Re: Identifying an Identifying Relationship

John Dilley wrote:
> "Jan Hidders" <hidders_at_REMOVE.THIS.win.tue.nl> wrote in message
> news:9mjg11$7a7$1_at_news.tue.nl...
> >
> > That is actually not a very good example. A relationship between A and
> > B is only an identifying relationship if two (!) things hold:
> > 1. Bs cannot exist without an associated A
> > 2. Bs can only be identified through their relationship with As
> >
> This definition really clears it up for me. Is there a good source for this
> kind of information. My copy of SQL for Smarties, and my way out of date
> Date books (1977) don't mention identifying relationships.

Well, a short explanation can be found on:

  http://www.gc.maricopa.edu/business/sylvester/cis164/er2b.htm

If you look in Date and other literature the concept is usually known as "ID-dependent entity" and sometimes also as "weak entity". (In the sixth edition that's what Date called it, I believe.) Some authors define these terms as synonymous, others make a distinction.

See for example:

  http://gizmo.lbl.gov/DM_TOOLS/OPM/ER/node2.html   http://www.cs.jcu.edu.au/ftp/web/teaching/Subjects/cp3020/2/node6.html   

> This leads to a personal nitpick for me. It would seem to me that the number
> of relationships that meet both 1 and 2 above is somewhat limited.

It depends upon the domain you are modeling, of course, but in general I think I would agree with that.

> IMHO, most relationships would be non-identifying. That being the
> case, why are the rare identifying relationships drawn with a solid
> line, (easy to draw and for the eye to follow) and the common
> non-identifying relationships drawn with a dashed line (more
> difficult to draw and follow with the eye).

I agree that the dashed lines are a bad idea, but the basic idea that identifying relationships are somehow more important than others is not that illogical. In most cases these relationships are 'part-of' relationships. Remember the kidney? :-) So the entities connected by these relationships often form a small hierarchy of things that constitute a whole. So it can be useful to indicate that these things somehow are more related to each other than to the other entities.

But like I said, as far as I am concerned the dashed lines are simply a mistake. There's no other word for it.

> When I use ERWIN to model
> the database for management, they wonder why I keep using dashes. And
> since apparently there is no way in ERWIN to prevent it from
> appending keys, I live with and explain the ugly lines every time
> somebody new looks at the chart. Like I said, a nitpick.

I sympathize. :-}

-- 
  Jan Hidders