On Sun, 15 Nov 2026 05:10:43 GMT, Sendao Yan <[email protected]> wrote:
>> On macOS 27 window animation takes a bit longer and if we attempt to perform >> the mouse event on the window at this moment the coordinates conversion can >> be incorrect. Uncommenting the one second delay before mouse click fixing >> the problem. >> >> >> >> >> --------- >> - [x] I confirm that I make this contribution in accordance with the >> [OpenJDK Interim AI Policy](https://openjdk.org/legal/ai). > > test/jdk/java/awt/event/MouseEvent/FrameMouseEventAbsoluteCoordsTest/FrameMouseEventAbsoluteCoordsTest.java > line 79: > >> 77: robot.mouseMove(mousePositionAbsolute.x, >> 78: mousePositionAbsolute.y ); >> 79: robot.delay(1000); > > **[suggestion] Consider moving the delay ahead of the coordinate computation** > > As written, the delay sits between `mouseMove()` and `mousePress()`, so > `button.getLocationOnScreen()` and the `mouseMove()` itself still execute > while the window may be in the middle of the show animation — only the press > is deferred. That is consistent with the failure in the bug (the absolute > screen-coordinate check passed; only the component-relative conversion was > off), so this does fix the reported problem. > > Still, moving the delay to right after `setAutoWaitForIdle(true)` would be > strictly more robust and easier to reason about: both the geometry read and > the mouse interaction would then happen only after the animation has settled, > so the fix no longer depends on the window's screen position being stable > while it animates. The cost is identical. > > > robot = new Robot(); > robot.setAutoWaitForIdle(true); > robot.delay(1000); // let the macOS window show animation settle > mousePositionAbsolute = new Point(button.getLocationOnScreen().x + > button.getWidth()/2, > button.getLocationOnScreen().y + > button.getHeight()/2); > mousePosition = new Point(button.getWidth()/2, > button.getHeight()/2); > robot.mouseMove(mousePositionAbsolute.x, > mousePositionAbsolute.y); > robot.mousePress(InputEvent.BUTTON1_MASK); > > > Not blocking — the current placement is fine if you prefer the minimal diff. It is not required because of the way we calculate the absolute position of the component - it is not affected by the transition coordinates changes. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/33910#discussion_r4590720845
