Feedback-ID: ia2694551:Fastmail
Date: Wed, 12 Mar 2025 17:23:25 +0100
From: =?utf-8?Q?=C3=81lvaro?= Herrera <alvherre@alvh.no-ip.org>
To: Nathan Bossart <nathandbossart@gmail.com>
Cc: pgsql-hackers@postgresql.org
Subject: Re: remove open-coded popcount in acl.c
Message-ID: <202503121623.i4enp6airsfl@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <Z9GtL7Nm6hsYyJnF@nathan>
Archived-At: 
 <https://www.postgresql.org/message-id/202503121623.i4enp6airsfl%40alvherre.pgsql>
Precedence: bulk

On 2025-Mar-12, Nathan Bossart wrote:

> There's a count_one_bits() function in acl.c that can be replaced with a
> call to pg_popcount64().  This isn't performance-critical code, but IMHO we
> might as well use the centralized implementation.

Makes sense.  Patch looks good to me.

> @@ -5532,7 +5514,7 @@ select_best_grantor(Oid roleId, AclMode privileges,
>  		 */
>  		if (otherprivs != ACL_NO_RIGHTS)
>  		{
> -			int			nnewrights = count_one_bits(otherprivs);
> +			int			nnewrights = pg_popcount64(otherprivs);

Strange: this code is not covered by any tests.

https://coverage.postgresql.org/src/backend/utils/adt/acl.c.gcov.html#5533
https://coverage.postgresql.org/src/backend/utils/adt/acl.c.gcov.html#5438

-- 
Álvaro Herrera        Breisgau, Deutschland  —  https://www.EnterpriseDB.com/