Hey there,
Please use the user mailing list for user-related questions (this list is
for Flink internals only).
At the moment outer joins are not directly supported in Flink, but there
are good indications that this will change in the next 4-8 weeks. For the
time being, you can use a CoGroup with a custom UDF to implement the
semantics of a left outer join.
If you dig through the mailing list archives for the past 2-3 weeks and
search for "outer join" you will find a thread discussing the details of
the workaround implementation.
Regards,
Alexander
2015-04-26 21:07 GMT+02:00 hager sallah <loveallah1...@yahoo.com.invalid>:
> how can handle left outer join for any two dataset this dataset inlcude
> any filed number
> example on two dataset data set one
> ExecutionEnvironment env = ExecutionEnvironment.getExecutionEnvironment();
> DataSet<Tuple4<Integer, String, String,String>>
> customer=env.readCsvFile("/home/hadoop/Desktop/Dataset/customer.csv")
> .fieldDelimiter('|')
> .includeFields("11110000").ignoreFirstLine()
> .types(Integer.class,String.class,String.class,String.class);dataset two
> ExecutionEnvironment orders =
> ExecutionEnvironment.getExecutionEnvironment();
> DataSet<Tuple3<Integer, String, String>
> customer=env.readCsvFile("/home/hadoop/Desktop/Dataset/order.csv")
> .fieldDelimiter('|')
> .includeFields("11110000").ignoreFirstLine()
> .types(Integer.class,String.class,String.classs);
>
