Content-Type: text/plain;
	charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 16.0 \(3776.700.51.11.2\))
Subject: Re: [PATCH] initdb: Treat empty -U argument as unset username
From: Daniel Gustafsson <daniel@yesql.se>
In-Reply-To: 
 <CAAZLFmSY_s6TOsiTwjzqrc2Y4uShATuWcEEN8O8cdtpkS=zfLg@mail.gmail.com>
Date: Wed, 2 Jul 2025 16:01:26 +0200
Cc: Robert Treat <rob@xzilla.net>,
 "David G. Johnston" <david.g.johnston@gmail.com>,
 pgsql-hackers@lists.postgresql.org
Content-Transfer-Encoding: quoted-printable
Message-Id: <CEDBFBDE-2BBA-4568-8B9B-42AD74E377EA@yesql.se>
References: 
 <CAAZLFmRK+XFp=mqCeruyNVkqGq5mH45CP+e-8oNttPRtLuB5eQ@mail.gmail.com>
 <CAKFQuwbVK=oANp9XLdQknd7o9jJOUFQanUtG3QAh-3kuaM5tJw@mail.gmail.com>
 <CAAZLFmSKrcovMSauueWg=8VdZN-fHPH2ZengZRKA8AfgST6o6w@mail.gmail.com>
 <CAKFQuwaCPkb8=2_V3121Ei0MBrg02FLRNQbcAcyzCO3dAha=Gw@mail.gmail.com>
 <CABV9wwOOpOijL9sY03KV3WNvgMxVFsyfvfYbfy2MYfn40z0yyQ@mail.gmail.com>
 <0231C713-50C8-4BE5-B7C8-106F57A79655@yesql.se>
 <CAAZLFmSY_s6TOsiTwjzqrc2Y4uShATuWcEEN8O8cdtpkS=zfLg@mail.gmail.com>
To: Jianghua Yang <yjhjstz@gmail.com>
Archived-At: 
 <https://www.postgresql.org/message-id/CEDBFBDE-2BBA-4568-8B9B-42AD74E377EA%40yesql.se>
Precedence: bulk

> On 2 Jul 2025, at 15:52, Jianghua Yang <yjhjstz@gmail.com> wrote:
>=20
> Hi hackers,
>=20
> Based on the suggestion that we should explicitly reject empty =
usernames instead of silently falling back, I=E2=80=99ve updated the =
patch accordingly.
>=20
> ### Changes in v2:
>=20
> - `initdb` now errors out immediately if the `-U` or `--username` =
argument is an empty string.
> - The error message is:
>  =20
>       superuser name must not be empty
>=20
> - A regression test is added to `src/bin/initdb/t/001_initdb.pl` to =
verify that the case `initdb -U ''` fails as expected.
>=20
> This approach avoids any ambiguity about whether an empty username is =
valid, and fails early with a clear message. It also brings consistency =
with existing checks, such as the one disallowing superuser names =
starting with `pg_`.
>=20
> Let me know if this looks acceptable or if further refinement is =
needed.

+	pg_log_error("superuser name must not be empty");
+	exit(1);

I would prefer pg_fatal() for this.

--
Daniel Gustafsson