On Tue, May 3, 2011 at 8:03 PM, Greg Smith <greg@2ndquadrant.com> wrote:
> With a uniqueness constraint in this situation, the unexpected data--row
> with a non unique MAC--will be rejected and possibly lost when the insertion
> happens. You say that's a good thing, plenty of people will say that's the
> worst possible thing that can happen.
But remember the original discussion is on using these are PK/FK.
That's where things get really ugly. I can change my data model to
not have a unique MAC or to do something to make them unique (add IP
or something) much more easily if they're NOT a PK/FK. That's the
real issue to me.