adriangb commented on code in PR #25032: URL: https://github.com/apache/datafusion/pull/25032#discussion_r3968299584
########## benchmarks/sql_benchmarks/predicate_eval/predicate_eval.suite: ########## @@ -1,4 +1,4 @@ -description = "Conjunctive filter evaluation micro-benchmarks covering predicate cost, selectivity, cardinality, width, scale, correlation, and drift" +description = "Conjunctive filter evaluation micro-benchmarks covering predicate cost, selectivity, cardinality, width, scale, correlation, drift, and nullable predicates" Review Comment: Sorry for not following up (had to call it a night). After looking at what was possible, I decided against it in this PR. https://github.com/apache/datafusion/commit/9307536bb5319064bdae5439d85855208016c756 is a good chunk more of code, and is not actually testing the real results. The real results are hard to test because it is slow to run the queries at the right scale to get those counts. I also tried randomly sampling queries but some take up to ~2s, which I felt was not worth introducing into the test suite. And realistically the results are not really that important here, there are plenty of other tests that assert the correctness of filters. Thus I propose: let's leave results untested. If they drift / cause issues in the future we can evaluate removing them, testing them, etc. -- 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]
