Migrate Chronicle Map tests to JUnit 5 - #588
Merged
Merged
Conversation
- root cause: test suites still depended on JUnit 4 conventions and runners. - fix: migrate tests and supporting module code to JUnit 5. - impact: the module now follows the JUnit 5 branch baseline.
|
# Conflicts: # src/test/java/net/openhft/chronicle/map/AbstractMarshallableKeyValueTest.java # src/test/java/net/openhft/chronicle/map/AutoResizeTest.java # src/test/java/net/openhft/chronicle/map/BooleanValuesTest.java # src/test/java/net/openhft/chronicle/map/BuildVersionTest.java # src/test/java/net/openhft/chronicle/map/BytesMarshallableValueTest.java # src/test/java/net/openhft/chronicle/map/CHMUseCasesTest.java # src/test/java/net/openhft/chronicle/map/ChronicleMapEqualsTest.java # src/test/java/net/openhft/chronicle/map/ChronicleMapImportExportTest.java # src/test/java/net/openhft/chronicle/map/ChronicleMapSanityCheckTest.java # src/test/java/net/openhft/chronicle/map/ChronicleMapTest.java # src/test/java/net/openhft/chronicle/map/ConstantSizeBySampleTest.java # src/test/java/net/openhft/chronicle/map/DefaultValueTest.java # src/test/java/net/openhft/chronicle/map/DemoChronicleMapTest.java # src/test/java/net/openhft/chronicle/map/ExitHookTest.java # src/test/java/net/openhft/chronicle/map/IntValueMapTest.java # src/test/java/net/openhft/chronicle/map/Issue354bTest.java # src/test/java/net/openhft/chronicle/map/Issue42Test.java # src/test/java/net/openhft/chronicle/map/Issue43Test.java # src/test/java/net/openhft/chronicle/map/Issue60Test.java # src/test/java/net/openhft/chronicle/map/LargeEntriesTest.java # src/test/java/net/openhft/chronicle/map/ListenersTest.java # src/test/java/net/openhft/chronicle/map/MapCloseTest.java # src/test/java/net/openhft/chronicle/map/MarkTest.java # src/test/java/net/openhft/chronicle/map/MarshallableReaderWriterTest.java # src/test/java/net/openhft/chronicle/map/MemoryLeaksTest.java # src/test/java/net/openhft/chronicle/map/NestedContextsTest.java # src/test/java/net/openhft/chronicle/map/RecursiveRefereneChMapTest.java # src/test/java/net/openhft/chronicle/map/ReplicatedChronicleMapTest.java # src/test/java/net/openhft/chronicle/map/TrickyContextCasesTest.java # src/test/java/net/openhft/chronicle/map/ValueAlignmentRelocationTest.java # src/test/java/net/openhft/chronicle/map/fromdocs/OpenJDKAndHashMapExamplesTest.java # src/test/java/net/openhft/chronicle/map/fromdocs/acid/revelations/DirtyReadVictimTest.java # src/test/java/net/openhft/chronicle/map/issue/ParallelStartupTest.java # src/test/java/net/openhft/chronicle/map/issue/PutIfAbsentNoGarbageTest.java # src/test/java/net/openhft/chronicle/map/jsr166/JSR166TestCase.java # src/test/java/net/openhft/chronicle/map/jsr166/map/ChronicleMapTest.java # src/test/java/net/openhft/chronicle/map/locks/DirtyReadOffenderIPCTest.java # src/test/java/net/openhft/chronicle/map/locks/DirtyReadOffenderTest.java # src/test/java/net/openhft/chronicle/map/locks/DirtyReadOffender_ReaderWriterTest.java # src/test/java/net/openhft/chronicle/map/locks/DirtyReadOffender_WriterReaderTest.java # src/test/java/net/openhft/chronicle/map/locks/DirtyReadOffender_WriterWriterTest.java # src/test/java/net/openhft/chronicle/map/locks/DirtyReadVictimIPCTest.java # src/test/java/net/openhft/chronicle/map/locks/DirtyReadVictimTest.java # src/test/java/net/openhft/chronicle/map/locks/ReaderToo.java # src/test/java/net/openhft/chronicle/map/locks/WriterToo.java # src/test/java/net/openhft/chronicle/set/Issue3Test.java
|
tgd
removed their request for review
August 26, 2026 18:01
This was referenced Sep 8, 2026
peter-lawrey
added a commit
that referenced
this pull request
Sep 8, 2026
Use Buffer-typed cursor calls so modern source/target-8 compilation does not link Map header I/O against Java 9+ ByteBuffer methods. Include the global-state flip required by reopening, exposed by exercising the packaged artifact on Java 8. Extract the compatibility changes identified in #588 and add create/reopen/recover/reopen coverage. The Java-21-built baseline fails on Java 8; the corrected packaged JAR passes, as does full Java 21 verification.
