gt; ago in operational table to keep table size manageable and ensure the query
> performance not to be decreased. These deleted records for maintain purpose
> should be not synced into our streaming warehouse. So we have to filter our
> them in our flink sql jobs. But currently it is no
in purpose
should be not synced into our streaming warehouse. So we have to filter our
them in our flink sql jobs. But currently it is not convenient to do ROWKIND
filtering. That is why I ask flink support read ROWKIND metadata by ROW_KIND()
function. Then we can use the following flink
