On 2022-Apr-08, Amit Kapila wrote:
> On Thu, Mar 17, 2022 at 3:36 AM Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:
> >
> > Did you see some old code I wrote towards this goal?
> > https://www.postgresql.org/message-id/20150215044814.GL3391@alvh.no-ip.org
> > The intention was that DDL would produce some JSON blob that accurately
> > describes the DDL that was run;
>
> I have read that thread and found one of your emails [1] where you
> seem to be saying that JSON representation is not required for BDR.
> Will in some way going via JSON blob way make this project
> easier/better?
I don't know if replication support will be easier by using JSON; I just
think that JSON makes the overall feature more easily usable for other
purposes.
I am not familiar with BDR replication nowadays.
> For runtime conditions, one of the things you have mentioned in that
> thread is to add schema name in the statement at the required places
> which this patch deals with in a different way by explicitly sending
> it along with the DDL statement.
Hmm, ok. The point of the JSON-blob route is that the publisher sends a
command representation that can be parsed/processed/transformed
arbitrarily by the subscriber using generic rules; it should be trivial
to use a JSON tool to change schema A to schema B in any arbitrary DDL
command, and produce another working DDL command without having to know
how to write that command specifically. So if I have a rule that
"schema A there is schema B here", all DDL commands can be replayed with
no further coding (without having to rely on getting the run-time
search_path correct.)
--
Álvaro Herrera 48°01'N 7°57'E — https://www.EnterpriseDB.com/
"El sudor es la mejor cura para un pensamiento enfermo" (Bardia)
