[JENKINS-55081] SEC-904 related test failing on java 11

[Wadeck]

• just running the tests for the moment

See JENKINS-55081.

Proposed changelog entries

• Internal: correct test failing with Java 11: FilePathSEC904Test.isDescendant_regularFiles

Submitter checklist

• JIRA issue is well described
• Changelog entry appropriate for the audience affected by the change (users or developer, depending on the change). Examples
  * Use the Internal: prefix if the change has no user-visible impact (API, test frameworks, etc.)
• Appropriate autotests or explanation to why this change has no tests
• [n/a] For dependency updates: links to external changelogs and, if possible, full diffs

[oleg-nenashev]

LGTM as a test fix, but probably we need a user-facing changelog entry

[batmat]

I have checked the commits, and bear with me, I might have missed something. So IIUC, both tests and main code need to be changed because of the upstream relativize change in behavior.

To assess things stay fixed in Java 8, I think we should have the following:

1. first commit only changes src/main/java code, not touching src/test/java yet
   • => we expect a PR build success on Java 8 (still ignore for now the failures with JDK11, given that's we're fixing)
2. Once this shows we are still green on !Java-11, we also change the test so we remove failures from JDK11.

This way we do know that the code on Java 8 was still checked to work with the original tests it was written with.

WDYT?

[Wadeck]

@batmat I did not change the failing tests. I just added new ones.

[batmat]

Right indeed, I don't know why I thought I had seen replacements.

[batmat]

Ping @jenkinsci/java11-support BTW given this does relate to an issue only surfacing with Java 11.

[batmat]

I think we have enough approvals to merge it as soon as, or not long after, the PR succeeds. I think we should merge this early so that other PRs stop being red for unrelated failures (and we easily see which one keep failing for actual reasons, that is)

[Wadeck]

Tests failing due to:

• hudson.model.QueueTest.testBlockBuildWhenUpstreamBuildingLock <= flaky / time-dependent test
  • new SleepBuilder(5000), [...] new SleepBuilder(10000)
  • NOT displayed in BO Tests tab
• hudson.util.AtomicFileWriterPerfTest.poorManPerformanceTestBed <= Baptiste is already on it

[batmat]

I'm seeing only the latter failing:

Which indeed I am still analyzing in #3788

Let's merge this.

[batmat]

Filed as #3797