On Mon, 25 Jun 2001, [iso-8859-1] Daniel �kerud wrote:
>
> OK, I've been discussing this with a collegue of mine... and I'm starting to
> see the light here...
>
> I will, first of all, make a new, simpler, 1<->1 realtionship to test FK
> constraints... no 2<->1<->2 relasionship here...
>
> Person -> Item/item_fkc
>
> And I will do no bulk-delete. Instead these tests:
>
> Fill person. no time measuring.
> Fill item, no foreign keys constraints, time measuring.
> Fill item, foreign keys constraints
> Compare last to measurments. How much do you loose in performance having the
> Foreign Key check?
> For all persons {
> if (fkc)
> delete from person where id = $id
> else
> delete from person where id = $id; delete from item where personid=$id
> }
> Compare measurements. How much do you loose having the foreign keys
> constraints delete the item?
>
> This, I think, this is a more fair comparison.
>
> Can you call the FK itself a foreign key constraint, as it actually is
> constraining something?
Yeah, that's a more even test. Since it sounds like you're writing
something on foreign key constraints (presumably about postgres), you
might want to mention the differences with table clearing deletes
but point out the foreign key constraints aren't generally used in
schemas where you're doing that :)
As for naming, I'd think so. It is a constraint, just that
"foreign key constraint" is much longer to type than FK :)
