[Windows] Trackers # is not announced on mouseover in the Privacy Widget
Categories
(Firefox :: New Tab Page, defect, P3)
Tracking
()
People
(Reporter: csasca, Assigned: rhamoui, NeedInfo)
References
(Blocks 1 open bug)
Details
(Keywords: access, Whiteboard: [hnt])
Attachments
(1 file, 1 external file)
|
1.97 MB,
video/quicktime
|
Details |
Affected versions
- Firefox Nightly 156.0a1 (2026-08-13)
Affected platforms
- Windows 11
Preconditions
- NVDA screenreader on
- Have some blocked trackers shown inside the Privacy widget
Steps to reproduce
- Reach the Privacy widget in a new tab page
- Hover over the actual blocked trackers number
Expected result
- The number is read by the screenreader
Actual result
- The number is not read by the screenreader
Regression range
- New feature
Additional Notes
- The issue with sound can be seen in the following attachment
Comment 1•3 months ago
|
||
The number is not read on hover which isn't ideal but not as severe as not read at all.
Updated•3 months ago
|
Updated•3 months ago
|
| Assignee | ||
Comment 3•3 months ago
|
||
I'm not personally sure how to emulate this on a Macbook, and can't find the right resources.
Eitan, is there a way of emulating read-on-hover on MacOS Voiceover that can help me fix this bug?
Comment 4•3 months ago
|
||
I'm not 100% sure this is reproducible on mac, you can try. Search for the accessibility preference "speak item under the pointer" and see if it works.
| Assignee | ||
Comment 5•3 months ago
|
||
Ah okay I was able to have the MacOS voiceover work with cursor navigation, and i was unable to reproduce the issue because the numbers were actually being read on my end. I've just attached a file for it.
Catalin, would you confirm if the issue is still persisting on your end?
| Assignee | ||
Comment 6•3 months ago
|
||
| Assignee | ||
Updated•3 months ago
|
| Reporter | ||
Comment 7•3 months ago
•
|
||
Well yes, macOS and Ubuntu works as expected, this one is specific for Windows and can confirm that this is still reproducible on Firefox 156.0a1 (2026-08-18) and Firefox 154.0 (with 156.1.20260818.43226 new tab addon) . For macOS we have only logged Bug 2063274.
Comment 8•2 months ago
|
||
Hey Reem + all! I tested this on Windows with NVDA and Nightly 156.0a1 (2026-08-26). Here is a recording from the test. I was able to tab to the privacy widget and hear '88 trackers blocked today across 6 sites link' from NVDA. I encountered the same behavior Eitan mentioned where the number is not read on mouse hover (demonstrated in video).
Let me know if you need anything else!
Updated•2 months ago
|
Comment 9•2 months ago
|
||
Tested on Windows with Narrator and Nightly 157.0a1 (2026-08-31). I enabled 'Read and interact with the screen using the mouse' in Narrator settings and hovered over the trackers # by mouse. Narrator does not read the number, only reads 'trackers blocked today across 9 sites'. Given that I encountered the issue with NVDA and Narrator, I'm thinking this is a platform bug. I'll move this over to Core > Disability Access API
Updated•2 months ago
|
Updated•2 months ago
|
Updated•2 months ago
|
Comment 10•2 months ago
|
||
I don't see a privacy widget on the new tab page in Nightly, neither by default nor on the Manage Widgets page. browser.newtabpage.activity-stream.widgets.privacy.enabled is true in about:config. Is there some other pref I need to set?
| Assignee | ||
Comment 11•2 months ago
|
||
Hi James!
here are the following prefs you'll need to enable to view the widget on New Tab:
browser.newtabpage.activity-stream.widgets.system.enabledbrowser.newtabpage.activity-stream.widgets.system.privacy.enabled
Comment 12•2 months ago
|
||
This is an interesting one. The reason this doesn't work with NVDA is that the visually rendered number - i.e. the node you would get when hit testing the screen coordinates - is hidden from the accessibility tree:
<span class="privacy-count-number" aria-hidden="true">1</span>
For mouse tracking, NVDA will always look for the deepest text node, but in this case, it gets nothing because the text node at those coordinates is aria-hidden.
