RFR: 8276058: Some swing test fails on specific CI macos system [v2]

[Phil Race]

On Thu, 28 Oct 2021 20:54:54 GMT, Sergey Bylokhov <serb at openjdk.org> wrote:

> > I have attached 2 screen capture in JBS taken when PressedIconTest and JInternalFrame test failed and the screenshot at least in this machine shows presence of cursor.
> 
> So maybe something was changed in the native API? It looks like a tck issue that breaks the specified behavior. Is it happen only on this system? Could it be related to the cursor configuration in the settings, size/visibility/scale etc?

See the comment I just added to https://bugs.openjdk.java.net/browse/JDK-8252813.

It still needs verification to see if it affects the macOS API used. But it seems probable something like that is going on.
Regardless I think we may need to revise the Robot spec a little ... I don't know why it is so insistent that the cursor is NOT captured since it seems something outside control of the Robot ..

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

PR: https://git.openjdk.java.net/jdk/pull/6140