rich7420 opened a new pull request, #6655: URL: https://github.com/apache/datafusion-comet/pull/6655
## Which issue does this PR close? Part of #6634. ## Rationale for this change The existing SQL fixture covers `bit_count` on INT and BIGINT, but omits TINYINT, SMALLINT, and BOOLEAN. Move the Scala min/max test to a typed Parquet SQL fixture so these cases are covered by the SQL test framework. ## What changes are included in this PR? Add `bitwise/bit_count_types.sql` and remove `CometBitwiseExpressionSuite`'s `bitwise_count - min/max values` test in the same change. Preserve the original three rows, all four integer types, boolean literal queries, and both Parquet dictionary settings. Add a boolean column, NULLs, -1/0/1, and BIGINT values around 2^32. Spark 3.4.3 generates invalid Java for boolean `bit_count`. The fixture uses interpreted Spark reference execution by disabling whole-stage codegen and selecting `NO_CODEGEN` for projections. Plain queries still require Comet operator replacement. Allowing codegen fallback alone is insufficient because Spark disables whole-stage fallback in test mode. This migrates one test. The remaining hash and bitwise migrations in #6634 are separate work. ## How are these changes tested? Plain `query` records compare answers against Spark and assert Comet operators. The SQL runner disables constant folding, including for the boolean literal queries. The original Scala test and both new dictionary variants passed together on Spark 3.5 before removing the Scala test. The final fixture passed on Spark 3.4, 3.5, 4.0, 4.1, and 4.2 (two dictionary variants per profile, 10 tests total, no failures or skipped cases). Spark 3.5 also passed `-Pstrict-warnings`. Spotless, Scalastyle, Apache RAT, and `git diff --check` passed. Local verification uses base `35de0b9ec` and its matching native library; PR CI against the latest main is pending. -- 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]
