rkrishn7 opened a new pull request, #24714: URL: https://github.com/apache/datafusion/pull/24714
## Which issue does this PR close? - Closes: #24602 ## Rationale for this change Long-lived grouped aggregation state currently must be drained and rebuilt to read a subset of groups. Preserving reads avoid that full-state round trip. Although the diff is quite large, the conceptual change is fairly contained. Most of the code updates apply the same preserving-read contract across DataFusion’s existing `GroupColumn` and `GroupsAccumulator` implementations. ## What changes are included in this PR? - Adds validated `GroupSelection` API for all groups or ordered group indices. - Adds optional preserving-read APIs and capability checks to `GroupValues` and `GroupsAccumulator`. - Implements selected reads for built-in group values and grouped accumulators. - Documents the required `GroupColumn` migration. ## Are these changes tested? Yes ## Are there any user-facing changes? `GroupColumn::values_preserving` is a new required public trait method, so this is technically a breaking change. However, I think the downstream impact is low since `GroupColumn` seems to be a fairly low-level trait, and DataFusion doesn't expose a public path for supplying a custom implementation to `GroupValuesColumn` -- 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]
