Re: Add privileges test for pg_stat_statements to improve coverage

Fujii Masao <masao.fujii@oss.nttdata.com>

From: Fujii Masao <masao.fujii@oss.nttdata.com>
To: kuroda.keisuke@nttcom.co.jp, Michael Paquier <michael@paquier.xyz>, pgsql-hackers@postgresql.org
Date: 2024-07-23T05:28:14Z
Lists: pgsql-hackers

On 2024/07/23 10:40, kuroda.keisuke@nttcom.co.jp wrote:
> I agree.
> The information of different userids is mixed up.
> It is easier to understand if the role name is displayed.
> Join with pg_roles (view of pg_authid) to output the role name.

+       rolname       | queryid_bool |                       query                        | calls | rows
+---------------------+--------------+----------------------------------------------------+-------+------
+ postgres            | t            | SELECT $1 AS "ONE"                                 |     1 |    1
+ postgres            | t            | SELECT pg_stat_statements_reset() IS NOT NULL AS t |     1 |    1
+ regress_stats_user1 | t            | SELECT $1+$2 AS "TWO"                              |     1 |    1

Using "postgres" as the default superuser name can cause instability.
This is why the Patch Tester reports now test failures again.
You should create and use a different superuser, such as "regress_stats_superuser."

Regards,

-- 
Fujii Masao
Advanced Computing Technology Center
Research and Development Headquarters
NTT DATA CORPORATION



Commits

  1. pg_stat_statements: Add regression test for privilege handling.