rich7420 opened a new pull request, #6864: URL: https://github.com/apache/datafusion-comet/pull/6864
## Which issue does this PR close? Part of #6634 and #6615. ## Rationale for this change Move the required-child hash regressions to SQL fixtures while preserving the Parquet shape that exposes #5754: a NULL parent struct whose required children still contain values. ## What changes are included in this PR? Use Spark CTAS to preserve required children and assert native `hash`/`xxhash64` execution for flat structs, recursive structs and valid array elements wrapping NULL structs, including two-element hash chaining. Run both dictionary settings and remove the three migrated Scala tests; the other 39 test bodies are unchanged. ## How are these changes tested? Spark 4.1.3 / JDK 21 with matching native/JVM builds: both fixture variants and the remaining hash suite passed, totaling 41 tests. Formatting checks pass. Independent Spark-reference probes verified actual REQUIRED Parquet children on Spark 3.4, 3.5, 4.0, 4.1 and 4.2 under both dictionary settings. The small files use plain encoding, so this does not claim dictionary decoding coverage. In an isolated native negative control with parent-null propagation removed, all four queries failed on wrong results under both settings. Comet runtime on other Spark profiles remains pending; request `run-all-spark-profiles` before queueing. -- 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]
