Update a table with a trigger after update

Question

I have two tables

 batch (batch_id,start_date,end_date,batch_strength,is_locked)
 sem (user_id,is_active,no_of_days)

I have executed the tr

Answer 1

There are several ways to prevent the infinite recursion you built into your trigger, the most elegant and performant probably adding a WHERE clause to the UPDATE statement in your trigger function:

CREATE OR REPLACE FUNCTION em_batch_update()
  RETURNS trigger AS
$func$
BEGIN

UPDATE batch b
SET    is_locked = TRUE
FROM   sem s
WHERE  s.is_active
AND    s.user_id = 'OSEM'
AND    b.start_date <= (current_date - s.no_of_days)
AND    b.is_locked IS DISTINCT FROM TRUE; -- prevent infinite recursion!

RETURN NULL;

END
$func$  LANGUAGE plpgsql;

CREATE TRIGGER em_sem_batch
BEFORE UPDATE ON batch
FOR EACH STATEMENT
EXECUTE PROCEDURE em_batch_update();

I changed a few other things to move towards sanity:

• Since the trigger function does the same for every row, I changed it into a potentially much cheaper statement-level trigger.

• Consequently, I made the trigger function RETURN NULL, because, I quote the manual here:

Trigger functions invoked by per-statement triggers should always return NULL.

• batch.is_locked and sem.is_active look like boolean columns. Use a proper boolean data type for them. My code is building on it.

• I also rewrote your UPDATE query completely. In particular the condition on batch.start_date so that an index can be used if available.

• If batch.is_locked is defined NOT NULL, the WHERE condition can be simplified to:

      AND    b.is_locked = FALSE;