bvolpato opened a new pull request, #26839: URL: https://github.com/apache/datafusion/pull/26839
> [!IMPORTANT] > > **Fork CI setup:** For pull requests from a fork to `main`, enable workflows in your fork's **Actions** tab before you push. > See the [fork CI setup guide](https://datafusion.apache.org/contributor-guide/#pull-requests-from-forks). ## Which issue does this PR close? Part of #26759 and #13813. ## Rationale for this change The clustered grouping differential matrix finishes late in regular test runs. Under forced hash collisions, its hash-based reference also becomes expensive, although the clustered implementation does no hashing. ## What changes are included in this PR? Split the key types into deterministic shards that the test harness can schedule in parallel. Every original type, nullability, direction, batch-size, and emit-limit combination remains in regular and extended runs. Gate the matrix and its exclusive helper/imports under `force_hash_collisions`. Other clustered grouping tests remain active in that mode. ## What is the testing strategy for this PR? With `--test-threads=8`, the focused matrix took 33.6s before the change and 14.9s after it. These are single runs on a shared Linux host. The physical-plan unit suite also passes with `force_hash_collisions` enabled. Formatting, all-target/all-feature Clippy, and the extended workspace suite passed. Full workspace validation included the RSS sampling prerequisite already carried in #26445. ## Are there any user-facing changes? None. This changes test scheduling and redundant collision-mode coverage. AI-assisted contribution. -- 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]
