[ 
https://issues.apache.org/jira/browse/FLINK-21310?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Tianshi Zhu updated FLINK-21310:
--------------------------------
    Description: 
[https://gist.github.com/tianshizz/fd1bdbcc4676783b7300984c0bf532c1]

The gist shows three unit tests, that have different union behaviors. In the 
first one, a data stream unions with itself. In the second one, two identical 
data streams are unioned. However, the two operations will lead to different 
end results, which seems surprising to me. 

In the third test when the list is long enough, union can show nondeterministic 
behavior. When running the third test for multiple times, each time the test 
will fail with a different `i` position.

Are these expected? Could someone help me understand the different behaviors 
here?

Thanks!

  was:
[https://gist.github.com/tianshizz/fd1bdbcc4676783b7300984c0bf532c1]

The gist shows three unit tests, that have different union behaviors. In the 
first one, a data stream unions with itself. In the second one, two identical 
data streams are unioned. However, the two operations will lead to different 
end results, which seems surprising to me. 

In the third test when the list is long enough, union can show indeterministic 
behavior. When running the third test for multiple times, each time the test 
will fail with a different `i` position.

Are these expected? Could someone help me understand the different behaviors 
here?

Thanks!


> unexpected behavior of DataFrame union
> --------------------------------------
>
>                 Key: FLINK-21310
>                 URL: https://issues.apache.org/jira/browse/FLINK-21310
>             Project: Flink
>          Issue Type: Bug
>    Affects Versions: 1.12.0
>            Reporter: Tianshi Zhu
>            Priority: Major
>
> [https://gist.github.com/tianshizz/fd1bdbcc4676783b7300984c0bf532c1]
> The gist shows three unit tests, that have different union behaviors. In the 
> first one, a data stream unions with itself. In the second one, two identical 
> data streams are unioned. However, the two operations will lead to different 
> end results, which seems surprising to me. 
> In the third test when the list is long enough, union can show 
> nondeterministic behavior. When running the third test for multiple times, 
> each time the test will fail with a different `i` position.
> Are these expected? Could someone help me understand the different behaviors 
> here?
> Thanks!



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to