Hi Alex, right now, there is not native support for outer joins in Flink. This is on our road map though.
You can however do an outer join using a CoGroup function. CoGroup groups two data sets on a join-key and calls a user function for each pair of groups with identical keys. In the user function, each group is represented by an iterator and you can check if the iterator is empty which is the special case of an outer join. For the example you showed with the union of the names and myNull data sets. Did you check the result directly after the union or after the join? Regarding the exception you posted, is it possible that you post the program that caused the exception? Thank you very much, Fabian 2014-08-17 18:48 GMT+02:00 Alexander Sirotin <[email protected]>: > Hello everybody, > > I am reading a text-file and like to outer-join it with another text-file. > Because I do not know, how to call an outer-join throw Flink, I am adding a > dummy-row "-1" via union to one of these Dataset: > > DataSet<Tuple3<String, Integer, Integer>> myNull = env > .fromElements(new Tuple3<String, Integer, Integer>("-1", > new Integer(-1), new Integer(-1))); > > names = names.union(myNull); > > But if I look to the results, I never can see this value or the values > from the other datasets, who should be joined with it. > > Thank you very much. I hope someone knows how I can solve this problem. > > Best regards, > Alex >
