HippoBaro commented on PR #11267: URL: https://github.com/apache/arrow-rs/pull/11267#issuecomment-5955975776
@alamb Let us know what you think. I’m happy to go either way on this. For example, I can drop the `1pct` benchmarks and instead add fixed-cardinality equivalents for the other physical types, not just strings. The motivation is that in #11273, the null vs non-null and high vs low cardinality dictionary paths, across the different physical type backends, start diverging more than they do on `main`, as the Arrow writer becomes better able to take advantage of the physical shape of the input data. Having coverage across those shapes is therefore useful for measuring that work. That said, I’m already adding quite a few benchmarks here and don’t want the suite to become unwieldy. Happy to trim or reshape the matrix based on what you think will be most useful. -- 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]
