[PR] perf(scan): stop split construction after limit coverage [paimon-rust]

["hhhizzz (via GitHub)" <gi...@apache.org>]

hhhizzz opened a new pull request, #470:
URL: https://github.com/apache/paimon-rust/pull/470

   ### Purpose
   
   Linked issue: close #469
   
   Push safe LIMIT planning into split construction so limited scans can stop building candidate splits once known split-level row counts cover the requested limit.
   
   This keeps manifest planning complete and preserves the existing safety gates: LIMIT early-stop is only applied when split-level `merged_row_count()` remains conservative, so scans with data predicates or effective row ranges still build the full candidate set.
   
   ### Brief change log
   
   - Add an internal LIMIT pushdown accumulator that runs while scan planning constructs `DataSplit`s.
   - Stop split construction once known `merged_row_count()` values cover the requested LIMIT.
   - Preserve fallback behavior when the LIMIT is not covered by known row counts, so the original candidate split set is returned instead of being unsafely pruned.
   - Return an empty plan for `LIMIT 0` without building split candidates.
   - Extend scan trace output with `split_candidates_built` and `limit_early_stopped` so tests and diagnostics can distinguish final split pruning from construction-time early stop.
   - Add/extend data-evolution integration coverage for successful early stop and row-range safety.
   
   ### Tests
   
   Targeted tests run:
   
   - `rtk cargo test -p paimon test_incremental_limit_accumulator -- --nocapture`
     - `2 passed`
   - `rtk cargo test -p paimon-integration-tests test_limit_pushdown -- --nocapture`
     - `3 passed`
   
   Full `cargo test` was not run locally.
   
   ### API and Format
   
   No storage format changes.
   
   The public scan trace surface is extended with additive fields for LIMIT planning diagnostics:
   
   - `split_candidates_built`
   - `limit_early_stopped`
   
   Existing read results and LIMIT semantics are unchanged; callers still need to enforce the final LIMIT at the query/read layer.
   
   ### Documentation
   
   No user-facing documentation update is required. This is an internal scan-planning optimization with trace diagnostics covered by 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: issues-unsubscribe@paimon.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org

Re: [PR] perf(scan): stop split construction after limit coverage [paimon-rust]

["JingsongLi (via GitHub)" <gi...@apache.org>]

JingsongLi merged PR #470:
URL: https://github.com/apache/paimon-rust/pull/470


-- 
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: issues-unsubscribe@paimon.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org