Re: Declaring Unenforced Constraints

Kenneth Downs wrote:
> Perhaps someday I'll convince you of how much easier that is to
handle when
> the tables are ruthlessly normalized, so that much more checking can
be
> done with key constraints.

I don't need any convincing on that score. Unfortunately, the database design pre-dates my involvement in the project. I suppose I could investigate and perhaps propose a re-design of the tables, but I very much doubt I'd be able to convince the management on that; much easier to persuade them to "merely" add some constraints.

> Yikes. Do you have extensive regression testing now? Unless you can
make
> it completely automatic....

I wouldn't call it extensive, and it is not automated yet (there are plans to do that I understand). Essentially the testers run through some stock scenarios and see if they still work!

> > It is early days yet, and I am wondering whether to suggest we go
for a
> > "big bang" approach where we implement tons of constraints at once,
> > regression test like mad, and then wait for the screams; or a
drip-feed
> > where we implement a smaller number of constraints with each
upgrade,
> > regression test like mad, and hopefully get less issues, spread
over a
> > longer time frame.
>
> In my experience drip-feed will more likely be accepted by the more
people,
> giving you a greater overall good. Losing the fight for big-bang
brings
> zero improvement, but going one-by-one gives you concrete and
permanent
> improvements that will never cause trouble again. I wish you luck.
Thanks - and I think you are right. Received on Fri Nov 19 2004 - 11:29:21 CET