MIME-Version: 1.0
References: 
 <CACJufxGoAmN_0iJ=hjTG0vGpOSOyy-vYyfE+-q0AWxrq2_p5XQ@mail.gmail.com>
 <CA+renyU=ukksgiMQioaUixG+DJY1w1cwGdssKy9_hj_uaYDkBA@mail.gmail.com>
 <CACJufxEABtUZf=M9h3XpvjiWKFNDxObPpe6Ly9_dd-k3_xJJ+w@mail.gmail.com>
In-Reply-To: 
 <CACJufxEABtUZf=M9h3XpvjiWKFNDxObPpe6Ly9_dd-k3_xJJ+w@mail.gmail.com>
From: Paul A Jungwirth <pj@illuminatedcomputing.com>
Date: Mon, 22 Dec 2025 19:25:46 -0800
Message-ID: 
 <CA+renyW+ia6Uewa5+9f+cgTinndj8wt7yiG=3yZ_FG-hmiHn3w@mail.gmail.com>
Subject: Re: domain for WITHOUT OVERLAPS
To: jian he <jian.universality@gmail.com>
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: 
 <https://www.postgresql.org/message-id/CA%2BrenyW%2Bia6Uewa5%2B9f%2BcgTinndj8wt7yiG%3D3yZ_FG-hmiHn3w%40mail.gmail.com>
Precedence: bulk

On Wed, Dec 10, 2025 at 8:23=E2=80=AFPM jian he <jian.universality@gmail.co=
m> wrote:
> polished tests for
> > - WITHOUT OVERLAPS on a rangetype whose subtype has a domain
> > - WITHOUT OVERLAPS on a rangetype with a domain on itself
> > - WITHOUT OVERLAPS on a multirangetype whose subtype has a domain
> > - WITHOUT OVERLAPS on a multirangetype whose rangetype has a domain
> > - WITHOUT OVERLAPS on a multirangetype with a domain on itself
>
> no tests for
> > - WITHOUT OVERLAPS on a multirangetype whose rangetype has a domain
> now.
>
> because the expected behavior assumes that CREATE DOMAIN also generates a
> corresponding multirange type for the domain range type, which is not tru=
e.
>
> --example, we expect create domain also create a multirange type for xxin=
t4
> CREATE DOMAIN xxint4 AS int4range CHECK (VALUE <> '[10,11)');
> SELECT typname FROM pg_type WHERE typname ~* 'xxint4';

I'm not sure whether creating a domain on a range should create a
multirange or not. I asked the list on another thread. Since it does
not create a new range constructor either (or an entry in pg_range),
maybe not creating a new multirange is correct. Anyway I think we can
fix the WITHOUT OVERLAPS issue without dealing with that.


> diff --git a/src/backend/utils/cache/typcache.c b/src/backend/utils/cache=
/typcache.c
> index 6a347698edf..e27ca1466f2 100644
> --- a/src/backend/utils/cache/typcache.c
> +++ b/src/backend/utils/cache/typcache.c
> @@ -944,6 +944,8 @@ lookup_type_cache(Oid type_id, int flags)
>          typentry->domainBaseTypmod =3D -1;
>          typentry->domainBaseType =3D
>              getBaseTypeAndTypmod(type_id, &typentry->domainBaseTypmod);
> +        typentry->domainBaseTyptype =3D
> +            get_typtype(typentry->domainBaseType);
>      }
>      if ((flags & TYPECACHE_DOMAIN_CONSTR_INFO) &&
>          (typentry->flags & TCFLAGS_CHECKED_DOMAIN_CONSTRAINTS) =3D=3D 0 =
&&

Is there any performance concern about adding this lookup?