980307 - Remove DataContainerEvent dependency from accessibility

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[Masatoshi Kimura [:emk]]

Attachment: patch

Comment 1

[Trevor Saunders (:tbsaunde)]

Comment on attachment 8386756 [details] [diff] [review]
patch

I was hoping for something that didn't use dom events at all, but this is fine for now.

>+  customEvent->GetDetail(getter_AddRefs(detailVariant));
>+  if (!detailVariant)
>     return;
> 
>-  nsCOMPtr<nsIVariant> countVariant;
>-  dataEvent->GetData(NS_LITERAL_STRING("count"),
>-                     getter_AddRefs(countVariant));
>-  if (!countVariant)
>+  nsCOMPtr<nsIPropertyBag2> propBag;
>+  detailVariant->GetAsISupports(getter_AddRefs(propBag));

doesn't that use the operator nsISupports** on getter_AddRefs we'd like to get rid of? any way I think it would be clearer if you did the QI manually.

>+  nsCOMPtr<nsIPropertyBag2> propBag;
>+  detailVariant->GetAsISupports(getter_AddRefs(propBag));

same

might not hurt to have someone review the layout bits, but they lgtm

Comment 2

[Masatoshi Kimura [:emk]]

Attachment: patch v2

An automated test (test_tree.xul) caught errors of the patch.

(In reply to Trevor Saunders (:tbsaunde) from comment #1)
> I was hoping for something that didn't use dom events at all, but this is
> fine for now.

Maybe it is impossible unless we rewrite the test using C++.

> >+  customEvent->GetDetail(getter_AddRefs(detailVariant));
> >+  if (!detailVariant)
> >     return;
> > 
> >-  nsCOMPtr<nsIVariant> countVariant;
> >-  dataEvent->GetData(NS_LITERAL_STRING("count"),
> >-                     getter_AddRefs(countVariant));
> >-  if (!countVariant)
> >+  nsCOMPtr<nsIPropertyBag2> propBag;
> >+  detailVariant->GetAsISupports(getter_AddRefs(propBag));
> 
> doesn't that use the operator nsISupports** on getter_AddRefs we'd like to
> get rid of? any way I think it would be clearer if you did the QI manually.

It was not a style nit but an actual fatal bug. We can't receive the interface pointer without QI'ing. This bug caused a cryptic test crash. Thanks to pointing out. I wish nsCOMPtr could have caught this...

I think I made enough changes that are qualified to have another round of review.

Try runs with the new patch:
https://tbpl.mozilla.org/?tree=Try&rev=60b58c4ea9e2
https://tbpl.mozilla.org/?tree=Try&rev=81580be40d79

Comment 3

[Masatoshi Kimura [:emk]]

(In reply to Masatoshi Kimura [:emk] from comment #2)
> I wish nsCOMPtr could have caught this...

I found bug 345123.

Comment 4

[Masatoshi Kimura [:emk]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/9b0216df6e2a

Comment 5

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/9b0216df6e2a