[
https://issues.apache.org/jira/browse/FLINK-24924?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17445660#comment-17445660
]
Jark Wu commented on FLINK-24924:
---------------------------------
> I agree that exceptions are not optimal for streaming pipelines. But even for
> batch jobs, this is a reoccurring problem. Take a batch job for example that
> runs every night and suddenly produces an exception due to some corrupt data
> in the current batch. It is basically the same problem.
I don't think so. It's very easy for batch jobs to update SQL and re-run.
However, this is super hard for streaming jobs, because the state may not
compatible and the source may expired the needed data.
"COALESCE(CAST('aa' AS INT), 0) => NULL" is a bug of return type of CAST, we
should fix the result type signatures of all functions which returning null.
I think it's not a bad case or hard to explain for returning null.
> TO_TIMESTAMP and TO_DATE should fail
> ------------------------------------
>
> Key: FLINK-24924
> URL: https://issues.apache.org/jira/browse/FLINK-24924
> Project: Flink
> Issue Type: Sub-task
> Reporter: Francesco Guardiani
> Priority: Major
>
> In a similar fashion to what described
> https://issues.apache.org/jira/browse/FLINK-24385, TO_TIMESTAMP and TO_DATE
> should fail instead of returning {{null}}.
> In particular for these two functions, a failure in parsing could lead to
> very unexpected behavior, for example it could lead to records with null
> rowtime.
> We should change these functions to fail by default when parsing generates an
> error. We can let users handle errors by letting them use TRY_CAST for the
> same functionality:
> {code:sql}
> -- This fails when input is invalid
> TO_TIMESTAMP(input)
> -- Behaves the same as above
> CAST(input AS TIMESTAMP)
> -- This returns null when input is invalid
> TRY_CAST(input AS TIMESTAMP)
> {code}
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