peter-lawrey
added a commit
that referenced
this pull request
Sep 8, 2026
* Require successful initial opening in the recovery test Extract the scoped recovery assertion identified in #588. Initial opening and cleanup failures must fail independently; close any Map unexpectedly returned by recovery outside assertThrows. * Simplify cleanup of unexpected recovery results Use Closeable.closeQuietly for the nullable recovery result, keeping cleanup outside the scoped assertion. Addresses review feedback on PR #614.
peter-lawrey
added a commit
that referenced
this pull request
Sep 8, 2026
* Record JUnit assertion failures from JSR166 worker helpers Catch the AssertionError thrown by JUnit 4, record it for teardown, and preserve the original worker throwable. Extract the helper correction identified in #588 without translating the shared fixture or its consumers. An owning-JUnit regression fails six of eight cases before the fix; all eight pass afterwards. No active pre-existing consumer of the affected worker helpers was found. * Potential fix for pull request finding Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
Preserve the existing branch history and use the reviewed conflict resolution from 4f3d57351c2498024eb664335552ee1f0a0b1fa5. The resulting tree is identical to that validated candidate.
|
peter-lawrey-admin
approved these changes
Sep 17, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Migrate Chronicle Map tests and shared fixtures to Jupiter while retaining execution of the remaining JUnit 4 tests. Conditional removal has a real map-owning regression; nested-context and worker-write assertions target the intended operations and cannot pass because setup failed or a future timed out.
Current source:
59e6a00fdeaa890230c0cac01ac65ee5e0ce8567. It contains developec649062438ef6f706a493945608d44dbb3df3dcand the updated #617 prerequisite at505a7b507b6e8744024705c66e127f1ac324cf64. Land that independently useful repair separately, then inspect the migration's remaining diff. This branch as a whole still contains the shared production repair.The production audit found that only
VanillaChronicleHashchanges behaviour relative to develop. Changes inInternalMapFileAnalyzer,MapFileAnalyzerandChronicleMapBuilderare formatting. The shared close repair now has//!blocks explaining every behavioural region and referencing seven regressions. This review changes no production tokens beyond comments and whitespace.The migration retains:
testRemove2, with the owning map closed.ExecutionExceptioncaused specifically byInterProcessDeadLockExceptionfor the worker write. Context release precedes bounded worker termination and map close; secondary cleanup failures remain attached to the original assertion.The review added three Jupiter
CloseContextSnapshotTestcases inherited from #617's JUnit 4 regressions. They cover absent resources, released/collected context slots and cleared holders. Each fails when its corresponding guard is removed. The four persisted query/iteration regressions fail with develop's old production file and pass with the repair.It also fixed an existing fixture leak in
MapCloseTest.vanillaChronicleHashAllContextsExpungeTest: the final query and owning map now close in order. A probe against the original fixture found one open map and one active context after a passing invocation; the corrected fixture leaves neither. The final validation runs required no forced JVM shutdown.Fresh validation on Linux x86_64, Maven 3.9.11, isolated dependency cache, zero retries:
clean verifyThe full-run total increases from the previous 1,330 by the three new snapshot regressions. Jupiter executes the new tests and Vintage executes the eight legacy worker-failure variants. The September 12 scenario-level qualification remains a historical comparison at its stated revisions. This review did not rebuild that complete discovery/parameter manifest, so the fresh total alone is not a new parity proof.
Reproduction, using a populated isolated Maven cache:
Authenticated TeamCity review at 15:03 UTC on 17 September: eight builds on this exact head have finished successfully (Linux ARM, Java 8/11/17/21/25 and Zing 8/11); macOS build 1363098 remains queued. Linux Java 8 build 1363094 completed the full suite: 1,333 cases, 82 skips, zero failures/errors.
Windows qualification failed. Build 1363101 finished with
BUILD FAILUREand process exit 1, reporting three failures in fiveParallelStartupTestrepetitions. Worker map creation failed withsizeWord is not ready: 0, an impossible header size (446921713versus a65536-byte file), and a header-checksum mismatch; the owning assertion observed 15, 15 and 1 successful workers instead of 16. TeamCity groups these under one test identity. Surefire lists all five runs (three failures, two passes) under that method; its final aggregate is 1,329 results, one failure and 75 skips. The four persisted-close and three snapshot regressions passed. The startup algorithm and five-repetition annotation are unchanged from #617, whose Windows run passed; causation by the migration has not been established. A controlled Windows comparison with matching dependencies is needed before changing startup behaviour or weakening this test.The earlier Windows/macOS failure statuses correspond to cancelled builds. The old Java 26/27 failures are now diagnosed:
JSR166TestCase.java:273callsThread.stop(), sotestCompilefails withcannot find symbol: method stop(). The same call remains in this head and develop. JDK 26 removed that API; this is an outstanding test-fixture compatibility issue. Compiler diagnostics were recovered from retained TeamCity artifacts after normal old-log download endpoints returned 404. No new Java 26/27 execution is established for this head.The existing relocation boundary remains: a new entry can be published before write-lock acquisition, so lock rejection does not establish transactional rollback. The separate other-thread close-timeout path also remains outside #617's same-thread rejection repair. No merge approval is recorded by this update.