There is a risk that our reflection code will brake on sub version upgrade, however it's not that big a problem, since in our connectors we are freezing Kafka client's version that we are using. Keep in mind that Kafka client's version that we are using is independent of Kafka server's that we are talking to. For example our Kafka 0.11.3 connector as far as I know is able to work with both Kafka 1.0 and 1.1 (with lower performance however).
However, I thought a little bit more about arguments for providing separate module for each major Kafka version and there is one huge argument in favour of that: bug fixes in Kafka. There are quite a lot of bugs in Kafka and sometimes they are not backported to older versions. I think this reason is good enough for adding `flink-connector-kafka-1.1` module, even if our code is almost identical. Also this would reduce our user's confusion "Does flink-connector-kafka-0.11 works with Kafka 1.0?". [ Full content available at: https://github.com/apache/flink/pull/6577 ] This message was relayed via gitbox.apache.org for [email protected]
