Closed Bug 1625254 Opened 4 years ago Closed 4 years ago

NVDA screen reader doesn’t read the intervention cards “Get Help” button on mouseover

Categories

(Firefox :: Address Bar, defect, P1)

All
Windows
defect
Points:
2

Tracking

()

RESOLVED WONTFIX
Iteration:
76.2 - Mar 23 - Apr 5
Tracking Status
firefox-esr68 --- unaffected
firefox74 --- disabled
firefox75 --- wontfix
firefox76 --- wontfix

People

(Reporter: emilghitta, Assigned: adw)

References

(Blocks 1 open bug)

Details

Affected versions

  • Firefox 76.0a1 (BuildId:20200326093308)
  • Firefox 75.0b9 (BuildId:20200325213426)

Unaffected versions

  • Firefox 68.6.0esr (BuildId:20200305175243).
  • Firefox 74.0 (disabled).

Affected platforms

  • Windows 10 64bit

Unaffected platforms

  • macOS 10.14
  • Ubuntu 18.04 64bit

Preconditions

  • Install the latest version of NVDA screen reader.

Steps to reproduce

  1. Launch Firefox.
  2. Trigger an intervention card by typing update firefox inside the Address Bar.
  3. Mouseover the “Get Help” button.

Expected result

  • NVDA successfully reads the “Get Help” button.

Actual result

  • NVDA doesn't read the “Get Help” button.
Points: --- → 2
Priority: -- → P1
Assignee: nobody → adw
Status: NEW → ASSIGNED
Iteration: --- → 76.2 - Mar 23 - Apr 5

Jamie, sorry to bother you, this seemed like a simple fix, but it looks like a <span> needs text content in order for NVDA to read something when you hover over it, is that right? To refresh your memory, some tips in the urlbar have a help button with no text content and a question mark image. They're implemented using <span role="button" title="Get help"> without content, but NVDA doesn't read anything when you hover over them. I tried setting aria-label too, including just hardcoding a value for it, and that also doesn't work. When I add text content to the <span>, that is read.

Tips also have icons, which are implemented as <img alt="Some text">, and the alt is read on hover. I tried using alt for the help button <span>, but that doesn't work either.

So this seems like a constraint in how a11y works.

Flags: needinfo?(jteh)

The problem is that if there is a text interface, NVDA's mouse tracking will use that. Most elements (span, button, etc.) get a text interface. The reason img alt works is that img elements do not get a text interface, since it never makes sense for those.

I'd say this is more a bug in NVDA. It should probably ignore the text interface and use the name for cases where the name (label, title, etc.) clearly overrides the content. It already does this for browse mode, but not for mouse tracking.

Flags: needinfo?(jteh)

Thanks Jamie. From our POV for Firefox, do you think the current behavior is acceptable for our users in the meantime? Or should we (on the urlbar team) try to find a work-around? fwiw when the help button is focused with the keyboard, NVDA does read "Get help."

Flags: needinfo?(jteh)

I think it's acceptable because the majority of NVDA users are keyboard users. Also, this bug applies to mouse tracking elsewhere in Firefox as well. I think we can close this.

Flags: needinfo?(jteh)
Status: ASSIGNED → RESOLVED
Closed: 4 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.