Message-ID: <560b391a-2b3a-4009-a90b-b5e5863aa89d@uni-muenster.de>
Date: Sun, 10 May 2026 13:10:32 +0200
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: Fix wrong error message from pg_get_tablespace_ddl()
To: Chao Li <li.evan.chao@gmail.com>, =?UTF-8?Q?=C3=81lvaro_Herrera?=
 <alvherre@kurilemu.de>, Andrew Dunstan <andrew@dunslane.net>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
References: <af4YmulGI3pq2Rm0@alvherre.pgsql>
 <4B28CBF6-7470-456A-A635-62FE28067AEE@gmail.com>
Content-Language: en-US, de-DE
From: Jim Jones <jim.jones@uni-muenster.de>
In-Reply-To: <4B28CBF6-7470-456A-A635-62FE28067AEE@gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/560b391a-2b3a-4009-a90b-b5e5863aa89d%40uni-muenster.de>
Precedence: bulk

Hi Chao

On 09/05/2026 04:01, Chao Li wrote:
> Álvaro seems to bring the question to a deeper level, and I feel that might be worth a dedicated discussion. For example, I am not sure ACL_CREATE on the tablespace is enough to imply visibility of the tablespace DDL. My understanding is that CREATE on a tablespace allows the user to create objects within that tablespace, but it does not necessarily mean the user is allowed to inspect the definition of the tablespace itself.


Yeah, this is a good point. I don't have a strong opinion about it, but 
I'd be inclined to simply deny access to the DDL if the user does not 
have enough privileges -- at least I wouldn't mind seeing an error 
message in my logs :)


> How about keeping the scope of this patch narrow, as only adding a hint to guide users on how to fix the error if they really need to view the DDL of the tablespace? I will start a separate thread for the discussion of the access-checking model.
> 
> The attached v2 keeps the original error message and adds a hint. I took Jim’s comment about avoiding hardcoding "pg_tablespace”. And I also added a hint in pg_get_role_ddl_internal. With v2, the messages are like:
> ```
> evantest=> select * from pg_get_tablespace_ddl('ts1');
> ERROR:  permission denied for tablespace "ts1"
> HINT:  Grant SELECT on catalog "pg_tablespace" to read tablespace properties.


I'm not sure that telling unprivileged users to grant themselves access 
to pg_tablespace is an improvement -- IMO, a HINT here is supposed to be 
actionable. Perhaps a DETAIL would be a better fit, e.g. "DETAIL:  The 
function requires SELECT privilege on catalog "pg_tablespace"."

On top of that, I'm also not sure that replacing the aclcheck_error with 
an ereport just for the hint/detail is an option, since aclcheck_error 
is supposed to provide "Standardized reporting of aclcheck permissions 
failures." (from the aclcheck_error header comment)

Thanks!

Best, Jim