Bug 205210 – Isolated object implementation of parameterized attribute SelectTextWithCriteria.

• Component: New Bugs
• Status: RESOLVED FIXED

[Andres Gonzalez]

Isolated object implementation of parameterized attribute SelectTextWithCriteria.

[Andres Gonzalez]

Created attachment 385625 [details]
Patch

[chris fleizach]

Comment on attachment 385625 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=385625&action=review

> Source/WebCore/accessibility/AccessibilityRenderObject.cpp:-3577
> -    ASSERT(roleValue() == AccessibilityRole::TabList);

worried that this might populate tabChildren for elements not expected to have any... maybe we can change this to

if (role != TabList)
   return

> Source/WebCore/accessibility/isolatedtree/AXIsolatedTreeNode.cpp:696
> +Vector<RefPtr<Range>> AXIsolatedObject::findTextRanges(AccessibilitySearchTextCriteria const& criteria) const

what's the implication of returning a Range from the main thread to a secondary thread?

> Tools/WebKitTestRunner/InjectedBundle/AccessibilityController.cpp:101
>              m_semaphore.signal();

can you rename execute to something that is more meaningful in explaining that it will dispatch to the main thread

> Tools/WebKitTestRunner/InjectedBundle/AccessibilityController.cpp:104
> +        CFRunLoopRunInMode(kCFRunLoopDefaultMode, .25, false);

put a comment why we need to spin the run loop

[Andres Gonzalez]

Created attachment 385642 [details]
Patch

[Andres Gonzalez]

(In reply to chris fleizach from comment #2)
> Comment on attachment 385625 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=385625&action=review
> 
> > Source/WebCore/accessibility/AccessibilityRenderObject.cpp:-3577
> > -    ASSERT(roleValue() == AccessibilityRole::TabList);
> 
> worried that this might populate tabChildren for elements not expected to
> have any... maybe we can change this to
> 
> if (role != TabList)
>    return

Done.
> 
> > Source/WebCore/accessibility/isolatedtree/AXIsolatedTreeNode.cpp:696
> > +Vector<RefPtr<Range>> AXIsolatedObject::findTextRanges(AccessibilitySearchTextCriteria const& criteria) const
> 
> what's the implication of returning a Range from the main thread to a
> secondary thread?

This function is only used in the wrapper and will be always be executed inside a block dispatched to the main thread. You are right, it won't work if we return this value outside the main thread, but that is not happening.
> 
> > Tools/WebKitTestRunner/InjectedBundle/AccessibilityController.cpp:101
> >              m_semaphore.signal();
> 
> can you rename execute to something that is more meaningful in explaining
> that it will dispatch to the main thread

Renamed it executeOnAXThreadIfPossible.
> 
> > Tools/WebKitTestRunner/InjectedBundle/AccessibilityController.cpp:104
> > +        CFRunLoopRunInMode(kCFRunLoopDefaultMode, .25, false);
> 
> put a comment why we need to spin the run loop

Done.

[Andres Gonzalez]

Created attachment 385766 [details]
Patch

[chris fleizach]

Comment on attachment 385766 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=385766&action=review

> Source/WebCore/accessibility/isolatedtree/AXIsolatedTree.h:50
> +    AXObjectCache* axObjectCache() const { return m_axObjectCache; }

feel like we should assert main thread in this getter

[Andres Gonzalez]

(In reply to chris fleizach from comment #6)
> Comment on attachment 385766 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=385766&action=review
> 
> > Source/WebCore/accessibility/isolatedtree/AXIsolatedTree.h:50
> > +    AXObjectCache* axObjectCache() const { return m_axObjectCache; }
> 
> feel like we should assert main thread in this getter

Will do. Thanks.

[WebKit Commit Bot]

Comment on attachment 385766 [details]
Patch

Clearing flags on attachment: 385766

Committed r253565: <https://trac.webkit.org/changeset/253565>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/57976699>