Knut Suebert <knut.suebert@web.de> writes:
> I thought that a table-constraint unique (a,b) in a table like
> a | b
> --+--
> 1 | 1
> 2 | 2
> 3 | 3
> would allow an insert of (1,2) as it is an unique combination. That seems to be wrong.
It works fine for me:
regression=# create table ff (a int, b int, unique(a,b));
NOTICE: CREATE TABLE/UNIQUE will create implicit index 'ff_a_key' for table 'ff'
CREATE
regression=# insert into ff values(1,1);
INSERT 1401043 1
regression=# insert into ff values(2,2);
INSERT 1401044 1
regression=# insert into ff values(3,3);
INSERT 1401045 1
regression=# insert into ff values(1,2);
INSERT 1401046 1
regression=# select * from ff;
a | b
---+---
1 | 1
2 | 2
3 | 3
1 | 2
(4 rows)
regression=# insert into ff values(1,2);
ERROR: Cannot insert a duplicate key into unique index ff_a_key
If you're having a problem you'll need to be more specific.
regards, tom lane
