Bug 136273 – overflow:scroll elements should not latch to the body if the body is overflow:hidden

• Status: RESOLVED
• Resolution: FIXED
• Component: Layout and Rendering

[Beth Dakin]

overflow:scroll elements should not latch to the body if the body is overflow:hidden. Right now, since we do latch to the body, this prevents you from rubber-banding an overflow area when you are pinned to the edge of the direction you are trying to rubber-band. That behavior feels wrong when the gesture results in absolutely nothing happening, which is the case when the body is overflow:hidden. This affects the web inspector.

[Beth Dakin]

Created attachment 237191 [details]
Patch

[WebKit Commit Bot]

Attachment 237191 [details] did not pass style-queue:


ERROR: Source/WebCore/ChangeLog:9:  You should remove the 'No new tests' and either add and list tests, or explain why no new tests were possible.  [changelog/nonewtests] [5]
Total errors found: 1 in 10 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Beth Dakin]

Created attachment 237233 [details]
Patch

Should fix Windows and style problems.

[Darin Adler]

Comment on attachment 237233 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=237233&action=review

> Source/WebCore/page/FrameView.cpp:3268
> +bool FrameView::isScrollable(Scrollability defineScrollable)

I would call this definitionOfScrollable or scrollabilityDefinition or some other noun phrase. “define scrollable” is a verb phrase.

> Source/WebCore/page/FrameView.cpp:3310
> +    return parentView->enclosingLayer() ? parentView->enclosingLayer()->hasScrollableOrRubberbandableAncestor() : false;

Could the local variable be the layer rather than the parentView? The point of the local variable is to make the null check possible, and apparently the parent view can’t be null.

I prefer && for these rather than ? : false.

    RenderLayer* layer = parentFrameView()->renderView()->enclosingLayer();
    return layer && layer->hasScrollableOrRubberbandableAncestor();

> Source/WebCore/page/FrameView.h:405
> +    bool isScrollable(Scrollability defineScrollable = Scrollability::Scrollable);

Should omit the argument name here. Unless the new one makes it really easy to understand.

> Source/WebCore/page/FrameView.h:406
> +    virtual bool hasScrollableOrRubberbandableAncestor() override;

Can this be private instead of public?

> Source/WebCore/rendering/RenderLayer.cpp:3068
> +bool RenderLayer::hasScrollableOrRubberbandableAncestor()

Seems messy for this function to special-case RenderView and RenderBox rather than using a virtual function to deal with that.

> Source/WebCore/rendering/RenderLayer.cpp:3078
> +            return nextLayer;

Should say return true here. This is just a roundabout way to write it!

> Source/WebCore/rendering/RenderLayer.h:442
> +    virtual bool hasScrollableOrRubberbandableAncestor() override;

Does this need to be public?

> Source/WebCore/rendering/RenderListBox.cpp:792
> +    return enclosingLayer() ? enclosingLayer()->hasScrollableOrRubberbandableAncestor() : false;

I prefer && instead of ? : false.

[Darin Adler]

Comment on attachment 237233 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=237233&action=review

>> Source/WebCore/rendering/RenderLayer.h:442
>> +    virtual bool hasScrollableOrRubberbandableAncestor() override;
> 
> Does this need to be public?

Oops, meant to remove that comment. Clearly it does.

[Beth Dakin]

(In reply to comment #4)
> (From update of attachment 237233 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=237233&action=review
> 
> > Source/WebCore/page/FrameView.cpp:3268
> > +bool FrameView::isScrollable(Scrollability defineScrollable)
> 
> I would call this definitionOfScrollable or scrollabilityDefinition or some other noun phrase. “define scrollable” is a verb phrase.
> 

Changed this to definitionOfScrollable

> > Source/WebCore/page/FrameView.cpp:3310
> > +    return parentView->enclosingLayer() ? parentView->enclosingLayer()->hasScrollableOrRubberbandableAncestor() : false;
> 
> Could the local variable be the layer rather than the parentView? The point of the local variable is to make the null check possible, and apparently the parent view can’t be null.
> 

Actually, I think we do need null checks here! Eek! Added those.

> I prefer && for these rather than ? : false.
> 
>     RenderLayer* layer = parentFrameView()->renderView()->enclosingLayer();
>     return layer && layer->hasScrollableOrRubberbandableAncestor();
> 

I changed this to use &&.

> > Source/WebCore/page/FrameView.h:405
> > +    bool isScrollable(Scrollability defineScrollable = Scrollability::Scrollable);
> 
> Should omit the argument name here. Unless the new one makes it really easy to understand.
> 

I think that the new one (definitionOfScrollable) is helpful, so I left it.

> > Source/WebCore/rendering/RenderLayer.cpp:3068
> > +bool RenderLayer::hasScrollableOrRubberbandableAncestor()
> 
> Seems messy for this function to special-case RenderView and RenderBox rather than using a virtual function to deal with that.
> 

I added a virtual function on RenderBox, overridden by RenderView.

> > Source/WebCore/rendering/RenderLayer.cpp:3078
> > +            return nextLayer;
> 
> Should say return true here. This is just a roundabout way to write it!
> 

Silly! The code is not here anymore now that I added the virtual function on RenderBox.

> > Source/WebCore/rendering/RenderListBox.cpp:792
> > +    return enclosingLayer() ? enclosingLayer()->hasScrollableOrRubberbandableAncestor() : false;
> 
> I prefer && instead of ? : false.

Changed to &&.

[Beth Dakin]

http://trac.webkit.org/changeset/173014