[FOLLOW-UP] HIVE-28373: Iceberg: Refactor the code of HadoopTableOptions

[BsoBird]

What changes were proposed in this pull request?

Modify the test case to ensure it can run on macOS environments.

Why are the changes needed?

Since a considerable number of developers use macOS as their local development environment, the test case code should avoid relying on Linux-specific commands.

Does this PR introduce any user-facing change?

No

[deniskuzZ]

please drop the unnecessary casts

[BsoBird]

got it

[deniskuzZ]

why such a massive copy-paste?

try {
        HadoopTableOperations tableOperations = (HadoopTableOperations) baseTable.operations();
        HadoopTableOperations spyOps = spy(tableOperations);
        doNothing().when(spyOps).tryLock(any(), any());
        doAnswer(x -> {
          countDownLatch.countDown();
          countDownLatch.await();
          return x.callRealMethod();
        }).when(spyOps).renameMetaDataFileAndCheck(any(), any(), any(), anyBoolean());
        doAnswer(x -> {
          Path srcPath = x.getArgument(1);
          Path dstPath = x.getArgument(2);
          Files.move(
              Paths.get(srcPath.toUri()), Paths.get(dstPath.toUri()),
              StandardCopyOption.ATOMIC_MOVE
          );
          return true;
        }).when(spyOps).renameMetaDataFile(any(), any(), any());
        TableMetadata metadataV1 = spyOps.current();
        SortOrder dataSort = SortOrder.builderFor(baseTable.schema()).asc("data").build();
        TableMetadata metadataV2 = metadataV1.replaceSortOrder(dataSort);
        spyOps.commit(metadataV1, metadataV2);
      } catch (CommitFailedException e) {
        expectedException.set(e);
      } catch (Throwable e) {
        unexpectedException.set(e);
      }

[deniskuzZ]

why 8 threads? Executors.newFixedThreadPool(8)
use virtual threads instead

[deniskuzZ]

    Runnable commitTask = () -> {
      try {
        HadoopTableOperations tableOperations = (HadoopTableOperations) baseTable.operations();
        HadoopTableOperations spyOps = spy(tableOperations);
        doNothing().when(spyOps).tryLock(any(), any());
        doAnswer(x -> {
          countDownLatch.countDown();
          countDownLatch.await();
          return x.callRealMethod();
        }).when(spyOps).renameMetaDataFileAndCheck(any(), any(), any(), anyBoolean());
        doAnswer(x -> {
          Path srcPath = x.getArgument(1);
          Path dstPath = x.getArgument(2);

          var src = Paths.get(srcPath.toUri());
          var dst = Paths.get(dstPath.toUri());
          Files.move(src, dst, StandardCopyOption.ATOMIC_MOVE);

          return Files.exists(dst) && Files.notExists(src);
        }).when(spyOps).renameMetaDataFile(any(), any(), any());

        TableMetadata metadataV1 = spyOps.current();
        SortOrder dataSort = SortOrder.builderFor(baseTable.schema()).asc("data").build();
        TableMetadata metadataV2 = metadataV1.replaceSortOrder(dataSort);
        spyOps.commit(metadataV1, metadataV2);
      } catch (CommitFailedException e) {
        expectedException.set(e);
      } catch (Throwable e) {
        unexpectedException.set(e);
      }
    };

    Tasks.range(2)
        .executeWith(executorService)
        .run(i -> commitTask.run());

[BsoBird]

got it

[deniskuzZ]

@BsoBird this test fails most of the time for me.

failed on jenkins as well: https://ci.hive.apache.org/job/hive-precommit/job/PR-6345/1/testReport/junit/org.apache.iceberg.hadoop/TestHiveHadoopCommits/Testing___split_20___PostProcess___testTwoClientCommitSameVersion/

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[BsoBird]

@deniskuzZ
Hello, I've updated the test code for cross-platform stability.
The root cause was that macOS lacks the renameat2 syscall available in Linux, forcing rename to follow basic POSIX behavior (overwriting targets). To eliminate these filesystem behavioral differences between test environments, I've moved the rename logic into Java code to simulate HDFS rename semantics.