On Thu, 22 May 2025 14:55:21 GMT, Mikhail Yankelevich <myankelev...@openjdk.org> wrote:
>> Barring other effects, it is likely that this test just needs to have some >> longer timeouts to ensure that a stall isn't likely to fail the test case. >> The shouldn't have any impact on test execution duration, as the test looks >> at tasks which *have not executed yet*. > > test/jdk/java/util/concurrent/tck/ScheduledExecutorTest.java line 703: > >> 701: final ScheduledThreadPoolExecutor p = new >> ScheduledThreadPoolExecutor(1); >> 702: List<ScheduledFuture<?>> tasks = new ArrayList<>(); >> 703: final int DELAY = 100; > > What do you think of trying to change this to `Utils.adjustTimeout(100)` so > the env could change this depending on the factor used? This way may be the > delay won't need to be as large. @myankelev The tasks are not expected to run, so the delay does not impact the duration of the test execution. The problem is rather that if they *do run* that the test fails, so if we move the delay further into the future, the test is less prone to failing spuriously. ------------- PR Review Comment: https://git.openjdk.org/jdk/pull/25394#discussion_r2102782380
