Bug 255138 – AX: Eliminate all ways AXCoreObject::elementRect could be called on an AXIsolatedObject

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

[Tyler Wilcock]

In a future patch, we will be caching the AXCoreObject::relativeFrame of each isolated object. The relative frame is the position of an object relative to the current viewport (i.e. relative frame is affected by things like scrolling). This is different than `AXCoreObject::elementRect`, which is the "absolute" rect (relative to the root document origin, not effected by scrolling).

To avoid the need to cache both `elementRect` and `relativeFrame`, we should eliminate all ways `elementRect` could be called on an `AXIsolatedObject` in the wrapper and elsewhere.

[Radar WebKit Bug Importer]

<rdar://problem/107741185>

[Tyler Wilcock]

Created attachment 465807 [details]
Patch

[chris fleizach]

Can you fix spacing after : here

1768        auto size = backingObject->size();
 1769        return [NSValue valueWithSize: NSMakeSize(size.width(),

[Tyler Wilcock]

Created attachment 465810 [details]
Patch

[chris fleizach]

Comment on attachment 465810 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=465810&action=review

> Source/WebCore/accessibility/mac/WebAccessibilityObjectWrapperMac.mm:1392
> +    return @(cgPoint);

do we need two lines of code here can can we do in one

also does @() turn this into a NSValue? I had always hoped it would, but don't know if I tested it

> Source/WebCore/accessibility/mac/WebAccessibilityObjectWrapperMac.mm:1769
> +        return [NSValue valueWithSize:NSMakeSize(size.width(), size.height())];

is there no overload for

(NSValue *)backingObject->size()

[Tyler Wilcock]

Created attachment 465815 [details]
Patch

[Tyler Wilcock]

(In reply to chris fleizach from comment #5)
> Comment on attachment 465810 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=465810&action=review
> 
> > Source/WebCore/accessibility/mac/WebAccessibilityObjectWrapperMac.mm:1392
> > +    return @(cgPoint);
> 
> do we need two lines of code here can can we do in one
> 
> also does @() turn this into a NSValue? I had always hoped it would, but
> don't know if I tested it
Fixed this. And yes, @() on "some" (quoting https://clang.llvm.org/docs/ObjectiveCLiterals.html) structs will result in an NSValue wrapper. The basic CG structs are included in this "some" (I just confirmed it). 

> > Source/WebCore/accessibility/mac/WebAccessibilityObjectWrapperMac.mm:1769
> > +        return [NSValue valueWithSize:NSMakeSize(size.width(), size.height())];
> 
> is there no overload for
> 
> (NSValue *)backingObject->size()
Checked and unfortunately there's no overload to automatically turn an IntSize into an NSValue.

[EWS]

Committed 262752@main (2d9b495ba023): <https://commits.webkit.org/262752@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 465815 [details].