danny0405 opened a new pull request, #19613: URL: https://github.com/apache/hudi/pull/19613
### Describe the issue this Pull Request addresses Hudi's metadata indexes can prune files or map indexed values to records, but Hudi does not currently provide snapshot-aware, relevance-ranked full-text search. Users must maintain a separate search system and coordinate its ingestion, retention, and document identity with the Hudi timeline. This PR proposes RFC-110 to define native full-text search whose lifecycle and snapshot coverage are managed by the Hudi metadata table. It also includes a small in-repository Rust prototype so the binary envelope and BM25 scoring contracts are executable during design review. ### Summary and Changelog - Add RFC-110 covering the SQL API, MDT partition and descriptor model, immutable Lance-inspired sidecar format, exact source-slice coverage, distributed BM25 execution, JNI boundary, rollout, and testing strategy. - Add RFC-110 to the RFC catalog as under review. - Add the dependency-free `hudi-native-text-index` Rust prototype with document-address, segment-statistics, BM25, and versioned binary-envelope contracts. - Add seven Rust unit tests for format validation, statistics, scoring, and block upper bounds. - No source code was copied. Lance influenced the architecture and is linked as a design reference in the RFC. ### Impact No existing Hudi runtime behavior, public API, storage format, dependency graph, or Maven build changes in this PR. The RFC is a proposal, and the Rust crate is an isolated prototype that is not yet wired into Hudi builds or query engines. ### Risk Level low. The change adds documentation and an isolated prototype only. The Rust crate is dependency-free, formatting-clean, and covered by unit tests; repository whitespace and PR-compliance checks also pass. ### Documentation Update The RFC is the design documentation for the proposed feature. No Hudi website update is appropriate until the RFC is accepted and user-facing implementation begins. ### Contributor's checklist - [x] Read through [contributor's guide](https://hudi.apache.org/contribute/how-to-contribute) - [x] Enough context is provided in the sections above - [x] Adequate tests were added if applicable -- 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]
