ParentFull threadaphyr·I (and apparently the Confluent docs?) may be wrong about this. I've added an update to the report.View on HN