Re: candidate keys in abstract parent relations

From: paul c <toledobythesea_at_oohay.ac>
Date: Fri, 20 Jan 2006 23:41:11 GMT
Message-ID: <ryeAf.295927$2k.205270@pd7tw1no>

David Cressey wrote:

> "Forrest L Norvell" <spankysyourpal_at_gmail.com> wrote in message
> news:1137705821.754921.280830_at_g47g2000cwa.googlegroups.com...
> 

>>Roy Hann wrote:
>>
>>>(Names are a bad example for us to use here because they are a poor key
>>>anyway, but let's ignore that.)
>>
>>If names are a poor key, and candidate keys are tied to identity, what
>>are good keys for a named entity (like a person)?
>>

> 
> 
> Funny you should ask.  There aren't any good natural keys for a person.

Putting words in your mouth, can't resist, sorry. To amplify, I would say a person can't have a key even though a relation about persons could.

I'd be interested to hear where the philosophers or mathematicians have been able to define 'identity'. If they can't, I don't see how a computer can. Outside of narrow context, it seems to me to be an indefinable. In fact, in most systems it's not even demonstrable.

Safeway here charges 50% more, no exaggeration if you aren't a 'member'.   Need to give your telephone number. They won't accept mine because I share it with friend who has already given it to them. Since friend is female and I'm not, I can't fool them. On the other hand they can't fool me, which is why I don't shop there.

Sometimes I think modellers go too far trying to ascribe dbms attributes to the natural world. Certainly big biz and governments do.

I do agree with whoever posted to the effect that it's good to avoid showing internal or 'artificial' keys even though it's sometimes necessary to subsequent transactions, some of the more common examples being social security numbers and postal codes.

p Received on Fri Jan 20 2006 - 17:41:11 CST