On Mon, 16 Jul 2001, Bryan Buchanan wrote: > Hi, > > Would anyone know if it's possible to define a rule to do the folowing. Why use a rule and not a trigger? > I have two tables: > > journal_master (serial #, account, date, debit $, credit $) > journal_summary (account, date, debit $, credit $) > > each time a transaction occurs, a new journal_master row is > added, and the cumulative amounts added to the summary table. > > I can have a rule like > > CREATE RULE js_update AS > ON INSERT TO journal_master > DO > UPDATE journal_summary set debit = debit + new.debit, > set credit = credit + new.credit where account = new.account and > date = new.date; > > My problem is that if the journal_summary row doesn't exist, I > really want to an INSERT. In pseudo code, something like: > > CREATE RULE js_update AS > ON INSERT TO journal_master > DO > if journal summary row exists for this date and account then > UPDATE journal_summary set debit = debit + new.debit, > set credit = credit + new.credit where account = new.account and > date = new.date > else > INSERT INTO journal_summary (account, date, debit, credit) > VALUES (new.account, new.date, new.debit, new.credit) > end > > Is it possible ? I'd do a trigger in plpgsql. The procedure body would more or less look like: table_row RECORD; SELECT INTO table_row * from journal_summary where account=NEW.account and date=NEW.date; IF NOT FOUND THEN //do an INSERT to journal_summary ELSE //do the UPDATE to journal_summary END IF; Read the online doc for the plpgsql description. cheers, thalis > > Thanks, > > Bryan > > > ---------------------------(end of broadcast)--------------------------- > TIP 1: subscribe and unsubscribe commands go to [EMAIL PROTECTED] > ---------------------------(end of broadcast)--------------------------- TIP 5: Have you checked our extensive FAQ? http://www.postgresql.org/users-lounge/docs/faq.html