JingsongLi opened a new pull request, #10410: URL: https://github.com/apache/paimon/pull/10410
### Purpose Enable native primary-key BLOB LIMIT reads using https://github.com/apache/paimon-rust/pull/1053. Remove the obsolete native read/write exclusion for finite FLOAT/DOUBLE sequence fields already supported by Rust core. - Keep BLOB reads serial across splits when a global limit can discard payloads, including managed scalar/ARRAY/MAP BLOB fields. - Retain explicit batch-iterator cleanup when LIMIT serializes native readers, so early close reaches the underlying reader. - Add real-file tests for all four merge engines, post-merge selection, output-only BLOBs, Boolean short-circuiting, null checks, cached payloads, global split quotas, descriptor mode, zero limit, and streaming row kinds. Managed BLOB creation, writing, and reading run through Rust, with Python fallback guarded. - Extend the existing cross-write/read sequence matrix with FLOAT and DOUBLE values. The CI workflow continues to build paimon-rust from `main`. Merge the linked Rust PR first, then rerun this PR's Native CI. No compatibility fallback is introduced. ### Tests Locally tested against the linked Rust branch: - Native regression suite with planning/read/write/update/commit enabled: 2,276 passed, 7 skipped, 125 subtests passed. - Final BLOB/reader regression suite against the quota-batched core: 110 passed. - Python binding suite: 311 passed. - Python core subset used by the 3.6/3.7 CI jobs, run with the local Python 3.13 interpreter: 126 passed, 1 skipped. - Flake8 for every changed Python file: passed. -- 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]
