Github user StephanEwen commented on the issue: https://github.com/apache/flink/pull/4896 The diffs looks good, but what I cannot judge in a final manner is whether some tests now get not executed any more (accidentally). What would be good is to take the Travis output from the profiles that run tests in `flink-runtime` and `flink-tests` and compare the number of executed tests (maven prints that in the "test" and "verify" summary). If they are still the same (or differ by a number explained through the refactoring), then +1 to merge
---