Rich-T-kid opened a new pull request, #27443: URL: https://github.com/apache/datafusion/pull/27443
## Which issue does this PR close? - Part of #1178. ## Rationale for this change DataFusion's physical regex expression already has a scalar-pattern dictionary path, but literal regex predicates are currently planned with a cast that materializes dictionary strings first. Before changing that planner behavior, the performance contribution workflow requires canonical benchmark coverage, including the retained-dictionary case where direct value evaluation can be slower. ## What changes are included in this PR? This benchmark-only PR extends `binary_op` with paired dictionary and explicit cast-away regex cases. It covers 65,536 rows at cardinalities 16, 256, 4,096, and 65,536, plus an 8,192-row batch retaining a 65,536-value dictionary. The benchmark can remain unchanged when the planner starts preserving dictionary input and when the physical expression adds its retained-dictionary guard. ## What is the testing strategy for this PR? ```text cargo fmt --all -- --check cargo check -p datafusion-physical-expr --bench binary_op cargo bench -p datafusion-physical-expr --bench binary_op -- dictionary_regex --sample-size 10 --warm-up-time 1 --measurement-time 1 git diff --check ``` Current `main` baseline from a quiet rerun (Criterion middle estimates): | Case | Dictionary | Cast away | |---|---:|---:| | NDV 16 | 32.447 us | 915.31 us | | NDV 256 | 35.331 us | 917.60 us | | NDV 4,096 | 78.960 us | 931.96 us | | NDV 65,536 | 745.73 us | 898.19 us | | retained 65,536 values / 8,192 rows | 722.83 us | 122.72 us | ## Are there any user-facing changes? No. This PR only adds benchmark coverage. -- 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]
