Bug 58419 – [chromium] Clicking on plugin for first time blocks focus shift

• Component: WebCore Misc.
• Status: RESOLVED FIXED

[Jeremy Moskovich]

Chromium bug: crbug.com/77172

Description:
The attached testcase contains a plugin and a textarea embedded in an iframe.

Open the testcase in Chromium, click on the plugin, then click back in the textarea and try to type some text.

Expected result:
You should be able to type into the textarea.

Actual result:
You can't type into the textarea, even though the focus ring is draw around it.

[Jeremy Moskovich]

Created attachment 89354 [details]
testcase

[Stuart Morgan]

Created attachment 89397 [details]
fix

Uses FocusController to handle the focus change, just as would happen for a non-plugin element (see the debugging in the Chromium bug for more details on why).

[Stuart Morgan]

Ping. Can one of you take a look at this? It's a small patch, and the issue it fixes is pretty nasty.

[Eric Seidel (no email)]

Comment on attachment 89397 [details]
fix

How do we test this?

[Eric Seidel (no email)]

If this is only reproducible in the full Chromium browser, we could/should still land a manual-test.

[Ryosuke Niwa]

Comment on attachment 89397 [details]
fix

View in context: https://bugs.webkit.org/attachment.cgi?id=89397&action=review

Can't we use eventSender to test this?

> Source/WebKit/chromium/ChangeLog:4
> +

You forgot bug title & bug url.

[Stuart Morgan]

(In reply to comment #4)
> How do we test this?

(In reply to comment #6)
> Can't we use eventSender to test this?

I have very little experience with WebKit development; I'll try to figure this out myself I guess (the docs seem pretty sparse), but pointers would be a lot more helpful than rhetorical questions I don't know the answers to.

(In reply to comment #6)
> You forgot bug title & bug url.

The bug link is there. I wasn't aware the title was necessary; I'll add it. (Perhaps the prepare-ChangeLog script should update the template if people are expected to include it, since I don't see any mention of it in either the template or the "how to make a patch" documentation.)

[Ryosuke Niwa]

(In reply to comment #7)
> I have very little experience with WebKit development; I'll try to figure this out myself I guess (the docs seem pretty sparse), but pointers would be a lot more helpful than rhetorical questions I don't know the answers to.

EventSender allows you to send arbitrary mouse events / keyboard events in layout tests.  I don't think there's any documentation about eventSender but you can see a couple of examples if you grep on LayoutTests.

> (In reply to comment #6)
> > You forgot bug title & bug url.
> 
> The bug link is there. I wasn't aware the title was necessary; I'll add it. (Perhaps the prepare-ChangeLog script should update the template if people are expected to include it, since I don't see any mention of it in either the template or the "how to make a patch" documentation.)

You're supposed to put the bug title & bug url in the consecutive lines and then have a blank line before describing what you're fixing.

[Stuart Morgan]

Created attachment 90114 [details]
fix v2

Test case added, based on looking at some of the surrounding test--thanks for the pointers. It runs as both a layout test and an interactive test (although the two modes work differently since I can't simulate clicks in normal mode).

[Tony Chang]

Comment on attachment 90114 [details]
fix v2

Adding cq+ since stuart isn't a committer.

[WebKit Commit Bot]

The commit-queue encountered the following flaky tests while processing attachment 90114 [details]:

http/tests/xmlhttprequest/cross-origin-no-credential-prompt.html bug 52249 (author: [e-mail])
The commit-queue is continuing to process your patch.

[WebKit Commit Bot]

Comment on attachment 90114 [details]
fix v2

Clearing flags on attachment: 90114

Committed r84270: <http://trac.webkit.org/changeset/84270>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[WebKit Commit Bot]

The commit-queue encountered the following flaky tests while processing attachment 90114 [details]:

http/tests/misc/favicon-loads-with-icon-loading-override.html bug 58412 (author: [e-mail])
The commit-queue is continuing to process your patch.

[WebKit Review Bot]

http://trac.webkit.org/changeset/84270 might have broken SnowLeopard Intel Release (Tests)
The following tests are not passing:
plugins/mouse-click-iframe-to-plugin.html

[Stuart Morgan]

Am I missing something in the test structure? It's all run in one pass, no waiting for callbacks, so I didn't think I needed notifyDone based on looking at other cases. It ran fine for me locally on Snow Leopard.

[Ryosuke Niwa]

I skipped the test in http://trac.webkit.org/changeset/84279 for WebKit2 but it doesn't seem to be the right fix then.  Please re-enable the test on WebKit2 when you fixed the test.

[Ryosuke Niwa]

Faling on Windows 7: http://build.webkit.org/results/Windows%207%20Release%20(Tests)/r84283%20(11851)/results.html

[Tony Chang]

Updated Skipped files in http://trac.webkit.org/changeset/84287.  The bug for tracking the mac/win failure is https://bugs.webkit.org/show_bug.cgi?id=58924