The last point is correct. I thought that we were already aware of this shortcoming... but now I don't remember who I was talking to about it.
About the parents... The immediate parent of suppress is a single KTable, but may not be the one with a defined grace period. It might be a filter, in which case we need to examine the parent of the filter, or it might be a join, in which case we need to examine *both* parents of the join. The fact that we currently have a design flaw that prevents this situation doesn't imply that we should encode this limitation here. Once we fix that design flaw, we would have to remember that we also coded that flaw into this method and come back to revert it to the state it's in right now! [ Full content available at: https://github.com/apache/kafka/pull/5567 ] This message was relayed via gitbox.apache.org for [email protected]
