[
https://issues.apache.org/jira/browse/FLINK-25459?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17465922#comment-17465922
]
qyw commented on FLINK-25459:
-----------------------------
[~wenlong.lwl] If it is a simple field, I agree with your solution However, for
"row" type fields, type matching depends on order. For example:
query schema: [ceshi: ROW<`name` INT, `id` STRING>]
Sink schema: [ceshi: ROW<`id` INT, `name` STRING>
According to my previous understanding, the data mapping relationship is: name
-> name and id-> id, which will be matched according to the field name
In fact, the data mapping relationship is: name ->id and id -> name, which are
matched in order. Because my data is in JSON format, I am very confused about
this phenomenon
> When inserting row type fields into sink, the order needs to be maintained
> --------------------------------------------------------------------------
>
> Key: FLINK-25459
> URL: https://issues.apache.org/jira/browse/FLINK-25459
> Project: Flink
> Issue Type: Improvement
> Components: Table SQL / Planner
> Affects Versions: 1.14.2
> Reporter: qyw
> Priority: Major
>
> When I insert a row type value into sink, why do I need to maintain the field
> order in row?
> This is the comparison between my query schema and sink schema:
> Query schema: [ceshi: ROW<`name` STRING, `id` INT, `age` INT, `test` ROW<`c`
> STRING>>]
> Sink schema: [ceshi: ROW<`id` INT, `name` STRING, `age` INT, `test` ROW<`c`
> STRING>>]
> An error will be thrown:
> Exception in thread "main" org.apache.flink.table.api.ValidationException:
> Column types of query result and sink for registered table
> 'default_catalog.default_database.kafka_target' do not match.
> Cause: Incompatible types for sink column 'ceshi' at position 0.
>
>
> Is this phenomenon reasonable?
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
