felipepessoto opened a new pull request, #12855: URL: https://github.com/apache/gluten/pull/12855
## What changes are proposed in this pull request? Automate maintenance of the Delta Spark UT `known-failures.txt` baseline from the scheduled daily run. We recently had to remove now-passing tests manually in two separate PRs: - [#12815](https://github.com/apache/gluten/pull/12815) removed four tests fixed by cast-mode preservation. - [#12851](https://github.com/apache/gluten/pull/12851) removed two type-widening tests fixed by a Velox update. Those fixes reached core libraries without triggering Delta CI when merged. The daily run eventually detected the now-passing tests, but updating the baseline still required a manual PR. This change makes the scheduled workflow open or update a `[MINOR][CI]` PR containing those removals. The generated update is deliberately deletion-only: it removes tests confirmed as passing, never adds new failures, and preserves skipped, unseen, and flaky entries as well as existing comments. Regressions continue to fail the Delta gate and cannot be accepted automatically. A focused unit test covers the pruning behavior, including preservation of regressions, skipped tests, stale entries, flaky tests, and baseline comments. Related to #12743. ## How was this patch tested? - Ran the `compare-test-results.py` unit test. - Checked both Python files with Black. - Validated `delta_spark_ut.yml` with actionlint. - Validated the embedded PR-update shell script. - Ran `git diff --check`. ## Was this patch authored or co-authored using generative AI tooling? Generated-by: GitHub Copilot CLI 1.0.80 -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
