On Thu, 15 May 2025 13:51:07 GMT, Anass Baya <ab...@openjdk.org> wrote:
>> **Analysis :**
>>
>> Whether the test passes on the main line or fails, the behavior is still
>> incorrect.
>> This test is meant to ensure that pressing ESC a second time while the file
>> dialog is open behaves correctly.
>>
>> However, the CountDownLatch is currently set to 1, which means the test only
>> waits for the first open/close interaction to complete. As a result, it does
>> not wait for the second attempt (opening the dialog again and pressing ESC
>> to close it), because the latch reaches zero after the first attempt.
>>
>> This causes the test to proceed immediately to the validation step:
>>
>> if (fd.isVisible()) {
>> throw new RuntimeException("File Dialog is not closed");
>> }
>>
>> At this point, whether the test passes or fails becomes unreliable and
>> undefined, as it depends on the state of the second attempt (whether the
>> file dialog is in the process of opening, being closed, or hasn't even
>> started yet)
>>
>> To ensure the test behaves correctly, the CountDownLatch should be set to 2,
>> so it waits for the two attempt of open-close interactions to be completed
>> before moving on to validation.
>>
>> **Proposed fix:**
>>
>> set the CountDownLatch to 2
>>
>> **Proposed enhancements :**
>>
>> Remove unnecessary threads (Thread and Thread-2)
>> Properly handle delays and robot.waitForIdle()
>> Avoid indefinite blocking on latch.await()
>>
>> With these enhancements, the test execution time is reduced from around 3
>> minutes to approximately 1 minute 30 seconds
>>
>> The adapted test uncovered a new bug in GTKFileDialog on Linux, which is
>> being tracked under
>> [JDK-8356981](https://bugs.openjdk.org/browse/JDK-8356981). As a result, it
>> has been added to the ProblemList. See
>> [JDK-8356981](https://bugs.openjdk.org/browse/JDK-8356981) for more details
>
> Anass Baya has updated the pull request incrementally with one additional
> commit since the last revision:
>
> Update test/jdk/ProblemList.txt
>
> Co-authored-by: Abhishek Kumar <abhishek.cx.ku...@oracle.com>
A sanity check: does the updated test reproduce the problem reported in
[JDK-5097243](https://bugs.openjdk.org/browse/JDK-5097243)? I'm pretty sure it
should.
test/jdk/java/awt/FileDialog/DoubleActionESC.java line 56:
> 54: private static volatile CountDownLatch latch;
> 55: private static final int REPEAT_COUNT = 2;
> 56: private static final long LATCH_TIMEOUT = 4;
Suggestion:
private static final int REPEAT_COUNT = 2;
private static final long LATCH_TIMEOUT = 4;
private static final CountDownLatch latch = new
CountDownLatch(REPEAT_COUNT);
test/jdk/java/awt/FileDialog/DoubleActionESC.java line 81:
> 79: robot.mouseRelease(InputEvent.BUTTON1_DOWN_MASK);
> 80: robot.waitForIdle();
> 81: robot.delay(1000);
Can the delay be reduced to `500`?
test/jdk/java/awt/FileDialog/DoubleActionESC.java line 89:
> 87: }
> 88:
> 89: latch.await(LATCH_TIMEOUT, SECONDS);
Throw an exception if the timeout is reached — the test already failed.
test/jdk/java/awt/FileDialog/DoubleActionESC.java line 90:
> 88:
> 89: latch.await(LATCH_TIMEOUT, SECONDS);
> 90: if (fd.isVisible()) {
I'm on the fence here… Calling `fd.isVisible()` on EDT is more reliable; but
it's not strictly required for AWT components.
test/jdk/java/awt/FileDialog/DoubleActionESC.java line 92:
> 90: if (fd.isVisible()) {
> 91: throw new RuntimeException("File Dialog is not closed");
> 92: }
Dispose of `fd` too (in the finally block).
-------------
Changes requested by aivanov (Reviewer).
PR Review: https://git.openjdk.org/jdk/pull/25184#pullrequestreview-2844332500
PR Review Comment: https://git.openjdk.org/jdk/pull/25184#discussion_r2091558746
PR Review Comment: https://git.openjdk.org/jdk/pull/25184#discussion_r2091564010
PR Review Comment: https://git.openjdk.org/jdk/pull/25184#discussion_r2091576832
PR Review Comment: https://git.openjdk.org/jdk/pull/25184#discussion_r2091578738
PR Review Comment: https://git.openjdk.org/jdk/pull/25184#discussion_r2091575682
