Bug 106814 – Positioned children of an overflow:visible container should ignore scroll offset when updating layer position

• Component: New Bugs
• Status: RESOLVED FIXED

[Tien-Ren Chen]

Positioned childs of an overflow:visible container should ignore scroll offset when updating layer position

[Tien-Ren Chen]

Created attachment 182609 [details]
Patch

[Simon Fraser (smfr)]

Comment on attachment 182609 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=182609&action=review

> Source/WebCore/ChangeLog:3
> +        Positioned childs of an overflow:visible container should ignore scroll offset when updating layer position

childs -> children

> Source/WebCore/ChangeLog:11
> +        and triggers a RenderGeometryMap assertion falure.

Was that seen on any real sites?

> LayoutTests/ChangeLog:3
> +        Positioned childs of an overflow:visible container should ignore scroll offset when updating layer position

children

[Tien-Ren Chen]

Created attachment 182628 [details]
Patch

[Simon Fraser (smfr)]

From the cr bug, this fixes assertions when loading
http://flightaware.com/live/flight/DLH454/history/20121222/0915Z/EDDF/KSFO

[Tien-Ren Chen]

(In reply to comment #4)
> From the cr bug, this fixes assertions when loading
> http://flightaware.com/live/flight/DLH454/history/20121222/0915Z/EDDF/KSFO

It is probably dependent on viewport size and how do we handle Element::focus(). Somehow Element::focus() calls into RenderLayer::scrollRectToVisible, and it tries to scroll whatever blocks in the container chain.

My opinion is that it would be nice if we also change RenderLayer::clampScrollOffset so it would always return (0, 0) when there is no overflow clip. Does that sounds reasonable to you?

[WebKit Review Bot]

Comment on attachment 182628 [details]
Patch

Clearing flags on attachment: 182628

Committed r139669: <http://trac.webkit.org/changeset/139669>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.