Date: Fri, 12 Nov 2021 15:13:58 -0300
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Tomas Vondra <tomas.vondra@enterprisedb.com>
Cc: Pg Hackers <pgsql-hackers@lists.postgresql.org>
Subject: Re: support for MERGE
Message-ID: <202111121813.44jbkflr7kgn@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <3f45ed11-8b55-70f0-0c01-ff863313d0d1@enterprisedb.com>
Archived-At: 
 <https://www.postgresql.org/message-id/202111121813.44jbkflr7kgn%40alvherre.pgsql>
Precedence: bulk

On 2021-Nov-12, Tomas Vondra wrote:

> On 11/12/21 18:57, Alvaro Herrera wrote:

> > Secondarily, and I'm now not sure that I really want to do it, is change
> > the representation for executor: instead of creating a fake join between
> > target and source, perhaps we should have just source, and give
> > optimizer a separate query to fetch tuples from target.
> 
> When you say you're not sure you want to change this, is that because you
> don't have time for that, or because you think the current approach is
> better?

I'm not sure that doing it the other way will be better.  We'll have two
queries to optimize: one is reading from the data source, and the other
is fetching tuples from the target table based on the conditions applied
to each row obtained form the data source.  Right now, we just apply a
join and let the optimizer do it's thing.  It's simpler, but ...

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/