Re: Purely declarative FKs

On 10/16/23 09:06, Tom Lane wrote:
> Christophe Pettus <xof(at)thebuild(dot)com> writes:
>> On Oct 16, 2023, at 00:51, Thiemo Kellner <thiemo(at)gelassene-pferde(dot)biz> wrote:
>>> Question: Are there plans to provide a feature in PostgreSQL that one
>>> can have foreign keys for purely documentation purpose - I know, one
>>> could use a modelling tool and just not implement the FKs, but my
>>> reality is, there is hardly ever a modelling tool involved.
>> The answer to the specific question is, no, I don't believe that there
>> any plans to implement purely documentary foreign keys.
> Recent versions of the SQL spec have a NOT ENFORCED option for
> constraints, which is what I think you're looking for here.
> If somebody came with a well-thought-out patch to add that feature,
> we'd likely take it. But this is definitely a "scratch your own
> itch" situation; I don't know of anyone already working on such
> a thing.

NOT ENFORCED would be a great boon to administrators writing regular
archive/purge operations.

--
Born in Arizona, moved to Babylonia.