Now that shadow DOM no longer hangs off of renderer, it seems more natural for HTMLTextFormControl instead of RenderTextControl to have inner*Element().
Created attachment 100167 [details] Patch
Comment on attachment 100167 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=100167&action=review > Source/WebCore/ChangeLog:14 > + shadow DOM for input and textarea elements clear. It accessing shadow DOM of input and textarea elements > + are now less error-prone because it doesn't depend on the lifetime of the render tree. Oops, typo. It should read "Accessing the shadow DOM of input and textarea elements are now less error prone because it no longer depends on the lifetime of the render tree".
Comment on attachment 100167 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=100167&action=review > Source/WebCore/rendering/RenderTextControl.h:56 > + HTMLElement* innerTextElement() const; Let's add a comment that this function should not be public. Otherwise, one who doesn't know the intention of this patch might make it public and revert the AccessibilityRenderObject.cpp change.
Comment on attachment 100167 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=100167&action=review >> Source/WebCore/rendering/RenderTextControl.h:56 >> + HTMLElement* innerTextElement() const; > > Let's add a comment that this function should not be public. > Otherwise, one who doesn't know the intention of this patch might make it public and revert the AccessibilityRenderObject.cpp change. That makes sense.
Now that r90662 has been reverted, I'd have to reland the patch for the bug 64133 first.
Committed r90775: <http://trac.webkit.org/changeset/90775>