BUG #1636: Foreign key referencing inherited table fails.

From: "Fredrik Olsson" <peylow(at)atari(dot)org>
To: pgsql-bugs(at)postgresql(dot)org
Subject: BUG #1636: Foreign key referencing inherited table fails.
Date: 2005-04-28 15:31:45
Message-ID: 20050428153145.9424EF1381@svr2.postgresql.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs


The following bug has been logged online:

Bug reference: 1636
Logged by: Fredrik Olsson
Email address: peylow(at)atari(dot)org
PostgreSQL version: 8.0.2
Operating system: Max OS X 10.3.9
Description: Foreign key referencing inherited table fails.
Details:

Some SQL statements says more then 1000 words :).

CREATE TABLE employees (
nr serial PRIMARY KEY,
name varchar(32)
);

CREATE TABLE managers (
loves_meetings boolean DEFAULT TRUE NOT NULL
) INHERITS (employees);

CREATE TABLE events (
summoner integer NOT NULL REFERENCES employees (nr),
at timestamp DEFAULT now() NOT NULL,
reason text NOT NULL,
PRIMARY KEY (summoner, at)
);

INSERT INTO employees (name) VALUES ('Bob');
INSERT INTO managers (name) VALUES ('Alice');

-- This makes an ugly assumption of nr starting at 1 and Alice having nr=2.

INSERT INTO events (summoner, reason) VALUES (2, 'Brag about
inheritance.');

The last statement will fail with this error message:
ERROR: insert or update on table "events" violates foreign key constraint
"events_summoner_fkey"
DETAIL: Key (summoner)=(2) is not present in table "employees".

The foreign key from events (summoner) referencing employees (nr) can be
removed and replaced by triggers with ease. But I still think it is a bug,
and that foreign key references to tables that are inherited from should be
legal, even if referenced rows orginates from child tables.

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Ingolf Knopf 2005-04-28 15:48:32 Patch for Bug#1611
Previous Message Bruno Wolff III 2005-04-28 01:32:28 Re: BUG #1633: about transactions and row level locking