On Fri, 2003-08-01 at 11:15, Tom Lane wrote:
> Ron Johnson <ron.l.johnson@cox.net> writes:
> >> I'd think that in most cases, the extra time spent checking to see
> >> whether the updated columns didn't change would be a net loss.
>
> > Would it always be a net loss, though?
>
> You're asking the wrong question. From my perspective, the question
> is whether it'd be a net win averaged across all UPDATEs at all
> installations everywhere. I can't believe that it would be.
>
> > CPUs are so fast, nowadays. How many microseconds *would* be spent?
>
> That's been a standard excuse for bad design for decades now :-(.
Very true! How is it bad design to try and save an IO, though?
> Yeah,
> the comparisons might be cheap (or not, on some datatypes) ... but the
> potentially-avoided computation is reduced by a faster CPU as well.
But we don't know. MS, IBM or Oracle have the resources to do
that kind of analysis. We don't.
> If you have a particular application and table where no-op UPDATEs occur
> often enough that it's really a win to suppress them, you can put in a
> trigger to do it. Or better, fix the application to not issue the
> UPDATE in the first place; that saves way more computation for the same
> basic comparison overhead.
Which is what I also said...
<QUOTE>
Of course, one could always say, "Hey, application! Don't update
unchanged values!!!!".
</QUOTE>
--
+-----------------------------------------------------------------+
| Ron Johnson, Jr. Home: ron.l.johnson@cox.net |
| Jefferson, LA USA |
| |
| "I'm not a vegetarian because I love animals, I'm a vegetarian |
| because I hate vegetables!" |
| unknown |
+-----------------------------------------------------------------+
