RFR: 8306464: [testbug] CustomSecurityManagerTest full screen tests fails on MacOS intermittently

Kevin Rushforth kcr at openjdk.org
Thu Apr 20 12:29:58 UTC 2023


On Thu, 20 Apr 2023 08:36:59 GMT, Karthik P K <kpk at openjdk.org> wrote:

> Usage of `getBounds()` method instead `getVisualBounds()` was giving unreliable screen bound values and color value was read very close to the edge of the window.
> 
> Updated the code to use `getVisualBounds()` instead of `getBounds()` and moved the coordinates inside the window from where the color value is read.
> 
> Ran the tests individually and along with all system tests in following systems. No failure found after the fix.
> Mac M1 with Ventura 13.3
> Window 11

This looks good (I still want to run a quick test), with a couple additional suggestions.

tests/system/src/test/java/test/robot/helloworld/CustomSecurityManagerTest.java line 202:

> 200:                 if (row == 0) {
> 201:                     // avoid the top area as it might contain OS-specific UI (Macs with a notch)
> 202:                     y = h / 3;

Now that this tests uses visualBounds, there is no need to worry about avoiding the top, so this can be something like "y = 4".

tests/system/src/test/java/test/robot/helloworld/CustomSecurityManagerTest.java line 208:

> 206: 
> 207:                 for (int col = 0; col < 2; col++) {
> 208:                     int x = col == 0 ? 1 : screenWidth.get() - 5;

For `col == 0` you might want to change the value from `1` to `4` for the same reason you changed the second value.

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

PR Review: https://git.openjdk.org/jfx/pull/1103#pullrequestreview-1393865672
PR Review Comment: https://git.openjdk.org/jfx/pull/1103#discussion_r1172506991
PR Review Comment: https://git.openjdk.org/jfx/pull/1103#discussion_r1172510742


More information about the openjfx-dev mailing list