Re: no partition pruning when partitioning using array type
Amit Langote <langote_amit_f8@lab.ntt.co.jp>
From: Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>
To: Alvaro Herrera <alvherre@2ndquadrant.com>
Cc: Robert Haas <robertmhaas@gmail.com>,
Pg Hackers <pgsql-hackers@postgresql.org>
Date: 2018-07-11T00:41:46Z
Lists: pgsql-hackers
On 2018/07/11 4:48, Alvaro Herrera wrote:
> On 2018-Jul-10, Alvaro Herrera wrote:
>
>> alvherre=# explain update p set a = a || a where a = '{1}';
>> QUERY PLAN
>> ──────────────────────────────────────────────────────────
>> Update on p (cost=0.00..54.03 rows=14 width=38)
>> Update on p1
>> Update on p2
>> -> Seq Scan on p1 (cost=0.00..27.02 rows=7 width=38)
>> Filter: (a = '{1}'::integer[])
>> -> Seq Scan on p2 (cost=0.00..27.02 rows=7 width=38)
>> Filter: (a = '{1}'::integer[])
>> (7 filas)
>>
>> Because UPDATE uses the predtest.c prune code, not partprune. So it's
>> not just some ruleutils beautification.
>
> I added this test, modified some comments, and pushed.
>
> Thanks for the patch.
Thank you for committing.
Regards,
Amit
Commits
-
Better handle pseudotypes as partition keys
- b6e3a3a492db 12.0 landed
- e7df94f317c9 11.0 landed
- 7c644b7d3f83 10.5 landed