discivigour opened a new pull request, #10976: URL: https://github.com/apache/paimon/pull/10976
### Purpose This is a prerequisite for #10867, which exposes `changelog-producer=on-read` through snapshot planning and Flink integration. Add a primary-key table reader that accepts an `IncrementalSplit` containing historical data files and new level-0 files. It reconstructs the previous value for each changed key, merges it with the delta, and emits the corresponding changelog. For example, if a compacted value of `13` already includes an older value of `10`, a new `SUM` contribution of `4` emits `-U 13` and `+U 17`. The reader reuses lookup files across splits, takes each file’s level from the split being read, and bounds the number of cached partition-bucket states. The caller closes the reader to release its cached lookup files. ### Tests The `paimon-core` Maven reactor build passed with 64 focused tests. The 13 new split-reader cases cover warm and cold lookups after compaction, deletes, equal sequence numbers, cache eviction and reload, and projected reads that spill with both sort engines. The run also covered existing lookup, merge, incremental read, and batch projection 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]
