AdamGS opened a new pull request, #20082: URL: https://github.com/apache/datafusion/pull/20082
## Which issue does this PR close? - Closes #20081. ## Rationale for this change Ran into this while investigating #20078, in my benchmarks this saves around %10, which can quickly add up to a few ms per query. The benchmarking code is [here](https://gist.github.com/AdamGS/4fc6612814786a95a673378de7d61202), measuring it on my laptop the effect is: ``` tpc-ds/q76/cs/16 time: [927.65 µs 928.86 µs 930.14 µs] change: [−12.309% −11.983% −11.667%] (p = 0.00 < 0.05) Performance has improved. tpc-ds/q76/ws/16 time: [929.00 µs 930.51 µs 932.13 µs] change: [−12.102% −11.808% −11.475%] (p = 0.00 < 0.05) tpc-ds/q76/cs/128 time: [6.8376 ms 6.8460 ms 6.8552 ms] change: [−12.626% −12.411% −12.207%] (p = 0.00 < 0.05) tpc-ds/q76/ws/128 time: [6.8394 ms 6.8536 ms 6.8710 ms] change: [−12.039% −11.813% −11.575%] (p = 0.00 < 0.05) ``` ## What changes are included in this PR? I've also changed `[cfg(test)]` to a debug assertion, which seems more in the spirit of https://github.com/apache/datafusion/issues/18001, and will make the check fail even if a test in another crate triggers, which effectively increases the coverage. ## Are these changes tested? The functionality is identical, and in addition to existing tests I've changed the defensive assertion into a debug check and ran tests for other crates that make use of this codepath (like `datafusion-datasource-parquet` and `datafusion-pruning`). ## Are there any user-facing changes? None -- 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]
