RFR: 8254841: [macos] Enter or Exit events are shouldn't be displayed while resizing [v2]
Alisen Chung
github.com+90066231+alisenchung at openjdk.java.net
Wed Sep 15 08:18:55 UTC 2021
On Wed, 15 Sep 2021 04:18:11 GMT, Sergey Bylokhov <serb at openjdk.org> wrote:
> I have a general question about this bug, initially the test was added for https://bugs.openjdk.java.net/browse/JDK-6479820 and that bug was fixed because of this:
>
> ```
> When resizing the window by dragging a corner with the mouse,
> the panel receives entry/exit mouse events randomly.
> These events should not occur because the mouse is not entering
> or exiting the window, it is dragging a corner of its border.
> ```
>
> And in the current bug description:
>
> `The window resizing is a bit slower than mouse movement. So before the window resize can catch up to the mouse, the mouse exits the window, then re-enters when the window catches up to it. This can be verified if you resize the window at the edge without entering the window: there will be no mouseEntered or mouseExit events. `
>
> So situation is different, if the mouse is moved faster than window is moving, then I do not see why we should not post an events. Especially if the native system did that.
In the original bug report, I believe the root cause of the bug is the same based on the comment in JBS:
`So far this happens when system is not fast enough to resize native component in a short period of time while message is on its way.`
And I tried to keep the same idea for the fix in windows:
`The idea of the fix would be to pick up AwtWindow.m_resizing value and not consider any messages in PreProcessMouseMsg until m_resizing is true.`
So I guess the question is whether or not we should be posting these events in the first place during resizing? I still see the original bug fix in the windows files so I believe there currently is a mismatch in behavior across platforms.
-------------
PR: https://git.openjdk.java.net/jdk/pull/5497
More information about the client-libs-dev
mailing list