In article <20050426203938.GA18628@xxxxxx>,
Stephane Bortzmeyer <bortzmeyer@xxxxxx> wrote:
% But it does not apply to primary keys containing a group of
% columns. In that case (my case), columns do not have to be UNIQUE. But
% they have to be NOT NULL, which puzzles me.
It does apply to primary keys containing groups of columns.
You can get the table definition you want by using a unique constraint,
but you should know that in SQL, unique constraints don't apply to
rows containing null values in the constrained columns. If you
do this:
create table x (
name TEXT NOT NULL,
address INET,
CONSTRAINT na UNIQUE (name, address)
);
your table definition will be as you want it, but the constraint you
want won't be there.
$ INSERT INTO x VALUES ('alpha');
INSERT 194224 1
$ INSERT INTO x VALUES ('alpha');
INSERT 194225 1
$ INSERT INTO x VALUES ('alpha');
INSERT 194226 1
$ INSERT INTO x VALUES ('alpha');
INSERT 194227 1
--
Patrick TJ McPhee
North York Canada
ptjm@xxxxxxxxxxxx
---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match
