On 16/07/11 05:21, Tom Lane wrote:
> Gavin Flower<GavinFlower@archidevsys.co.nz> writes:
>> On 15/07/11 11:25, Gavin Flower wrote:
>>> I think it should be possible to put a constraint on the master table
>>> to prevent rows being inserted.
>> Of course, minutes after I sent the above - I realized these constraints
>> are inherited, so the above is nonsense! :-(
> Yeah. I think there's been some discussion of inventing a non-inherited
> variety of check constraint, so that you could put something like
> "CHECK (false) NO INHERIT" on the parent table. There's not really
> consensus for this though.
>
> IMO we'd be better advised to spend our time on building an explicit
> partitioning mechanism to handle the common cases more simply and
> efficiently, instead of continuing to add frammishes to the inheritance
> mechanism.
>
> regards, tom lane
How about being able to mark individual columns, and optionally the
table itself, as ABSTRACT?
Though, I agree that a proper partitioning mechanism would be better.
Cheers,
Gavin