Hi, I worked on rewriting flink-test according to https://issues.apache.org/jira/browse/FLINK-2275
In "org.apache.flink.test.javaApiOperators.SortPartitionITCase" I hit something strange. When I rewrite the code slightly differently, the test passes or fails and I have no idea why. The following code works (result is of type java.util.List) > result = ds > .map(new IdMapper()).setParallelism(4) // parallelize input > .sortPartition(1, Order.DESCENDING) > .mapPartition(new OrderCheckMapper<Tuple3<Integer, Long, String>>(new > Tuple3Checker())) > .distinct().collect(); Rewriting the above as follows result in a failing test: > ds.map(new IdMapper()).setParallelism(4) // parallelize input > .sortPartition(1, Order.DESCENDING) > .mapPartition(new OrderCheckMapper<Tuple3<Integer, Long, String>>(new > Tuple3Checker())) > .distinct(); > result = ds.collect(); I have no clue what the problem might be. The code looks semantically identical to me. Can anyone explain the difference? Do I miss anything? Or is this a bug? You can find the working version of the code in my github repo: https://github.com/mjsax/flink/tree/flink-2275-migrateFlinkTests -Matthias
signature.asc
Description: OpenPGP digital signature
