[
https://issues.apache.org/jira/browse/FLINK-10873?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16694199#comment-16694199
]
Hequn Cheng commented on FLINK-10873:
-------------------------------------
[~fhueske] Thanks for your information. I have tested it with TableApi.
UnionAll would be ok now. However, join test failed due to field missing. And
this is because we use the same RelBuilder object to construct and the same
tableEnv to translate plan.
I concluded that it is more complicated than I expected.
> Remove tableEnv in DataSetConversions#toTable and
> DataStreamConversions#toTable
> -------------------------------------------------------------------------------
>
> Key: FLINK-10873
> URL: https://issues.apache.org/jira/browse/FLINK-10873
> Project: Flink
> Issue Type: Improvement
> Reporter: Jeff Zhang
> Assignee: Hequn Cheng
> Priority: Major
>
> What I would like to achieve is to change the following code
> {code}
> val table = data.flatMap(line=>line.split("\\s"))
> .map(w => (w, 1))
> .toTable(tEnv, 'word, 'count)
> {code}
> to this
> {code}
> val table = data.flatMap(line=>line.split("\\s"))
> .map(w => (w, 1))
> .toTable('word, 'count)
> {code}
> The only change is that tableEnv is removed in method toTable. I think the
> second piece of code is more readable. We can create TableEnvironment based
> on the ExecutionEnvironment of DataSet/DataStream rather than asking user to
> pass it explicitly.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
