> I have 2 tables and I must make asure unique values like this. > > table1 > id > nr - integer > > table2 > id > id_table1 - FK in Table 1 > valid_from - timestamp > > There must be unique values for: > - nr - from table1 > and > - YEAR(MIN(valid_from)) from table 2 In situations like this, I add the required column to the other table with a foreign key. Assuming that (id) is the primary key of table1, you would need another unique key on (nr, id). Than you can add nr column to table2 by changing the foreign key to (nr, id_table1) references table1 (nr, id). Obviously, its not an efficient solution. It requires an additional unique key and more storage on the referencing table. Though, I believe it is a safe one. It doesn't allow the duplicated column to be inconsistent. There are many things that can go wrong under concurrency with a trigger like you posted. -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
