I agree with @StephanEwen that the current approach is not very clean. I'm wondering whether introducing a Scala `DataSource extends DataSet` and changing the return type of `ExecutionEnvironment#createInput` from `DataSet` to `DataSource` would really be API breaking. All existing programs should still compile since we would extend `DataSet`. What do you think @yanghua?
[ Full content available at: https://github.com/apache/flink/pull/6738 ] This message was relayed via gitbox.apache.org for [email protected]
