On Wed, Nov 26, 2003 at 10:11:20PM -0800, ow wrote:
> --- Alvaro Herrera <alvherre@dcc.uchile.cl> wrote:
> > On Thu, Nov 27, 2003 at 12:40:28AM +0100, Andreas Pflug wrote:
> >
> > > A common mistake, can't count how often I created this one... And not
> > > easy to find, because EXPLAIN won't explain triggers.
> >
> > That's a pity. And the lack of EXPLAINing function execution, too.
> > Maybe it's not that hard to do?
>
> I'd like to emphasize again that NOT having an index on the FK column is a
> perfectly valid approach, despite some opinions to the contrary.
In what scenarios? I'd easily buy this if you are talking about small
tables.
> Also, FK column index DOES NOT, in general, solve performance issues
> with FK verification check. Someone may (and, I'm sure, will) simply
> have more data or more constraints.
More data? Hmm ... if you have a lot of data in the referenced table,
you'd better _have_ an index unless you want a lot of seqscans.
--
Alvaro Herrera (<alvherre[a]dcc.uchile.cl>)
"La tristeza es un muro entre dos jardines" (Khalil Gibran)