andygrove opened a new pull request, #6487: URL: https://github.com/apache/datafusion-comet/pull/6487
## Which issue does this PR close? Refs #5507 and #6402. Backport of merged #6468 (source c312b1c193a62065bff7350227ef99ce3ea67f22; main merge 300c7484a3a03df465ce11ee5f222532c9a58bde). ## Rationale for this change RC1's native RANK/DENSE_RANK cutoff can drop tied rows when order keys contain nested -0.0/+0.0 or NaN. Restore Spark's correct results for RC2 by falling back for these keys. ## What changes are included in this PR? Cherry-pick #6468 onto branch-1.1: the nested-floating-point fallback, compatibility documentation, and regression coverage. ROW_NUMBER and supported scalar-key paths keep their existing behavior. The cherry-pick applied without conflicts. ## How are these changes tested? On branch-1.1 with this patch, a fresh `cargo build --offline` passed. JDK 17/default Spark 4.1: `./mvnw -o -B test -Dtest=none '-Dsuites=org.apache.comet.exec.CometWindowExecSuite floating-point values nested'` passed (one selected test exercising the query matrix). `git diff --check` passed. Other Spark profiles and the full Spark SQL suite were not run locally; release-branch CI 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]
