On Mon, 2008-05-26 at 21:10 +0300, Hannu Krosing wrote:
> Either overlaps or contains seem valid join operators
>
"Overlaps" is the join operator as defined in [1]. In the case where you
have a temporal FK, they are the same. But yes, either is useful.
> Will there be UNIQUE (non-overlapping) indexes for period ?
>
I discussed this with several people and I believe this is possible with
some modifications to GiST. I will be discussing it further and I will
try to put together a proposal.
Regards,Jeff Davis
1. _Temporal Data and the Relational Model_, C.J. Date, Hugh Darwen,
Nikos A. Lorentzos, p. 149.