kumarUjjawal opened a new pull request, #26791: URL: https://github.com/apache/datafusion/pull/26791
## Which issue does this PR close? Related to #17261. Provides the benchmark baseline for #26790. ## Rationale for this change The existing union construction benchmarks use nullable fields, so selecting the output field stops at the first child. They do not expose the repeated metadata merges when every candidate is non-nullable. Adding these cases separately lets the benchmark bot compare the same workloads before and after the optimization in #26790. ## What changes are included in this PR? Extend `union_schema` with non-nullable flat schemas at 10, 100, and 1,000 children. Run the existing shared-schema, equal-schema, differing-metadata, and differing-name scenarios for each size. The benchmark exercises the public `UnionExec::try_new` constructor. ## What is the testing strategy for this PR? All 32 benchmark cases execute successfully in Criterion's test mode: ```sh cargo bench --profile release-nonlto -p datafusion-physical-plan --bench union_schema -- --test ``` The added cases were also measured against the unoptimized implementation during the investigation for #26790. ## Are there any user-facing changes? No. This adds benchmark coverage only. AI assistance: OpenAI Codex. -- 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]
