On Wed, 2005-09-21 at 19:46 -0500, A. Banks wrote:
> I have the following related tables:
>
> PEOPLE
> --------------
> peopleid pkey,
> name,
> etc
>
>
> GROUPS
> -------------
> groupid pkey,
> description,
> etc
>
>
> PEOPLEGROUPS
> -------------------
> peopleid pkey/fkey,
> groupid pkey/fkey
>
>
> What is the CORRECT way (in Postgres) to define the PEOPLEGROUPS table so
> that it has both the double primary key AND still acts as a foreign key for
> people.peopleid and groups.groupid? Can i specify both or is this not
> necessary? Any suggestions would be appreciated.
Your terminology is a bit wrong; there can only ever be one primary key
for a table; what you want is probably a single primary key made up of
two columns.
Your definition will look like this:
CREATE TABLE people (
peopleid SERIAL PRIMARY KEY, -- SERIAL is an INTEGER that autoincrements
-- if no value is supplied on insertion
name TEXT NOT NULL,
...
);
CREATE TABLE groups (
groupid SERIAL PRIMARY KEY,
name TEXT NOT NULL,
...
);
CREATE TABLE peoplegroups (
peopleid INTEGER
REFERENCES people (peopleid)
ON UPDATE CASCADE ON DELETE NO ACTION,
groupid INTEGER
REFERENCES people (peopleid)
ON UPDATE CASCADE ON DELETE NO ACTION,
PRIMARY KEY (peopleid, groupid)
);
--
Oliver Elphick olly@lfix.co.uk
Isle of Wight http://www.lfix.co.uk/oliver
GPG: 1024D/A54310EA 92C8 39E7 280E 3631 3F0E 1EC0 5664 7A2F A543 10EA
========================================
Do you want to know God? http://www.lfix.co.uk/knowing_god.html