andygrove commented on PR #4798: URL: https://github.com/apache/datafusion-comet/pull/4798#issuecomment-6151381130
@viirya on the points in your review body. I ran Spark's `DataFramePivotSuite`, `pivot.sql` and `udf/udf-pivot.sql` with Comet on, locally against 3.4.3, 3.5.9, 4.0.4, 4.1.3 and 4.2.0, and they pass. The exception is the two Python UDF variants of `udf-pivot.sql` on 3.5.9, which fail in Spark's own harness before any query runs, because pyspark 3.5's cloudpickle can't pickle the UDF under my local Python 3.14. On 4.2.0 the run includes the collation pivot tests from SPARK-56306, and both PivotFirst fixtures pass on every profile. To check that the suites reach native `PivotFirst`, I reran them on 4.1.3 against a native library whose grouped `PivotFirst` never writes a slot. 19 failed, including pivoting by a constant and grouping by TIMESTAMP_NTZ. The DecimalType and timestamp-with-count tests pass either way, because their first aggregation runs in Spark, so the final `PivotFirst` can't be native. `PivotFirst.sql` covers decimal values and timestamp pivot columns natively instead. I've applied `run-spark-4.1-tests` and `run-all-spark-profiles` for the CI verdict. -- 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]
