; probably wouldn't do it.
> AFAICS there's no chance to make this bit smarter until the estimates
> get much better to reality.
Got it. Thanks. I guess we'll have to emit correlated subqueries/CTEs.
Sincerely,
Bakhtiyar
On Wed, Jun 21, 2023 at 12:58 PM Tomas Vondra
wrote
ason for this, or is this something worth improving in the
planner?
Sincerely,
Bakhtiyar
On Wed, Jun 21, 2023 at 6:28 AM Tomas Vondra
wrote:
>
> On 6/21/23 05:37, Bəxtiyar Neyman wrote:
> > I define a table user_ranks as such:
> >
> > CREATE TABLE user_ranks (
> > id I
I define a table user_ranks as such:
CREATE TABLE user_ranks (
id INTEGER GENERATED ALWAYS AS IDENTITY PRIMARY KEY,
rank INTEGER NOT NULL,
CONSTRAINT "by (rank, id)" UNIQUE (rank, id)
);
INSERT INTO user_ranks (user_id, rank) SELECT generate_series(1, 1),
generate_series(1, 1);
Her
