On Fri, 17 Jan 2020 23:40:18 GMT, Kevin Rushforth <k...@openjdk.org> wrote:
>>> >>> >>> @arapte - This bug looks like a good candidate for JavaFX 14. Can you >>> retarget this PR to the jfx14 branch? >> >> Thanks for guiding Kevin, PR is now targeted to jfx14. > > The fix looks good. I'll take a closer look at the unit test later. > > Speaking of tests...since the addition of the `TabPane` reordering logic was > a victim of the already-existing leak in the `viewOrderChildren` list in > `Parent`, it should be possible to write a test case using a Group node and a > few Shape nodes, using setViewOrder directly on the Group node (this would be > in addition to the system test you wrote). Can you take a look at adding one? > It might even be possible to do it as a `javafx.graphics` module unit test > rather than a system test, although you would need to see if the bug > reproduced there (I suspect it will). > > > The fix looks good. I'll take a closer look at the unit test later. > > Speaking of tests...since the addition of the `TabPane` reordering logic was > a victim of the already-existing leak in the `viewOrderChildren` list in > `Parent`, it should be possible to write a test case using a Group node and a > few Shape nodes, using setViewOrder directly on the Group node (this would be > in addition to the system test you wrote). Can you take a look at adding one? > It might even be possible to do it as a `javafx.graphics` module unit test > rather than a system test, although you would need to see if the bug > reproduced there (I suspect it will). Hello Kevin, The bug can be reproduced with system test written using `Group` and `Shape` which is very similar to `TabPaneHeaderLeakTest` test. but it seems the bug is not reproducible with unit test. I tried a unit test very similar to the newly added system test `ShapeViewOrderLeakTest`, but looks like `Parent.viewOrderChildren` list does not get populated and so the issue does not occur. ------------- PR: https://git.openjdk.java.net/jfx/pull/79
