Re: Update on tables when the row doesn't change

Tom Lane wrote:
> =?ISO-8859-1?Q?Sebastian_B=F6ck?= <sebastianboeck(at)freenet(dot)de> writes:
>
>>Why does Postgres perform updates to tables, even if the row doesn't
>>change at all?
>
>
> Because testing for this would almost surely be a net loss for the vast
> majority of applications. Checking to see if the new row value exactly
> equals the old is hardly a zero-cost operation; if you pay that on every
> update, that's a lot of overhead that you are hoping to make back by
> sometimes avoiding the physical store of the new tuple. In most
> applications I think the "sometimes" isn't going to be often enough
> to justify doing it.
>
> If you have a particular table in a particular app where it is worth it,
> I'd recommend writing a BEFORE UPDATE trigger to make the comparisons
> and suppress the update when NEW and OLD are equal.

In any case, what if I have a trigger that's supposed to increment a
counter or similar if issue a supposedly "unneeded" update.

--
Richard Huxton
Archonet Ltd