bubba postgres <bubba.postgres@xxxxxxxxx> writes: > This is the reverse of what I thought I would find. > In short my check constraint is extracting the epoch from a start timestamp, > and an end timestamp to get the number of seconds difference. > It then uses this number to check the array_upper() of an array to make sure > it's the proper size > The SQL version uses a case statement, and the plpgsql uses an IF/ELSE > In a particular insert test > The plpgsql version adds 1 second over the no constraints case. > the sql version adds 10 seconds over the no constraints case. > Why would this be? It would not likely be faster unless it can be inlined, and maybe not even then, because of plan caching effects (plpgsql is a lot better about that). In this particular case, I'm suspicious whether all the operations are immutable; if they aren't, the marking of the function as immutable will definitely prevent inlining. regards, tom lane -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
