flink-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Till Rohrmann (JIRA)" <j...@apache.org>
Subject [jira] [Created] (FLINK-3538) DataStream join API does not enforce consistent usage
Date Mon, 29 Feb 2016 14:13:18 GMT
Till Rohrmann created FLINK-3538:

             Summary: DataStream join API does not enforce consistent usage
                 Key: FLINK-3538
                 URL: https://issues.apache.org/jira/browse/FLINK-3538
             Project: Flink
          Issue Type: Improvement
          Components: DataStream API, Scala API
    Affects Versions: 1.0.0
            Reporter: Till Rohrmann

In the Scala DataStream API the {{join}} operation does not enforce that the user has specified
a {{KeySelector}} for both input sides before applying a window function. Moreover, the order
of the {{where}} and {{equalTo}} clause is not fixed and it is possible to specify multiple
{{where}} and {{equalTo}} clauses. In the latter case, it is not clear which {{KeySelector}}
will eventually be used by the system.

So the following Flink programs compile without a compilation problem (the first two lines
will only fail at runtime):
inputA.join(inputB).equalTo{x => x}.window(TumblingProcessingTimeWindows.of(Time.seconds(10)))
      .apply(new DefaultFlatJoinFunction[String, String]()).print()

inputA.join(inputB).where{x => x}.window(TumblingProcessingTimeWindows.of(Time.seconds(10)))
      .apply(new DefaultFlatJoinFunction[String, String]()).print()

inputA.join(inputB).equalTo{x => x}.where{x => x}.where{x => "1"}.equalTo{x =>
      .apply(new DefaultFlatJoinFunction[String, String]()).print()

This is unlike the Java DataStream API where a clear pattern of {{join}} then {{where}} and
then {{equalTo}} is enforced. I would propose to do the same for the Scala API.

This message was sent by Atlassian JIRA

View raw message