That is a good point. I was thinking of "final results emitted" in aggregate, as in the downstream state is eventually consistent with the upstream state.
It sounds like your reading is more like "each result emitted is eventually consistent with its upstream version". I think this reading is actually the more likely one. And, as you point out, this statement is almost nonsense: if you look at each result when it gets emitted, it is fully consistent with the upstream state. And of course, "eventual consistency" as a term probably opens up a whole bag of worms we don't want to deal with anyway, as people may bring assumptions (and bad associations) from distributed databases. I'll just say "will match the upstream" instead. [ Full content available at: https://github.com/apache/kafka/pull/5567 ] This message was relayed via gitbox.apache.org for [email protected]
