On Fri, 2002-05-31 at 08:05, David Busby wrote:
>
> The GUID from Micro$oft is formatted like
> {01234567-89AB-CDEF-0123-456789ABCDEF} it represents a 16 byte number that
> is again unique in spacetime. Is there PostgreSQL solution for something
> like that or will I have to come up with my own.
There is an RFC defining GUID and UUID, and it isn't a major job to
generate them yourself. If you implement a function to do so you may
want to contribute it back to the PostgreSQL community.
The basic approach uses IP address, date and time and some other stuff I
forget right now.
GUID's are actually pretty neat, because they are sortable, but still
unique, I believe.
It sounds like a straight timestamp + IP address would be sufficient for
you. In a similar multi-system problem I am working on I actually just
use a two-field key with a serial and a machine name. Using a machine
name explicitly, and splitting it into a separate field, actually offers
the advantage of making it clearer where the transaction came from, and
consequently how my replication model works.
Steer well clear of OIDs - they are not necessarily persistent over
backup / restore, and they offer little real utility in PostgreSQL these
days being something of an appendix from it's time-travelling days. I
actually disable them on most user tables when running under 7.2.1 .
Regards,
Andrew.
--
--------------------------------------------------------------------
Andrew @ Catalyst .Net.NZ Ltd, PO Box 11-053, Manners St, Wellington
WEB: http://catalyst.net.nz/ PHYS: Level 2, 150-154 Willis St
DDI: +64(4)916-7201 MOB: +64(21)635-694 OFFICE: +64(4)499-2267
Are you enrolled at http://schoolreunions.co.nz/ yet?
