[
https://issues.apache.org/jira/browse/FLINK-18632?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17159992#comment-17159992
]
luoziyu commented on FLINK-18632:
---------------------------------
When i use pojo type, result are :
(true,Result\{user='Bob', cnt=1})
(true,Result\{user='Bob', cnt=1})
(true,Result\{user='Bob', cnt=1})
and tuple type's result are
(true,(1,Bob))
(false,(1,Bob))
(true,(1,Bob))
the picture test_code.png in attachment is my test case.
> RowData's row kind do not assigned from input row data when sink code
> generate and physical type info is pojo type
> ------------------------------------------------------------------------------------------------------------------
>
> Key: FLINK-18632
> URL: https://issues.apache.org/jira/browse/FLINK-18632
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Planner
> Affects Versions: 1.10.0, 1.11.0
> Environment: mac os 10.15.3
> Reporter: luoziyu
> Priority: Major
> Attachments: bug.jpg, test_code.png
>
>
> I use tuple type and pojo type to test retract stream, the test data is same
> and when i use
> toRetractStream(Table table, Class<T> clazz) api, the retract msg become to
> insert msg from sink conversion. I found the SinkCodeGenerator object did not
> give a row kind to afterIndexModify variable, so the delete msg become insert
> msg when it comes into processElement function generator by SinkCodeGenerator.
> At last i add line of code like the pic in attachment and it works, so is it
> a bug?
>
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
