JingsongLi opened a new pull request, #10779: URL: https://github.com/apache/paimon/pull/10779
## Purpose Keep existing row-ID deletion on the selected execution backend without materializing Python integers along the Native path. ## Changes - Accept integer Arrow Array/ChunkedArray through existing direct, batch and stream deletion APIs. Keep shared input validation outside the executors and preserve existing sequence coercion. - Pass Ray delete columns directly to the writer; materialize action row IDs only when collection was requested. Preserve bucket ownership, duplicate-source rejection, captured snapshots and commit messages. - Retain Native execution for empty deletes instead of initializing a Python updater. - Move Python-only DV liveness validation out of the Ray module into Python execution behind dispatch; Native retains its Core validation. - Refresh historical/incremental update tests to guard the Python executor and assert Native selection through the public wrapper. - Add Arrow slicing, chunking, buffer ownership, overflow, NULL, idempotence, snapshot and DV regressions; document the accepted inputs. ## Verification - Joint current SDK/PyPaimon regression run: 912 passed, 662 native plans exercised. - Pure Python focused execution/boundary tests: 156 passed (26 Native cases skipped without an installed SDK); existing row-ID delete tests: 8 passed. - Four actual Ray update/delete worker tests passed. - Repository Flake8 for all 15 changed Python files and `git diff --check`: passed. The joint run registered the SDK in a Cargo test executable without installing or rebuilding the extension in Python environments. Fresh timezone checks used that executable in separate processes. Depends on https://github.com/apache/paimon-rust/pull/1210. Native CI continues to build Rust main; this PR is draft until that dependency merges. No SDK-version probes or fallback compatibility code are added. -- 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]
