JingsongLi opened a new pull request, #10079: URL: https://github.com/apache/paimon/pull/10079
## Summary - Derive Java row-id conflict checks from CommitMessage.checkFromSnapshot at the commit entry points and remove the ordinary rowIdCheckConflict setter. - Tag Flink and Spark Data Evolution commit messages with their read snapshot. Reject inconsistent baselines and mixed tagged/untagged row-id messages. - Bump CommitMessageSerializer to v14 while keeping v13 reads. Preserve the dedicated materialize-DV compaction check. ## Validation - Focused core DataEvolutionTableTest, CommitMessageSerializerTest, and DataEvolutionDeletionVectorTest passed with normal Maven checks. - Flink 2 CommittableSerializerTest passed. - Spark 3 main and test sources compiled; Spark 4 main sources compiled; Spotless checks passed. - Full Spark runtime suites were not run. ## Compatibility and rollout - Existing v13 in-flight Flink messages deserialize without checkFromSnapshot. Restoring a row-id commit from an old checkpoint can lose its conflict-check baseline, or a mixed batch can be rejected. Do not roll out against such in-flight messages until a recovery migration and regression test exist. - Older runtimes cannot read v14 messages. Avoid mixed Paimon runtime versions on the same message channel, including downstream Morax channels. - Downstream callers of the removed public rowIdCheckConflict method need a source migration. Draft pending validation of the in-flight recovery and mixed-version rollout cases. -- 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]
