2010YOUY01 commented on PR #22729: URL: https://github.com/apache/datafusion/pull/22729#issuecomment-4618766692
> But seems it may not help very much to reduce codes about supporting blocked memory management? That's true, the actual code change is about the same. The major thing get reduced is the cognitive overhead when maintaining this feature, I'll try to explain better: I think the real pain point is: 1. To review and merge #15591, reviewers need to understand how the actual blocked state change interacts with `row_hash.rs`. 2. `row_hash.rs` is way too complex, so it is hard for reviewers to see the boundary between the actual blocked state-management changes and the existing hash-aggregation state machine. The purpose of this refactor is to make that boundary explicit. With this split, the key blocked state-management changes (`Accumulator` + `GroupValues`) should only interact with the new dedicated stream, rather than being wired into the complex state machine in `row_hash.rs`. > I think maybe #15591 and refactor can be pushed forward in parallel, they don't have very much in common. Yes I agree. I was thinking maybe #15591 can be updated after this refactor, if we can ensure the blocked `Accumulator` and `GroupValues` can only get planned into the new aggregate streams in this refactor, and the old `row_hash.rs` won't get affected, that change would be easier to review. -- 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]
