[
https://issues.apache.org/jira/browse/FLINK-29748?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17624151#comment-17624151
]
Aitozi commented on FLINK-29748:
--------------------------------
My connector has implemented the \{{SupportsPartitionPushDown}}. But, due to
the \{{PartitionPushDown}} optimization, the effective partition count can be
different from the original source table. If I want to validate the source do
not consume too much partition, from the connector's perspective, It do not
know whether the optimization is finished, so it do not know when to apply the
validation on the final optimization results.
> Expose the optimize phase in the connector context
> --------------------------------------------------
>
> Key: FLINK-29748
> URL: https://issues.apache.org/jira/browse/FLINK-29748
> Project: Flink
> Issue Type: Improvement
> Components: Table SQL / Planner, Table SQL / Runtime
> Reporter: Aitozi
> Priority: Minor
>
> Currently, in the connector it can not know whether the whole optimize is
> finished.
> When the optimize finished, the all information is static, eg: the reading
> partitions. If I want to validate the final optimized result (like whether
> the reading partition is too much or empty), it needs the context of what is
> the current phase. I think the {{ScanContext}} is ok to expose this
> information.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