Of course, the count is exposed to the accessibility tree, but that count is off-screen visually, so it can't be hit tested:
<span class="privacy-count-number-a11y">1</span>
I'm not really sure why there is a separate node (privacy-count-number-a11y) for accessibility purposes, rather than using the real count. That is generally not a good idea, precisely because it breaks hit testing. However, I guess there must have been a good reason for this. Reem, do you know why?
| Assignee | ||
Comment 13•2 months ago
|
||
Hi Jamie, sorry this slipped from my radar! I'm catching up with it now.
I wasn't the one who implemented these 2 different span elements, however my conjecture from taking a deeper look, I think the second span tag (<span class="privacy-count-number-a11y">1</span>) was placed there because the first one (<span class="privacy-count-number" aria-hidden="true">1</span>) carried a number that would be updating in real-time/refresh if the number of trackers being blocked was increasing.
But, I don't really think it's necessary to have 2 separate tags to display the count. What we can do instead is keep the first one, remove the aria-hidden="true" attribute, and replace it with aria-live="polite" or aria-live="assertive" to indicate that if the number is updating.
That would probably clean up a bit of code, lessen future confusion, and screenreaders would be able to read what is neccessary.
What do you think?
Comment 14•2 months ago
|
||
(In reply to Reem Hamoui from comment #13)
I wasn't the one who implemented these 2 different span elements,
Right, and unfortunately, git blame isn't useful here. It seems this was landed as part of the initial code with no explanation as to why. I wonder if it's worth asking dgrant?
I think the second
spantag (<span class="privacy-count-number-a11y">1</span>) was placed there because the first one (<span class="privacy-count-number" aria-hidden="true">1</span>) carried a number that would be updating in real-time/refresh if the number of trackers being blocked was increasing.
So the second "a11y" one doesn't get updated when the count increases? If I understand correctly, that means it might well contain stale information, which seems bad.
But, I don't really think it's necessary to have 2 separate tags to display the count. What we can do instead is keep the first one, remove the
aria-hidden="true"attribute,
Sounds good to me, unless there's something else I'm missing.
and replace it with
aria-live="polite"oraria-live="assertive"to indicate that if the number is updating.
I wouldn't add aria-live here. That would cause the number to be reported each time it updates, which will potentially result in excess verbosity that i don't think is useful to the user. Note that a screen reader user can still query for the information and they will receive the latest count. aria-live just determines whether the number gets reported automatically as soon as it changes.
That would probably clean up a bit of code, lessen future confusion, and screenreaders would be able to read what is neccessary.
I agree.
Comment 15•2 months ago
|
||
Moving this back to HNT given that the fix here belongs in HNT code.
Comment 16•1 month ago
|
||
The severity field is not set for this bug.
:thecount, could you have a look please?
For more information, please visit BugBot documentation.
| Assignee | ||
Comment 17•1 month ago
|
||
Updated•1 month ago
|
Comment 18•1 month ago
|
||
Comment 19•1 month ago
|
||
| bugherder | ||
Comment 20•1 month ago
|
||
Since nightly and release are affected, beta will likely be affected too.
For more information, please visit BugBot documentation.
Comment 21•1 month ago
|
||
The patch landed in nightly and beta is affected.
:rhamoui, please make an uplift decision for beta:
- Nominate the patch for beta uplift approval if the fix should be included in this release, or
- Set
status-firefox157towontfixif the fix can wait for the next release.
See Requesting an Uplift for documentation on how to request an uplift.
For more information, please visit BugBot documentation.
| Reporter | ||
Comment 23•1 month ago
|
||
Verified that the blocked trackers number is read as expected with NVDA on WIndows 11 with Firefox 159.0a1 (2026-09-25).
| Reporter | ||
Comment 24•1 month ago
|
||
Verified that this is fixed as well on Firefox 158.0b2 on the same machine as in Comment 23.
Description
•