[jdk11u-dev] RFR: 8339728: [Accessibility, Windows, JAWS] Bug in the getKeyChar method of the AccessBridge class

[Frederic Thevenet]

On Mon, 24 Feb 2025 09:51:09 GMT, Antonio Vieiro <duke at openjdk.org> wrote:

> Backport of [JDK-8339728](https://bugs.openjdk.org/browse/JDK-8339728) from 17u that solves some accessibility issues on Windows, for parity with 11.0.27/28-oracle  (and because it's marked as CPU25-critical-SQE-OK too). Low risk.
> 
> The test does fail without this fix (when read-out, the short-cuts are truncated, so for instance the menu short-cut "Ctrl+Shift+Period" was read out as "Control plus Shift plus P", instead of "Control plus Shift plus Period".
> 
> The new accessibility test passes on Windows 10 (tested with [JAWS v. 2025.2412.102 - February 2025](https://support.freedomscientific.com/Downloads/JAWS) ).
> 
> Clean but for an update in copyright header and for a multi-line string (that has been replaced) in `AccessBridge.java`.

Okay so today I learned that the accessibility bridge isn't enabled by default, and that you have to turn it on explicitly using `jabswitch -enable`[1]
With this turned on, I was finally able to test this change and I can confirm that  the screen reader software would read out the shortcuts correctly with the change, while it would only reads the first letter of the last key when running it in 11.0.26.
As a side note, I ran this test with NVDA instead of JAWS, so it is good to see this works with a larger panel of accessibility solutions.

I'm tempted to say that adding a short mention to turning on the accessibility bridge in the comments that describe how to run the test would go a long way to saving some time to people who, like me, aren't familiar with screen readers, but I'm not sure that this is something that belong into a backport. WDYT @jerboaa ?

[1] https://docs.oracle.com/en/java/javase/11/access/enabling-and-testing-java-access-bridge.html

-------------

PR Comment: https://git.openjdk.org/jdk11u-dev/pull/3001#issuecomment-2724377076