Re: Referential integrity: broken rule?

Yes, I believe it is mentioned in the docs (well at least the
sgml source, I don't have a compiled doc set) as a bug
of the implementation somewhere in the section on references.
7.1 should fail for this case, although it doesn't properly deal with
dropping the unique constraint later.

Stephan Szabo
sszabo(at)bigpanda(dot)com

On Wed, 11 Oct 2000, Franz J Fortuny wrote:

> This table:
>
> create table things
> (
> idthing integer not null,
> isthis boolean not null
> primary key(idthing,isthis)
> ....
>
> )
>
> would be referenced by this one:
>
> create table forthings
> (
> fromthing integer not null references things(idthing),
>
> ....
>
> )
>
> The above SHOULD NOT be accepted, since table "things" did not declare
> idthing as UNIQUE. However, it IS accepted under PostgreSQL (7.0.2).