On Tue, Feb 04, 2003 at 11:24:03AM -0500, Tom Lane wrote:
> *not* necessarily give the same plan, although it may be close enough
> to give you a clue about major problems like needing to add an index.)
Yes, this was what I meant. Sorry, I should have been clearer. What
I find frequently, actually, is that it more often makes you notice
something about the distribution of data. The cases which really
kill you with FKs are (a) there is contention on one of the tables,
so that you end up with everything waiting for their turn at the lock
and (b) referenced tables which are really big but have a small
number of values. Case (b) is interesting, because it's possible to
(mis)design a system which never encounters the symptom directly, and
only runs into it with a trigger.
> As of CVS tip there's finally a real solution: you can PREPARE the
> parameterized query and then EXPLAIN EXECUTE it. For example,
Oh, _nice_. I look forward to that.
A
--
----
Andrew Sullivan 204-4141 Yonge Street
Liberty RMS Toronto, Ontario Canada
<andrew@libertyrms.info> M2P 2A8
+1 416 646 3304 x110