Re: another simple data modelling question

thanks for everyone's help. i guess i will just have to decide between the lesser of two not-so-good's.

Heinz Huber <hhuber_at_racon-linz.at> wrote in message news:<3B08BE05.1E06CDC1_at_racon-linz.at>...
> Larry Coon wrote:
> >
> > crappy wrote:
> >
> > > thanks for your reply. i understand what you're saying. but i think my post
> > > was confusing now that i read it again.
> > >
> > > the characteristic_personality junction lists what characteristics make sense
> > > for which personalities. the reason for the person_characteristic junction is
> > > to list what characteristics a person *actually* has. i don't necessarily want
> > > to say that a person of such-and-such personality automatically has *all* the
> > > characteristics that make sense for that personality type.
> >

 [snipped]
> >
> > This is off the top of my head, and there's probably a million
> > reasons why this is a bad idea, but I'm not willing to spend
> > the time to think it through, so someone else can critique it.
> >
> > You have a table personality-characteristic, which identifies
> > the valid characteristics for a given personality.
> >
> > You have a table person-personality-characteristic, which
> > represents the ternary relationship between the three entities.
> > For example, ("Joe","Introvert","Quiet") is a valid tuple here.
> > The personality & characteristic are a composite foreign key to
> > the personality-characteristic table, which ensures that Joe
> > has a personality & characteristic combination which is valid.
> >
> > So one table represents the set of combinations which are valid.
> > The other table represents the combinations that a given person
> > actually has. No redundancy. No triggers.
> >
> > I'd also be curious whether both personality and characteristic
> > are really entities, as opposed to one just being an attribute.
> > But since you said the real problem is in an entirely different
> > subject matter, the point is probably moot.
>
> I think (and from an answer of crappy to an equivalent proposal he does too)
> that this might not be completely normalized, since you either need the
> personality in the person table or you have to ensure that a person is only
> related to one personality in the person-personality-characteristic table.
>
> Regards,
> Heinz
Received on Sun Jul 22 2001 - 01:25:43 CEST