On Mon, 27 Sep 2004, Aleksey Fedorchenko wrote:
> > And in current sources, functions 2 and 3 error with foreign key
> > violations rather than unique constraint violations. If the constraint
> > was deferred you'd still get the unique violations, but we're still
> > looking to see which error is correct in that case.
>
> Does it mean that scripts that i sent you yesterday works fine and their
> results are proper (in mean of proper reaction on improper user
> activities) and different from described by me in first letter?
>
> Its strange because i tuned SET CONSTRAINTS in different modes but it
> didn't take any effect - i still have same results!
>
> Did you apply any patches on your pgsql instance? Because i have 3
> distribs (7.4.2/8.0.0b1/8.0.0b2) and all fails the same way.
IIRC, b2 was before Tom changed the code for doing triggers between
function statements, current sources have that code. So, b3 should have
the change I mentioned which makes the above work (for immediate
constraints).