On 2017/08/01 10:52, Robert Haas wrote:
> On Mon, Apr 3, 2017 at 6:28 AM, Amit Langote
> <Langote_Amit_f8@lab.ntt.co.jp> wrote:
>> Since nowhere has the user asked to ensure unique(b) across partitions by
>> defining the same on parent, this seems just fine. But one question to
>> ask may be whether that will *always* be the case? That is, will we take
>> ON CONFLICT DO NOTHING without the conflict target specification to mean
>> checking for conflicts on the individual leaf partition level, even in the
>> future when we may have global constraints?
>
> No. We'll take it to mean that there is no conflict with any unique
> constraint we're able to declare. Currently, that means a
> partition-local unique constraint because that's all there is. It
> will include any new things added in the future.
So is the latest patch posted upthread to process ON CONFLICT DO NOTHING
using locally-defined unique indexes on leaf partitions something to consider?
Maybe, not until we have cascading index definition working [1]?
Thanks,
Amit
[1]
https://www.postgresql.org/message-id/c8fe4f6b-ff46-aae0-89e3-e936a35f0cfd%40postgrespro.ru