1293443 - Coalesce show/hide pairs for the same child node

Status: RESOLVED INVALID

(Core :: Disability Access APIs, defect)

Description

[Michael Li]

If a child node gets shown and then hidden after, like when it is a text node that gets moved to be the child of another node before the document has completely loaded, then the show and hide events for the child under the original parent are unnecessary and will cause assertions; the child will be shown with the new parent anyway.

Comment 1

[Michael Li]

Attachment: Bug 1293443: Coalesce show/hide pairs for the same child node

Review commit: https://reviewboard.mozilla.org/r/70124/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/70124/

Comment 2

[alexander :surkov (:asurkov)]

Comment on attachment 8779081 [details]
Bug 1293443: Coalesce show/hide pairs for the same child node

https://reviewboard.mozilla.org/r/70124/#review67652

::: accessible/base/EventTree.cpp:453
(Diff revision 1)
> +#ifdef A11Y_LOG
> +        logging::MsgBegin("EVENTS_TREE", "coalescing show/hide pair");
> +        logging::AccessibleInfo("removing show/hide pair for", ev->GetAccessible());
> +        logging::MsgEnd();
> +#endif
> +        mDependentEvents.RemoveElementAt(idx);

it's not that easy as a previous hide event may keep text change event from the coalescing hide event. Also this is n^2 processing, that'd be a perf killer, if we don't keep the numbers low.

Is it something must or nice to have? how does it lead to the assertions?

Comment 3

[alexander :surkov (:asurkov)]

Comment on attachment 8779081 [details]
Bug 1293443: Coalesce show/hide pairs for the same child node

cancelling review for now

Comment 4

[Michael Li]

Closing since we didn't go with this approach.