Rachelint opened a new pull request, #26081: URL: https://github.com/apache/datafusion/pull/26081
## Which issue does this PR close? None. Benchmark-only ablation for #25959; do not merge. ## Rationale for this change Measure whether the new Partial working-set policy retains its benefit when Final uses ordinary hash aggregation. The earlier requests on #25959 failed during checkout because the runner only fetched that PR history. This draft makes both experimental refs reachable. ## What changes are included in this PR? - `126109f50158b9dbb4e9fd13ce46d9520268b789`: old Partial policy from `cf6819d`, with Final bucketing disabled. - `70ac2eed7ad82bd0915ad526cd735bdcacbf7458`: new Partial policy from `efd384f`, with Final bucketing disabled. - The merge commit preserves the new variant tree and includes the baseline in its ancestry for the runner. - The only ablation change forces the Final constructor bucket threshold to zero; Partial still reads the experimental configuration. Original #25959 is unchanged. ## What is the testing strategy for this PR? Formatting and focused diff inspection passed. Two partitioned ClickBench comparisons: old versus new Partial with Final bucketing disabled on both sides; then complete `efd384f` versus new Partial without Final buckets. Shared threshold: 262144. No full test suite rerun for this experimental switch. ## Are there any user-facing changes? Experiment only, not intended for merge. -- 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]
