Bug 113591 – REGRESSION (r145870): When parent has border-radius, the overflow part of its children will not be able to catch hits.

• Component: CSS
• Status: RESOLVED FIXED

[Kevin M. Dean]

This in WordPress.

Go to Appearance -> Widgets

In the Right Side bar, click on the downward triangle to open the Text widget. Click in the Title input field and the textarea back and forth. Sometimes the character insertion point will appear, but eventually it won't after a few clicks. Sometimes it doesn't initially appear at all.

I've set up a wordpress test area if you want to login and take a look. Just respond here or send me an email directly for the access information.

[Alexey Proskuryakov]

> Just respond here or send me an email directly for the access information.

I'd like to take a look.

[Kevin M. Dean]

(In reply to comment #1)
> I'd like to take a look.

Access info emailed.

[Alexey Proskuryakov]

Regressed in <http://trac.webkit.org/changeset/145870>.

[Alexey Proskuryakov]

<rdar://problem/13546326>

[Xidorn Quan]

> Just respond here or send me an email directly for the access information.

I would like to have a look.

[Kevin M. Dean]

(In reply to comment #5)
> I would like to have a look.

Access info emailed.

[Xidorn Quan]

Created attachment 196079 [details]
testcase

This testcase shows the bug. The red area is a anchor which should all be clickable, but when it is overflow and the parent has border-radius, the outside part will be unclickable.

[Xidorn Quan]

Created attachment 196090 [details]
patch

This patch should fix this bug.

[Xidorn Quan]

Created attachment 196092 [details]
patch

It's my fault that some unrelated code was added in the previous patch.

[Simon Fraser (smfr)]

Comment on attachment 196092 [details]
patch

View in context: https://bugs.webkit.org/attachment.cgi?id=196092&action=review

> LayoutTests/ChangeLog:9
> +        * fast/borders/border-radius-child-expected.txt: Added.
> +        * fast/borders/border-radius-child.html: Added.

I'm concerned that this test will be very sensitive to slight differences between platforms.

Also, you'd be able to make a test without eventSender and handlers by using elementFromPoint.

Did the bug affect non-form controls with border radii and overflow? That might make for a more reliable test case.

[Xidorn Quan]

(In reply to comment #10)
> (From update of attachment 196092 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=196092&action=review
> 
> > LayoutTests/ChangeLog:9
> > +        * fast/borders/border-radius-child-expected.txt: Added.
> > +        * fast/borders/border-radius-child.html: Added.
> 
> I'm concerned that this test will be very sensitive to slight differences between platforms.

As you can see from the patch status, the test has been passed on all platforms.

> Also, you'd be able to make a test without eventSender and handlers by using elementFromPoint.

I'll try it.

> Did the bug affect non-form controls with border radii and overflow? That might make for a more reliable test case.

Yes, it did, as you can see from the testcase I uploaded before. Consequently, some of the information provided by the reporter is inaccuracy: component should be CSS instead of Forms, and platform should be All. You may assign this bug to me so that I can correct it.

[Xidorn Quan]

Created attachment 196264 [details]
patch

Test has been modified according to Simon Fraser's suggestion.

[Alexey Proskuryakov]

> As you can see from the patch status, the test has been passed on all platforms.

I think that currently, only one of the chromium EWS bots runs tests, everyone else just builds.

[Xidorn Quan]

(In reply to comment #13)
> > As you can see from the patch status, the test has been passed on all platforms.
> 
> I think that currently, only one of the chromium EWS bots runs tests, everyone else just builds.

Anyway, I do not think that the test is platform-sensitive. Could you give the reason and, if possible, some suggestion to this test?

[Simon Fraser (smfr)]

(In reply to comment #14)
> (In reply to comment #13)
> > > As you can see from the patch status, the test has been passed on all platforms.
> > 
> > I think that currently, only one of the chromium EWS bots runs tests, everyone else just builds.
> 
> Anyway, I do not think that the test is platform-sensitive. Could you give the reason and, if possible, some suggestion to this test?

Make a really big div with larger rounded corners, and overflowing content.

[Xidorn Quan]

(In reply to comment #15)
> (In reply to comment #14)
> > (In reply to comment #13)
> > > > As you can see from the patch status, the test has been passed on all platforms.
> > > 
> > > I think that currently, only one of the chromium EWS bots runs tests, everyone else just builds.
> > 
> > Anyway, I do not think that the test is platform-sensitive. Could you give the reason and, if possible, some suggestion to this test?
> 
> Make a really big div with larger rounded corners, and overflowing content.

But the bug is, in fact, absolutely unrelated to how large the rounded corners are, and the overflowed part has been big enough IMO.

[Xidorn Quan]

Created attachment 196445 [details]
patch

[WebKit Review Bot]

Comment on attachment 196445 [details]
patch

Clearing flags on attachment: 196445

Committed r147635: <http://trac.webkit.org/changeset/147635>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.