[
https://issues.apache.org/jira/browse/FLINK-41384?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18143663#comment-18143663
]
Sergey Nuyanzin commented on FLINK-41384:
-----------------------------------------
Merged as
[e4fe6ce2f146938556d2f0f950c88b170e7ef90e|https://github.com/apache/flink/commit/e4fe6ce2f146938556d2f0f950c88b170e7ef90e]
> Extra parse methods for JSON and XML functions in CASE branches
> ---------------------------------------------------------------
>
> Key: FLINK-41384
> URL: https://issues.apache.org/jira/browse/FLINK-41384
> Project: Flink
> Issue Type: Technical Debt
> Components: Table SQL / Runtime
> Reporter: Sergey Nuyanzin
> Assignee: Sergey Nuyanzin
> Priority: Major
> Labels: pull-request-available
>
> Several JSON and XML functions on the same input share one parse method in
> the generated code.
> This does not work inside {{CASE}}, {{OR}}, {{COALESCE}} when the input is an
> expression, for example TRIM(j). The expression is generated again in each
> branch. Each branch then gets its own parse method.
> {code:sql}
> SELECT CASE WHEN c1 THEN JSON_VALUE(TRIM(j), '$.a')
> WHEN c2 THEN JSON_VALUE(TRIM(j), '$.b')
> ELSE NULL END
> FROM t
> {code}
> The generated code has two identical parse methods instead of one.
> The result is correct. The generated code is just bigger than needed, and it
> grows with every branch.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
