andygrove commented on PR #4720: URL: https://github.com/apache/datafusion-comet/pull/4720#issuecomment-5005107973
Pushed `a499164d0` adding test coverage and clearer comments for the `RESPECT NULLS` fallback. New SQL file test `spark/src/test/resources/sql-tests/expressions/aggregate/collect_respect_nulls.sql`, gated with `-- MinSparkVersion: 4.2`: - `collect_list(v) RESPECT NULLS` and `collect_set(v) RESPECT NULLS` (global, per-group, and an all-null group) use `query expect_fallback(...)`, asserting both the exact fallback reason (`... with RESPECT NULLS (ignoreNulls = false) is not supported`) and Spark-identical results. - A closing default-behavior `collect_list` query stays native (`checkSparkAnswerAndOperator`) to prove the fallback is specific to `RESPECT NULLS` rather than a blanket collect fallback. This runs in the `Spark 4.2, JDK 17` job's `expressions` bucket (which includes `CometSqlFileTestSuite`) and is skipped on 3.4 through 4.1, where the `ignoreNulls` field does not exist and the branch is an unreachable no-op. I also expanded the `getSupportLevel` comments on `CometCollectList`/`CometCollectSet` to spell out that the fallback is only reachable on 4.2+. For transparency: the analysis and these tests were done with LLM assistance. -- 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]
