Bug 161734 – AX: Aria-activedescendant not supported

• Component: Accessibility
• Status: RESOLVED FIXED

Hello,

I am working for Amadeus, a company that provides IT solutions for the travel industry.
We are developing a flight booking website that should be accessible to screen reader users.

We are currently encountering a specific issue with VoiceOver with the support of aria-activedescendant attribute.
This has been used to implement a datepicker and an autocomplete widget.
When navigating on the active element, VoiceOver is not reading it (suggestion in the autocomplete list and date in the datepicker).

Please launch the URL below to reproduce the issue.
The concerned widget are the city picker and the datepicker in the search bar at the top of the page.

Testing environment
We are testing on MAC OS El Capitain 10.11 with Safari browser and VoiceOver.

Best regards

URL for test page

https://pcm.uat01.amadeus.com/plnext/demo/Override.action?SO_SITE_RUI_HIDE_MDF_SRC=FALSE&TRIP_FLOW=YES&BOOKING_FLOW=REVENUE&B_LOCATION_1=KEF&B_LOCATION_2=LON&B_DATE_1=201612110000&B_ANY_TIME_1=TRUE&E_LOCATION_1=LON&E_LOCATION_2=KEF&B_DATE_2=201612220000&B_ANY_TIME_2=TRUE&TRIP_TYPE=R&EXTERNAL_ID=BOOKING&PRICING_TYPE=O&EMBEDDED_TRANSACTION=FlexPricerAvailability&DISPLAY_TYPE=2&ARRANGE_BY=D&SO_SITE_MATRIX_CALENDAR=FALSE&REFRESH=0&COMMERCIAL_FARE_FAMILY_1=OWCOACHEU&COMMERCIAL_FARE_FAMILY_2=OWCOMFVIA&COMMERCIAL_FARE_FAMILY_3=OWCOMFVIA&DATE_RANGE_VALUE_1=3&DATE_RANGE_VALUE_2=3&DATE_RANGE_QUALIFIER_1=C&DATE_RANGE_QUALIFIER_2=C&TRAVELLER_TYPE_1=ADT&TRAVELLER_TYPE_2=&TRAVELLER_TYPE_3=&TRAVELLER_TYPE_4=&TRAVELLER_TYPE_5=&TRAVELLER_TYPE_6=&TRAVELLER_TYPE_7=&TRAVELLER_TYPE_8=&TRAVELLER_TYPE_9=&HAS_INFANT_1=FALSE&HAS_INFANT_2=FALSE&HAS_INFANT_3=FALSE&HAS_INFANT_4=FALSE&HAS_INFANT_5=FALSE&HAS_INFANT_6=FALSE&HAS_INFANT_7=FALSE&HAS_INFANT_8=FALSE&HAS_INFANT_9=FALSE&SO_SITE_EXT_PSPURL=&SO_SITE_EXT_MerchantID=FI&SO_GL=&SO_SITE_OFFICE_ID=KEFFI08AA&SO_SITE_QUEUE_OFFICE_ID=KEFFI08AA&SO_SITE_QUEUE_CATEGORY=0C0&SO_SITE_MOP_EXT=TRUE&SO_SITE_FOP_MODE=SITE&SITE=5APDDEMO&LANGUAGE=GB&SO_SITE_APIV2_SERVER_USER_ID=GUEST&SO_SITE_APIV2_SERVER=194.156.170.78&SO_SITE_APIV2_SERVER_PWD=TAZ&SO_SITE_CORPORATE_ID=OCGPDT&SO_SITE_SI_SAP=1ASIXJCP&SO_SITE_SI_SERVER_PORT=18033&SO_SITE_SI_SERVER_IP=193.23.185.67&SO_SITE_SI_USER=UNSET&SO_SITE_SI_PASSWORD=UNSET&SO_SITE_SI_1AXML_FROM=SEP_JCP&SO_SITE_FQ_INTERFACE_ACTIVE=FALSE&SO_SITE_HOST_TRACE_ACTIVE=TRUE&SO_SITE_FP_TRACES_ON=FALSE

[Radar WebKit Bug Importer]

<rdar://problem/28202679>

Hello,

Any news on this issue?

Could you please confirm that there is a better support of aria-activedescendant with the Sierra release?

Thanks & regards

[Jesse Beach]

We are running up against this issue as well at Facebook. Our top-level search bar experience is hampered on OSX because of this issue. I've put together a codepen that can be tested on OSX with VoiceOver. Note that the items targeted by the aria-activedescendant attribute on the combobox input are not announced by VoiceOver.

http://codepen.io/jessebeach/pen/OWBJRy

I tested this behavior with JAWS 16 on Windows using Chrome and the example codepen behaved as expected -- the items targeted by aria-activedescendant are announced when the id value is changed in the attribute.

This gap is functionality for Safari and Chrome on OSX is a serious impediment for low vision and blind web users.

[Jesse Beach]

Related to: https://bugs.webkit.org/show_bug.cgi?id=167680

[Jesse Beach]

Update: aria-activedescendant works in Chrome (not Safari) in combination with a role of textbox. Combined with a role of combobox, aria-activedescendant does not work in Chrome or Safari.

