On Tue, Dec 29, 2009 at 4:44 PM, Bryce Nesbitt <bryce2@obviously.com> wrote:
> Craig is correct in the OP attempt. All but one field is intended to be null
Ah; my misunderstanding, then. Sorry for the noise.
> Duplicate rows, or data in multiple columns, would wreck havoc on the
> scheme. If there is a better way, I am all eyes.
One thought would be to add a couple of constraints: a CHECK
constraint to ensure that one and only one of the columns is populated
(think "t_number IS NOT NULL AND t_string IS NULL AND t_boolean IS
NULL AND t_date IS NULL OR..."), and a UNIQUE constraint against
(context_key, t_number, t_string, t_boolean, t_date). That's not
really a "better" way, but it should prevent both duplicate and
"compound" data.
rls
--
:wq