On 22/06/11 18.30, David Johnston wrote:
> The only (obvious to me) way to really solve the problem - invisibly - is to
> allow for table-less unique indexes that multiple tables can share and that
> have a pointer to the "source" table for any particular entry in the index.
> The other method being discussed effectively uses a physical table to
> implement this behavior.
I Have the same problem: one ID must be unique.
Three tables inherits from the same parent table the id column, the ID
is is defined as:
id bigint DEFAULT nextval('sequence_name')
in the parent table.
Can I assume that a sequence ensures uniqueness?
And... I did some search on Google but can't figure "table-less unique
indexes that multiple tables can share" where can I find further
information regarding this thing?
thank you
Edoardo