Very Computer

1. MATCH FULL flawed?

You're not using it right.  You specified a separate MATCH FULL
constraint for each column:

This is essentially a no-op, since MATCH FULL and MATCH PARTIAL are only
meaningfully different for a multi-column key.  What you want is

regression=# create table test2 (a integer, b integer,
regression(# foreign key (a,b) references test(a,b) match full);
NOTICE:  CREATE TABLE will create implicit trigger(s) for FOREIGN KEY check(s)

which has the behavior you're after:

regression=# insert into test2 values (null, 1);
ERROR:  <unnamed> referential integrity violation - MATCH FULL doesn't allow mixing of NULL and NON-NULL key values
regression=# insert into test2 values (1, null);
ERROR:  <unnamed> referential integrity violation - MATCH FULL doesn't allow mixing of NULL and NON-NULL key values
regression=# insert into test2 values (null, null);
INSERT 262789 1
regression=# insert into test2 values (1, 1);
INSERT 262790 1
regression=#

                        regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

http://www.postgresql.org/users-lounge/docs/faq.html

2. ontape verification?

3. Flawed Pentium Chip

4. LINUX/sco AND oh yes Unixware

5. Interbase Fatally Flawed?

6. How do I extract data without using OLEDB or ODBC?

7. flawed FNKSHNs in version 1.01

8. Unix Databases?

9. Oracle Objects in O.R.DBs flawed to uselessness

10. importing slightly flawed files

11. Flawed Design of Allowing Same Table Name of Different Owner

12. KEYBOARD Command flawed? [2.0 DOS]

13. Verifying/Restoring flawed Online v4 archives