zhuqi-lucas commented on PR #186: URL: https://github.com/apache/datafusion-site/pull/186#issuecomment-4828457400
### Community walkthrough · DataFusion China Meetup I gave a walkthrough of this work at the **DataFusion China Meetup** ahead of opening this PR. Sharing the 24-slide deck here for reviewers / readers who want the talk-track version of the same material — same structure as the post (framework → Exact path → Inexact path → #22450 → future directions). 📎 **Slide deck** (`sort-pushdown-community-call.pdf`, 24 slides): _attached below — drag-and-drop preview_ Highlights from the talk that map to the post: - The architecture-3-eras framing for who drives the I/O + decode loop (pre-#20839 / #20839 / #22450) - The cascading-prune walkthrough — one row group fills `TopK`'s heap, every remaining RG cascades out at the next boundary - The three-layer pruning stack (file + RG + row), all driven by the same `TopK` dynamic filter - `topk_tpch` numbers (5 of 11 queries 3–4× faster, 0 regressions, total **−44%**) Happy to take feedback on either the post or the deck. -- 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]
