zhangyue19921010 opened a new pull request, #11210: URL: https://github.com/apache/paimon/pull/11210
### Purpose Fixes #11208. Batch append compaction consumes one eligible file as lookahead after each full scan chunk. That file is never passed to a task or retained for the next scan, so a batch job can leave it uncompacted. Cache the lookahead entry in the file iterator and consume it at the start of the next scan. The scan bound and streaming behavior stay unchanged. ### Tests Five real table cases cover multiple chunks, an exact scan boundary, fewer files than the limit and a final singleton. On unchanged `00b4d05a`, the three multiple-chunk cases fail exact planned/retained file coverage. With the fix, all five pass, execute and commit the compaction tasks, and verify all rows. Tasks remain within the configured five-file bound. 30 selected table and coordinator tests pass on JDK 11.0.18. Maven `verify` passes Checkstyle, Spotless, Enforcer and RAT without `fast-build`. AI-generated with TRAE CLI (GPT-6.1-Sol). -- 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]
