Re: Performance problems testing with Spamassassin 3.1.0

Josh Berkus wrote:

>Dennis,
>
>
>
>>      EXCEPTION WHEN unique_violation THEN
>>
>>
>
>I seem to remember that catching an exception in a PL/pgSQL procedure was a
>large performance cost.   It'd be better to do UPDATE ... IF NOT FOUND.
>
>
>
Actually, he was doing an implicit UPDATE IF NOT FOUND in that he was doing:

UPDATE

IF found THEN return;

INSERT
EXCEPT
...

So really, the exception should never be triggered.
John
=:->