NVDA no longer speaking aria-live "assertive" messages in ChatZilla

[nvaccessAuto]

Reported by MarcoZehe on 2014-06-02 08:50
I don't know when this broke, but whenever I am mentioned in ChatZilla, run in Firefox Nightly, NVDA no longer speaks those messages that get aria-live "assertive", or the children get container-live "assertive". The ones with aria-live or container-live "polite" are all spoken OK. So are those with an impoicit aria-live of "assertive" like in role "alert".

[nvaccessAuto]

Comment 2 by jteh on 2014-08-14 14:19
This occurs because aria-live="assertive" is set on the table row for mentions, which makes NVDA see this as the root of the region. We don't handle show events for region roots and this is a table, so it doesn't fire textInserted, which means the only event that we can use here gets ignored. Because this is actually inside a live region (even though the live type is overridden), we shouldn't treat this as the root.

One way to solve this is to check whether the parent is a live region. If it is, this obviously isn't the root. This is a bit ugly, but I can't think of a better way.

I'm attaching a test case which distills this. Pressing the button will add two rows: one which inherits polite from the table and one which is marked assertive.

[nvaccessAuto]

Attachment test4169.html added by jteh on 2014-08-14 14:19
Description:
Test case.

[nvaccessAuto]

Comment 3 by Michael Curran <mick@... on 2014-08-18 01:50
In [29dfe7f]:

IA2 live region support: an object should not be considered the root of a region because it has a valid "live" attribute, as this aria-live may simply be changing the live type. Rather check the object's parent for a "container-live" attribute which will be the propagated next-highest aria-live. If the parent has a valid "container-live" attribute, clearly the child object cannot be the root.
Re #4169

[nvaccessAuto]

Comment 4 by Michael Curran <mick@... on 2014-08-18 01:51
In [b3e017e]:

Merge branch 't4169' into next. Incubates #4169

Changes:
Added labels: incubating

[nvaccessAuto]

Comment 5 by Michael Curran <mick@... on 2014-08-22 02:12
In [af96d9f]:

Merge branch 't4169' into next. Incubates #4169

[nvaccessAuto]

Comment 6 by Michael Curran <mick@... on 2014-09-04 04:25
In [f93120b]:

Merge branch 't4169'. Fixes #4169

Changes:
Removed labels: incubating
State: closed

[nvaccessAuto]

Comment 7 by mdcurran on 2014-09-04 04:27
Changes:
Milestone changed from None to 2014.4