Re: [PG-11] Potential bug related to INCLUDE clause of CREATE INDEX

Yugo Nagata <nagata@sraoss.co.jp>

From: Yugo Nagata <nagata@sraoss.co.jp>
To: Yugo Nagata <nagata@sraoss.co.jp>
Cc: Andrey Borodin <x4mmm@yandex-team.ru>, Tom Lane <tgl@sss.pgh.pa.us>, Aditya Toshniwal <aditya.toshniwal@enterprisedb.com>, pgsql-hackers@postgresql.org
Date: 2018-07-12T07:20:13Z
Lists: pgsql-hackers
On Thu, 12 Jul 2018 15:58:08 +0900
Yugo Nagata <nagata@sraoss.co.jp> wrote:
 
> Yes, more simplly, the following query also works;
> 
>  CREATE INDEX ON test((i)) INCLUDE (i);
> 
> However, a problem is that when we use pg_dump for the database, this generate the following query
> 
>  CREATE INDEX test_i_i1_idx ON public.test USING btree (i) INCLUDE (i);
> 
> Of cause, this causes the "must not intersect" error, and we cannot restore this dump.
> 
> To fix this, we agree with Tom about getting rid of "must not intersect" restriction.
> A patch is attached for this

Should we add this to PG11 open items?




-- 
Yugo Nagata <nagata@sraoss.co.jp>


Commits

  1. Drop the rule against included index columns duplicating key columns.