Re: One-To-One Relationships

Phil Reynolds wrote:
> One thing that's not clear to me is when it's appropriate to create a
> one-to-one relationship. I mean, in some cases it's obvious, if there's a
> set of data that wouldn't always apply; then you'd want to create that set
> of fields in a separate table with a one-to-one relationship.

But that is not a one-to-one relative cardinality. That is actually a one-to-zero_or_one relative cardinality.

  But in what
> other cases? After the number of fields in a table is greater than X?
>
> I'm just curious about what thoughts/theories/ideas people have about
> one-to-one relationships, because that's something that's never been
> entirely clear to me.
>
> Thank you.

I have seen no theory-based arguments for partitioning a relation into two relations with one-to-one relative cardinality. At the same time, I haven't seen any compelling theory-based arguments against doing so either. After all, with join and project one can express either from the other. Received on Tue Oct 30 2007 - 13:21:15 CET