neilconway opened a new pull request, #25427: URL: https://github.com/apache/datafusion/pull/25427
## Which issue does this PR close? - Closes #25426. ## Rationale for this change When one argument of `array_has_any` is a scalar and the element type is not a string, the fast path `array_has_any_with_scalar_general` row-converts the list column's full child array, which does a lot of unnecessary work for sliced lists. Benchmarks: - array_has_any_scalar/i64_no_match_sliced/1: 4.14 ms -> 167.4 µs, −96.0% - array_has_any_scalar/i64_no_match_sliced/10: 4.11 ms -> 125.9 µs, −96.9% - array_has_any_scalar/i64_no_match_sliced/100: 4.25 ms -> 257.8 µs, −93.9% - array_has_any_scalar/i64_no_match_sliced/1000: 4.12 ms -> 141.1 µs, −96.6% Results on unsliced inputs were unchanged, within run-to-run noise. ## What changes are included in this PR? * Improve `array_has_any_with_scalar_general` to only touch the visible portion of the input list * Add a benchmark * Add unit tests for correct behavior on sliced lists ## What is the testing strategy for this PR? Existing tests pass, new tests added. Benchmark confirms that the inefficiency has been addressed. ## Are there any user-facing changes? No. -- 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]
