> > Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > > I have committed new warning code to alert users who auto-create
> > > relations without knowing it.
> > > The code issues the warning if it auto-creates a range table entry, and
> > > there is already a range table entry identified as coming from a FROM
> > > clause. Correlated subqueries should not be a problem because they are
> > > not auto-created.
> >
> > I still prefer the suggestion I made before: complain only if the
> > implicit FROM entry is for a table already present in the rangelist
> > (under a different alias, obviously). The fact that that choice
> > would not break any existing regression tests seems relevant...
>
> But it seems mine is going to complain if they forget one in a FROM
> clause, which sort of makes sense to me. I can do your suggestion, but
> this makes more sense. Can we get some other votes?
I like it the way you did it. Personally I would even throw an error,
but that would probably be too strict.
I would change the regressiontest to add onek to the from clause,
and not make it throw the warning.
Imho this example is only good to demonstrate how you can
misuse a feature.
There are good examples for using it, but all of those that I can think of
don't have a from clause.
Andreas