JingsongLi opened a new pull request, #1053: URL: https://github.com/apache/paimon-rust/pull/1053
### Purpose Linked issue: close #1052 Align primary-key BLOB reads with Java's filter-then-limit behavior and lazy descriptor access. Invalid payloads in discarded rows or unused Boolean branches must not fail a read. ### Brief change log - Apply one exact row quota after primary-key merging and filtering, before resolving output payloads. - Evaluate AND/OR/NOT with active row masks; resolve predicate BLOBs only in needed branches and cache payloads for repeated leaves and output projection. Null checks only inspect validity. - Resolve output-only BLOBs after selection. Identify metadata predicates by their reserved names rather than aliased positions. - Rebase ARRAY/MAP child offsets and select aligned map keys, removing hidden children while retaining non-nullable element/value schemas. - Use the shared BLOB plan for raw primary-key streaming row-kind reads and open mixed snapshot/delta groups lazily with their remaining quota. - Share the existing Data Evolution batch quota helper. ### Tests - `cargo test --locked -p paimon --lib`: 3,734 passed, 6 ignored. - `cargo test --locked -p paimon --doc`: 2 passed, 5 ignored. - `cargo clippy --locked --all-targets --workspace --features fulltext,vortex -- -D warnings`: passed. - `cargo fmt --all --check`: passed. - Python binding CI suite with an isolated warehouse fixture: 311 passed. ### API and Format No new public API, options, dependencies, or storage format changes. Existing ReadBuilder LIMIT is now enforced by primary-key readers. ### Documentation Updated the read-limit documentation. The paired PyPaimon contribution removes the native BLOB LIMIT fallback and adds direct Native regression tests. -- 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]
