail if the iframe has not finished loading before the web process hit tests to find the editable elements. To make the these tests more robust they need to wait until the iframe has loaded.

[Radar WebKit Bug Importer]

<rdar://problem/61303697>

[Daniel Bates]

Created attachment 395463 [details]
Patch

[Daniel Bates]

Created attachment 395464 [details]
Patch

[Darin Adler]

Comment on attachment 395464 [details]
Patch

Which test needs the inMainFrameOnly:NO behavior?

[Daniel Bates]

(In reply to Darin Adler from comment #4)
> Comment on attachment 395464 [details]
> Patch
> 
> Which test needs the inMainFrameOnly:NO behavior?

All the existing ones that call -performAfterLoading. I want to invert this relationship (see ChangeLog), but I hope you don't mind I defer for now.

[Daniel Bates]

Comment on attachment 395464 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=395464&action=review

> Tools/TestWebKitAPI/Tests/WebKitCocoa/RequestTextInputContext.mm:154
> +static webViewLoadHTMLStringAndWaitForDOMLoadEvent(TestWKWebView *webView, NSString *htmlString)

missing void

[Daniel Bates]

Created attachment 395465 [details]
Patch

[Darin Adler]

Comment on attachment 395464 [details]
Patch

I predict we will find no callers that want or need inMainFrameOnly:NO.

[Darin Adler]

Comment on attachment 395465 [details]
Patch

I predict we will find no callers that want or need inMainFrameOnly:NO.

I predict that all the existing callers would work the same or better with inMainFrameOnly:YES.

[Daniel Bates]

(In reply to Darin Adler from comment #8)
> Comment on attachment 395464 [details]
> Patch
> 
> I predict we will find no callers that want or need inMainFrameOnly:NO.

If I was betting, I bet you're right. I just didn't want to rock the boat...

[Daniel Bates]

(In reply to Darin Adler from comment #9)
> Comment on attachment 395465 [details]
> Patch
> 
> I predict we will find no callers that want or need inMainFrameOnly:NO.
> 
> I predict that all the existing callers would work the same or better with
> inMainFrameOnly:YES.

Okay, I'll throw a patch up....

[Daniel Bates]

Thanks for the review!

[Daniel Bates]

Created attachment 395466 [details]
[Alt] Patch that changes -performAfterLoading to add user script to main frame only

Let's see what the bots say...

[Daniel Bates]

Created attachment 395467 [details]
[Alt] Patch that changes -performAfterLoading to add user script to main frame only

[Daniel Bates]

Comment on attachment 395465 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=395465&action=review

> Tools/TestWebKitAPI/Tests/WebKitCocoa/RequestTextInputContext.mm:170
> +    webViewLoadHTMLStringAndWaitForDOMLoadEvent(webView, applyIframe(@"<input type='text' style='width: 50px; height: 50px;'>"));

Needs a .get() here.

> Tools/TestWebKitAPI/Tests/WebKitCocoa/RequestTextInputContext.mm:175
> +    webViewLoadHTMLStringAndWaitForDOMLoadEvent(webView, applyIframe(@"<textarea style='width: 100px; height: 100px;'></textarea>"));

and here.

> Tools/TestWebKitAPI/Tests/WebKitCocoa/RequestTextInputContext.mm:180
> +    webViewLoadHTMLStringAndWaitForDOMLoadEvent(webView, applyIframe(@"<div contenteditable style='width: 100px; height: 100px;'></div>"));

and here.

[Daniel Bates]

Thanks Simon! @Darin, both bet right, no test failed. Landing the alternative...

[Daniel Bates]

Created attachment 395513 [details]
To Land

[Daniel Bates]

Committed r259551: <https://trac.webkit.org/changeset/259551>