Rich-T-kid opened a new pull request, #26817: URL: https://github.com/apache/datafusion/pull/26817
## Which issue does this PR close? - Part of #26780. ## Rationale for this change #26816 needs to hash dictionary values once and spread the cached hashes through each batch’s keys. The aggregation and join follow-ups for #26780 need the same operation, so it should live in `hash_utils` instead of being copied into each PR. The ClickBench run on #26816 improved 12 queries, with the affected queries running 1.08x to 1.96x faster. ## What changes are included in this PR? Expose the existing dictionary scatter path as `scatter_dictionary_hashes`. It keeps the specialized paths for null keys, null values, and multi-column hashing, and checks the input lengths before the unchecked lookup. ## What is the testing strategy for this PR? The existing dictionary hash test now compares direct scattering with `create_hashes`. The focused hash-utils tests pass in normal and forced-collision builds. The dictionary Criterion cases showed no detectable performance change. ## Are there any user-facing changes? No query behavior changes. This adds a helper for the other dictionary hash-reuse PRs. -- 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]