Hello,

In our case, the "aria-activedescendant" is put on a simple text input without any role.

Any update please?

Thanks

[chris fleizach]

Created attachment 330217 [details]
patch

[chris fleizach]

Created attachment 330218 [details]
patch

[EWS Watchlist]

Attachment 330218 [details] did not pass style-queue:


ERROR: Source/WebCore/accessibility/AccessibilityRenderObject.cpp:2506:  One line control clauses should not use braces.  [whitespace/braces] [4]
Total errors found: 1 in 9 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[chris fleizach]

Created attachment 330219 [details]
patch

[chris fleizach]

Note: we still need a fix in VoiceOver before this will work.

[EWS Watchlist]

Comment on attachment 330219 [details]
patch

Attachment 330219 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.webkit.org/results/5844933

New failing tests:
accessibility/lists.html

[EWS Watchlist]

Created attachment 330221 [details]
Archive of layout-test-results from ews103 for mac-elcapitan

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews103  Port: mac-elcapitan  Platform: Mac OS X 10.11.6

[EWS Watchlist]

Comment on attachment 330219 [details]
patch

Attachment 330219 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-queues.webkit.org/results/5844946

New failing tests:
accessibility/lists.html

[EWS Watchlist]

Created attachment 330222 [details]
Archive of layout-test-results from ews107 for mac-elcapitan-wk2

The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews107  Port: mac-elcapitan-wk2  Platform: Mac OS X 10.11.6

[Jesse Beach]

```
if (activeDescendant->isDescendantOfObject(element.get()))
```

@cfleizach, is this assuming an ancestor/descendant relationship for the potential active descendent items of the controlling element? aria-owns would create such a relationship in the AX Tree, but aria-controls can also designate a container for "descendants", noted in ARIA 1.1 (https://www.w3.org/TR/wai-aria-1.1/#aria-activedescendant) and no such ancestor/descendant relationship would exist in the AX Tree in this case.

[EWS Watchlist]

Comment on attachment 330219 [details]
patch

Attachment 330219 [details] did not pass mac-debug-ews (mac):
Output: http://webkit-queues.webkit.org/results/5844989

New failing tests:
svg/wicd/rightsizing-grid.xhtml
accessibility/lists.html

[EWS Watchlist]

Created attachment 330223 [details]
Archive of layout-test-results from ews115 for mac-elcapitan

The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews115  Port: mac-elcapitan  Platform: Mac OS X 10.11.6

[Jesse Beach]

Aha!

```
<input type="text" role="combobox" id="combobox" aria-controls="list" aria-label="Combobox">
```

Sorry, I should have kept reading before commenting.

From the ARIA 1.1 spec:

"When a combobox is expanded, authors MUST ensure it contains or owns an element that has a role of listbox, tree, grid, or dialog."

You've accounted for `list` in this patch, not `listbox`. Are `tree`, `grid` and `dialog` accounted for as well?

[Jesse Beach]

@cfleizach, here are the official examples of comboboxes from that ARIA 1.1 Authoring Practices Guide. These examples could be used to create test cases.

https://www.w3.org/TR/wai-aria-practices-1.1/examples/combobox/aria1.1pattern/listbox-combo.html

Thank you for working on this!

[chris fleizach]

Will take a look at those other cases. thanks for double checking me!

[chris fleizach]

I don't think the dialog case needs anything to be done here, since focus needs to be managed by the author between dialog and combobox

[chris fleizach]

Created attachment 330230 [details]
patch

[EWS Watchlist]

Attachment 330230 [details] did not pass style-queue:


ERROR: Source/WebCore/accessibility/AccessibilityRenderObject.cpp:3380:  Boolean expressions that span multiple lines should have their operators on the left side of the line instead of the right side.  [whitespace/operators] [4]
ERROR: Source/WebCore/accessibility/AccessibilityRenderObject.cpp:3380:  Multi line control clauses should use braces.  [whitespace/braces] [4]
Total errors found: 2 in 10 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[chris fleizach]

Created attachment 330231 [details]
patch

[Joanmarie Diggs]

Note to gardeners: The new test (aria-combobox-control-owns-elements.html) will time out. That is due to bug 171185 and/or bug 171186.

[WebKit Commit Bot]

Comment on attachment 330231 [details]
patch

Clearing flags on attachment: 330231

Committed r227144: <https://trac.webkit.org/changeset/227144>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Jesse Beach]

WOOOOHHOOOOOO!!!

Hello Chris,

Could you please provide on which version the patch has been applied?

We tested on High Sierra 10.13.3 and the issue still persists.

Thank you

[chris fleizach]

(In reply to helene.vinh from comment #31)
> Hello Chris,
> 
> Could you please provide on which version the patch has been applied?
> 
> We tested on High Sierra 10.13.3 and the issue still persists.
> 
> Thank you

I don't know when it will be included in a public release, but it also requires a change in VoiceOver so there are a few moving parts.

[James Craig]

*** Bug 126058 has been marked as a duplicate of this bug. ***