On 06-Mar-2013, at 4:12 PM, Shigeru Hanada <shigeru.hanada@gmail.com> wrote:
> On Wed, Mar 6, 2013 at 12:35 PM, Pavan Deolasee
> <pavan.deolasee@gmail.com> wrote:
>> In the context of postgres_fdw, I am not sure if we need an additional
>> system column like a node_id. Would there be a possibility where
>> tuples to-be-modified are coming from different foreign tables and at
>> runtime we need to decide where to execute the UPDATE/DELETE operation
>> ? If we start supporting inheritance involving foreign tables as child
>> tables, this will become a reality.
>
> Foreign tables have tableoid system column which holds pg_class.oid of
> the foreign table. It seems sufficient to determine source server.
>
I think you are right. In the context of foreign tables, tableoid might be enough to get the source. Unfortunately,
Postgres-XChad not yet leveraged foreign tables fully and hence special handling was required.
Thanks,
Pavan
> --
> Shigeru HANADA