BUG #1514: Inheritance and Primary Keys - Mailing list pgsql-bugs

From Peter Newman
Subject BUG #1514: Inheritance and Primary Keys
Date
Msg-id 20050302152327.A7047F0F70@svr2.postgresql.org
Whole thread Raw
Responses Re: BUG #1514: Inheritance and Primary Keys  (Michael Fuhr <mike@fuhr.org>)
List pgsql-bugs
The following bug has been logged online:

Bug reference:      1514
Logged by:          Peter Newman
Email address:      pnewman@wanadoo.nl
PostgreSQL version: 7.4.7
Operating system:   Fedora Core 3
Description:        Inheritance and Primary Keys
Details:

Hi

Just exploring the use of inheritance to so some design problems and found
some strange behaviour.  Not sure if this is a bug or feature?

From
http://www.postgresql.org/docs/8.0/interactive/tutorial-inheritance.html

create table cities (id serial primary key, name varchar(45));
create table capitals (state char(2)) inherits (cities);

insert into cities values (1, 'Paris');
insert into cities values (1, 'London');

The second one fails naturally however if you were to
insert into capitals values (1, 'Paris', 'FR');
insert into capitals values (1, 'London', 'UK');

This is fine.  And if
create table capitals (state char(2), primary key (id)) inherits (cities);
create table non_capitals (state char(2), primary key (id)) inherits
(cities);
then
insert into capitals values (1, 'Paris', 'FR');
insert into non_capitals values (1, 'London', 'UK');

is also okay.

If you were to insert data like this:
insert into capitals (name, state) values ('Paris', 'FR');
insert into non_capitals (name, state) values ('London', 'UK');

then the ids in cities would be 1 and 2 using the serial sequencing.  This
seems inconsistent at least.

So my point is (if it is not clear) that the primary key constraint is not
enforced through subclass tables.  Is this intentional?

Cheers,
Pete.

pgsql-bugs by date:

Previous
From: Sergio Lob
Date:
Subject: Re: BUG #1523: precision column value returned from getTypeInfo()
Next
From: "Roy Badami"
Date:
Subject: BUG #1517: SQL interval syntax is accepted by the parser, but the interpretation is bogus