On Mon, 19 May 2025 08:15:29 GMT, Abhishek Kumar <abhis...@openjdk.org> wrote:
> Test was failing on macOS 14.7.1 system in CI pipeline. Logged output > suggests that the mouse exit event is not triggered and thus the test fails. > Increased number of list items and adjusted the X and Y offset for mouse move > events, so the mouse pointer will be within the frame and added robot delay > to stabilize the test. > > Verified the updated test in CI on macOS 14.7 machine and there is no failure. Changes requested by aivanov (Reviewer). test/jdk/java/awt/List/ListEnterExitTest.java line 1: > 1: /* I strongly suggest making `MouseEnterExitListener` a static nested class inside `ListEnterExitTest`. This prevents any duplicate class names from other classes when one works with tests in an IDE. test/jdk/java/awt/List/ListEnterExitTest.java line 90: > 88: robot.waitForIdle(); > 89: robot.delay(100); > 90: I don't really understand why `waitForIdle` and `delay` are needed here at all. Robot just moves mouse to generate mouse enter and exit events. It can do so without any delays, can't it? test/jdk/java/awt/List/ListEnterExitTest.java line 92: > 90: > 91: robot.mousePress(InputEvent.BUTTON1_MASK); > 92: robot.mouseRelease(InputEvent.BUTTON1_MASK); Is this mouse press needed only to release `mouseEnterExitListener.wait`? I suggest using a `CountDownLatch` to replace `MouseEnterExitListener.passed_1` and `passed_2`. The code would need to time-await for both latches to be released. Then `MouseEnterExitListener.mousePressed` can be removed completely together with the synchronized block above. ------------- PR Review: https://git.openjdk.org/jdk/pull/25299#pullrequestreview-2850541693 PR Review Comment: https://git.openjdk.org/jdk/pull/25299#discussion_r2095578628 PR Review Comment: https://git.openjdk.org/jdk/pull/25299#discussion_r2095570577 PR Review Comment: https://git.openjdk.org/jdk/pull/25299#discussion_r2095576436
