[
https://issues.apache.org/jira/browse/FLINK-38818?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ouyangwulin updated FLINK-38818:
--------------------------------
Description:
As FLINK-38045, only support mysql , and now we support pgsql.
We add postgres implement.
`
For historical reasons, the current implementation does not utilize the Source
to build a schema cache. This requires the TransformOperator to maintain its
own complex state.
An optimization is proposed: to build a reusable schema cache at the Source (or
in an operator immediately following it).
The primary benefit of this approach is that the transform operator can be made
completely stateless. This simplifies its design, improves architectural
modularity, and makes the entire pipeline more robust and maintainable.
`
was:As [FLINK-38045], only support mysql , and now we support pgsql.
> [posgtres-cdc] Emit initial table schemas from state
> -----------------------------------------------------
>
> Key: FLINK-38818
> URL: https://issues.apache.org/jira/browse/FLINK-38818
> Project: Flink
> Issue Type: Improvement
> Components: Flink CDC
> Reporter: ouyangwulin
> Assignee: ouyangwulin
> Priority: Major
> Labels: pull-request-available
>
> As FLINK-38045, only support mysql , and now we support pgsql.
>
> We add postgres implement.
> `
> For historical reasons, the current implementation does not utilize the
> Source to build a schema cache. This requires the TransformOperator to
> maintain its own complex state.
> An optimization is proposed: to build a reusable schema cache at the Source
> (or in an operator immediately following it).
> The primary benefit of this approach is that the transform operator can be
> made completely stateless. This simplifies its design, improves architectural
> modularity, and makes the entire pipeline more robust and maintainable.
> `
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
