Bug 123885 – AX: [ATK] <span> elements exposed through ATK when not needed

• Status: RESOLVED
• Resolution: FIXED
• Component: Accessibility

[Mario Sanchez Prada]

The patch for bug 121335 has introduced a change in how is determined when an accessibility object gets ignored that causes that all <span> elements show up in the accessibility tree even when they should not (e.g. not explicitly focusable due to tabindex, no ARIA rule mandating that, no meaningful accessible name...)

For that reason the following tests are failing now, as the StaticText objects inside those <span> elements are not getting properly folded into their parents anymore:

platform/gtk/accessibility/spans-paragraphs-and-divs.html
platform/gtk/accessibility/spans.html

[Radar WebKit Bug Importer]

<rdar://problem/15402535>

[Mario Sanchez Prada]

Created attachment 216177 [details]
Patch proposal

[chris fleizach]

Comment on attachment 216177 [details]
Patch proposal

View in context: https://bugs.webkit.org/attachment.cgi?id=216177&action=review

> Source/WebCore/accessibility/AccessibilityNodeObject.cpp:1925
> +

can all these checks move into AccessibilityObject?

> Source/WebCore/accessibility/AccessibilityObject.h:608
> +    virtual bool hasAccessibleName() const;

I think this should be called something like hasAccessibleAttributes() or hasRequiredInclusionAttributes() or hasAttributedRequiredForInclusion()

something to denote that this method should be used to determine whether to include it in the tree

[Mario Sanchez Prada]

Comment on attachment 216177 [details]
Patch proposal

View in context: https://bugs.webkit.org/attachment.cgi?id=216177&action=review

>> Source/WebCore/accessibility/AccessibilityNodeObject.cpp:1925
>> +
> 
> can all these checks move into AccessibilityObject?

Hmm... yes and no: we can move the MathML part but not the ariaAccessibilityDescription() call since that's a non-virtual method from AccessibilityNodeObject.

>> Source/WebCore/accessibility/AccessibilityObject.h:608
>> +    virtual bool hasAccessibleName() const;
> 
> I think this should be called something like hasAccessibleAttributes() or hasRequiredInclusionAttributes() or hasAttributedRequiredForInclusion()
> 
> something to denote that this method should be used to determine whether to include it in the tree

Believe it or not, I struggled to find a name for this and in the end this is the least bad I found.

hasAttributedRequiredForInclusion() sounds good to me, though

[chris fleizach]

(In reply to comment #4)
> (From update of attachment 216177 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=216177&action=review
> 
> >> Source/WebCore/accessibility/AccessibilityNodeObject.cpp:1925
> >> +
> > 
> > can all these checks move into AccessibilityObject?
> 
> Hmm... yes and no: we can move the MathML part but not the ariaAccessibilityDescription() call since that's a non-virtual method from AccessibilityNodeObject.
> 
> >> Source/WebCore/accessibility/AccessibilityObject.h:608
> >> +    virtual bool hasAccessibleName() const;
> > 
> > I think this should be called something like hasAccessibleAttributes() or hasRequiredInclusionAttributes() or hasAttributedRequiredForInclusion()
> > 
> > something to denote that this method should be used to determine whether to include it in the tree
> 
> Believe it or not, I struggled to find a name for this and in the end this is the least bad I found.
> 
> hasAttributedRequiredForInclusion() sounds good to me, though

should be: hasAttributesRequiredForInclusion()

[Mario Sanchez Prada]

Created attachment 216298 [details]
Patch proposal

(In reply to comment #5)
> [...]
> should be: hasAttributesRequiredForInclusion()

New patch attached renaming that function and moving the MathML bits up into AccessibilityObject

[EFL EWS Bot]

Comment on attachment 216298 [details]
Patch proposal

Attachment 216298 [details] did not pass efl-wk2-ews (efl-wk2):
Output: http://webkit-queues.appspot.com/results/21248312

[EFL EWS Bot]

Comment on attachment 216298 [details]
Patch proposal

Attachment 216298 [details] did not pass efl-ews (efl):
Output: http://webkit-queues.appspot.com/results/22738203

[kov's GTK+ EWS bot]

Comment on attachment 216298 [details]
Patch proposal

Attachment 216298 [details] did not pass gtk-ews (gtk):
Output: http://webkit-queues.appspot.com/results/22718202

[Mario Sanchez Prada]

Created attachment 216299 [details]
Patch proposal

[Mario Sanchez Prada]

Committed r158914: <http://trac.webkit.org/changeset/158914>