zaynt4606 commented on PR #10698: URL: https://github.com/apache/paimon/pull/10698#issuecomment-6306141779
Fixed the PR-specific Native CI failure in db2e8076bf. The native planner can return no splits for a zero scan limit, so the new datasource test must not index the first read task unconditionally. The test now checks the metadata of every returned task and asserts the total row count, including an empty plan. No runtime code was changed and no Native tests were skipped. Validation after this update: - 69 passed on official Ray 2.58.0: datasource, integration, sink, and RayData tests. - 14 passed with the split provider patched to reproduce Native's empty-plan behavior for LIMIT 0 (before this test fix: 1 failed, 13 passed). - Repository-configured Flake8, Python license-header checks, and git diff --check passed. The real Rust backend was not run locally; the new Native CI run will validate that backend. The other failures also occur on the unchanged upstream master commit eeb765d405963b619cc0194ecf79e2378df3b77e: - `BlobConsumerTest::test_blob_consumer_after_write_raises`: Native raises ValueError where the test expects RuntimeError. - `test_ltz_partition_strings_preserve_submicrosecond_fraction`: timestamp fraction differs from the expected nanosecond value. Both are present in the [upstream master Native CI log](https://github.com/apache/paimon/actions/runs/39399763121/job/118912025870); the timestamp failure is also present in the [upstream master Python 3.10 CI log](https://github.com/apache/paimon/actions/runs/39399763121/job/118912026269). I have kept these unrelated baseline failures out of this PR. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
