1362063 - replace delayed ValidateARIAOwned on sync DOM tree traversal

Status: RESOLVED FIXED

(Core :: Disability Access APIs, enhancement)

Description

[alexander :surkov (:asurkov)]

Attachment: patch

ValidateARIAOwned keeps dead DOM/frame's accessible in the tree longer than we want and it has more sophisticated logic than the straightforward traverse DOM tree approach; I suspect it may lead to the broken tree problem. Of course on a downside, it takes us to traverse a removed tree twice: traverse an accessible tree, and then traverse a DOM tree, but that shouldn't hurt us bad.

Comment 1

[David Bolter [:davidb] (NeedInfo me for attention)]

Comment on attachment 8864551 [details] [diff] [review]
patch

Review of attachment 8864551 [details] [diff] [review]:
-----------------------------------------------------------------

::: accessible/generic/DocAccessible.cpp
@@ +1986,5 @@
> +  mt.BeforeRemoval(aChild);
> +
> +  if (aChild->IsRelocated()) {
> +    nsTArray<RefPtr<Accessible> >* owned = mARIAOwnsHash.Get(parent);
> +    owned->RemoveElement(aChild);

Is it worth asserting owned?

Comment 2

[alexander :surkov (:asurkov)]

(In reply to David Bolter [:davidb] from comment #1)

> > +  mt.BeforeRemoval(aChild);
> > +
> > +  if (aChild->IsRelocated()) {
> > +    nsTArray<RefPtr<Accessible> >* owned = mARIAOwnsHash.Get(parent);
> > +    owned->RemoveElement(aChild);
> 
> Is it worth asserting owned?

you mean debug assert before crash? I can add it.

Comment 3

[alexander :surkov (:asurkov)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/6669f26d7f1366221b9edd3d82ac699e98101dc7
Bug 1362063 - replace delayed ValidateARIAOwned on straightforward DOM tree traversal, r=davidb

Comment 4

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/6669f26d7f13