GitHub user schenksj added a comment to the discussion: Delta Contribution Review Plan
Quick update — rather than keep waiting on the tracking branch, I went ahead and started landing the split. **Part 1 is open upstream** as #4700 (core SPI for contrib leaf scans, `CometScanWithPlanData`) — it's core-only and inert on default builds. **Parts 2–9 are split out and staged as a stacked review-draft chain on my fork** (each part builds on the previous); they're ready to open here in dependency order as each base merges to `main`. Every part is carved, fully verified, and reviewed clean. | Part | What | PR | |---|---|---| | 1 | Core SPI for contrib leaf scans (`CometScanWithPlanData`) — core-only, inert on default builds | #4700 | | 2 | Build gate + inert Delta wiring (Maven profile, Cargo feature, proto, `DeltaIntegration` bridge) | [schenksj#4](https://github.com/schenksj/datafusion-comet/pull/4) | | 3a | Rust: driver-side planning (log replay, predicate pushdown, JNI) | [schenksj#5](https://github.com/schenksj/datafusion-comet/pull/5) | | 3b | Rust: executor-side read path (kernel read + deletion vectors) | [schenksj#6](https://github.com/schenksj/datafusion-comet/pull/6) | | 4a | Scala: claim/decline + scan marker | [schenksj#7](https://github.com/schenksj/datafusion-comet/pull/7) | | 4b | Scala: serde + native exec — end-to-end native reads | [schenksj#8](https://github.com/schenksj/datafusion-comet/pull/8) | | 5 | Change Data Feed native reads | [schenksj#9](https://github.com/schenksj/datafusion-comet/pull/9) | | 6 | Contrib test battery + CI workflow | [schenksj#10](https://github.com/schenksj/datafusion-comet/pull/10) | | 7 | Delta own-suite regression harness | [schenksj#11](https://github.com/schenksj/datafusion-comet/pull/11) | | 8 | Docs (design docs + user guide) | [schenksj#12](https://github.com/schenksj/datafusion-comet/pull/12) | | 9 | `FAILED_READ_FILE` parity for Delta reads | [schenksj#13](https://github.com/schenksj/datafusion-comet/pull/13) | Full tracking (sizes, dependencies, status) is in the #4366 description. A review of **#4700** would get the chain moving — keeping the tracking branch current is the time sink I mentioned. Thanks again for all the guidance. @andygrove @mbutrovich @parthchandra GitHub link: https://github.com/apache/datafusion-comet/discussions/4648#discussioncomment-17386495 ---- This is an automatically sent email for [email protected]. To unsubscribe, please send an email to: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
