Re: BUG #15857: Parallel Hash Join makes join instead of exists

Pantelis Theodosiou <ypercube@gmail.com>

From: Pantelis Theodosiou <ypercube@gmail.com>
To: krujkov@gmail.com, pgsql-bugs@lists.postgresql.org
Date: 2019-06-18T09:08:44Z
Lists: pgsql-bugs
On Tue, Jun 18, 2019 at 9:56 AM Pantelis Theodosiou <ypercube@gmail.com>
wrote:

>
> ...
>
> I think it is correct result. This:
>
> > UPDATE c SET base_id = id / 10;
>
> would result in 9 rows (id from 1 to 9) to be updated with base_id = 0, as
> it should with integer division. These 9 rows will not match the condition:
>
> WHERE c.base_id = a.id
>
> as there is no row in a with a.id = 0
>
>
>
Please ignore the above, I  didn't read carefully.

>

Commits

  1. Prevent Parallel Hash Join for JOIN_UNIQUE_INNER.