From: | Ajit Awekar <ajitpostgres(at)gmail(dot)com> |
---|---|
To: | pgsql-hackers(at)postgresql(dot)org |
Subject: | Unnecessary scan from non-overlapping range predicates |
Date: | 2025-06-25 15:46:38 |
Message-ID: | CAER375MzJRka+WEo--jKAoqbP8E2rSrj5Ko7sW9pH8xKDzdf4A@mail.gmail.com |
Views: | Whole Thread | Raw Message | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Hi Team,
Please find below an observation regarding query planning that leads to
unnecessary table scan when the WHERE clause contains non-overlapping
conditions.
EXPLAIN (costs off)
select * from products where price < 100 AND price > 300;
Seq Scan on products
Filter: ((price < '100'::numeric) AND (price > '300'::numeric))
Since this condition is false and result will always be empty. Despite
this, we still perform unnecessary sequential scan over the table.
Can we detect such contradictory predicates during planning and optimize
them away using a Result node with One-Time Filter: false. This would avoid
scanning large tables unnecessarily and improve performance.
Similarly below non-overlapping ranges in query also lead to unnecessary
scan without any result.
select * from products where (price between 3 and 100) and (price between
900 and 1000);
Thanks
Ajit
From | Date | Subject | |
---|---|---|---|
Next Message | Jim Jones | 2025-06-25 15:50:39 | Re: display hot standby state in psql prompt |
Previous Message | Álvaro Herrera | 2025-06-25 15:45:06 | Re: pg_dump misses comments on NOT NULL constraints |