RI violated, bug or feature?

From: 100(dot)179370(at)germanynet(dot)de (Martin Jacobs)
To: PGSQL Mailinglist <pgsql-general(at)hub(dot)org>
Subject: RI violated, bug or feature?
Date: 2000-10-14 14:08:13
Message-ID: m13kRyv-000QcWC@Schnecke.Windsbach.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

Hi,

my 7.0.2 runs no stable, all tests except one numerical (least
signifcant digit was off by one) ran successfull. No I did play
with RI and discovered the following:

Lets create two tables:

test_ri=> create table publisher (name char(20));
CREATE

test_ri=> create index publ on publisher (name);
CREATE

test_ri=> create table books (title char(32), publisher char(20)
references publisher(name));
NOTICE: CREATE TABLE will create implicit trigger(s) for FOREIGN KEY check(s)
CREATE

Now let's insert some data:

test_ri=> insert into publisher (name) values ('prentice');
INSERT 35900 1
test_ri=> insert into publisher (name) values ('addison');
INSERT 35901 1
test_ri=> insert into publisher (name) values ('springer');
INSERT 35902 1

and at least

test_ri=> insert into books (title) values ('violated');
INSERT 35903 1

Hey, I have now a row in books, which has no corresponding
publisher entry. Let's do another test:

test_ri=> insert into books (title, publisher) values ('dummy', 'x');
ERROR: <unnamed> referential integrity violation - key referenced from books not found in publisher

Ok, that's what I expected.

test_ri=> insert into books (title, publisher) values ('XML handbook', 'prentice');
INSERT 35905 1

That's ok too.

If I create table books again and add a "DEFAULT ''" clause to
column publisher then inserting rows with empty field publisher
is prohibited.

Is this a bug or is it a feature?

Martin Jacobs

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Vince Vielhaber 2000-10-14 14:31:22 Re: Question about pg_dump
Previous Message Collin Peters 2000-10-14 14:06:10 JDBC Driver homepage down