Re: Surrogate Keys: an Implementation Issue

From: paul c <toledobythesea_at_oohay.ac>
Date: Sat, 29 Jul 2006 02:25:32 GMT
Message-ID: <wGzyg.258377$Mn5.187057@pd7tw3no>

JOG wrote:
> J M Davitt wrote:

>> JOG wrote:
>>> J M Davitt wrote:
>>> ...
>>
>> I don't think I agree that a "thing" changes because one of
>> its attributes changes.  Yes, the tuple representing it does
>> but the thing is the same.  For instance, the change in the
>> database may occur because the value for some attribute was
>> incorrectly recorded; that doesn't mean there was a different
>> thing.  Or instance of an entity.  Whatever

>
> Ok, I understand your viewpoint, but I think its a mistake. Liebniz,
> whose definition of identity is key to formal logic, states that
> something is identifiable and distinguishable through its attributes,
> and its attributes alone. If a comparison is made between item 1 and
> item 2 and they have identical attributes, then they are the same
> thing. If all attributes are not the same then they are by definition
> different things, and the fact that the items being considered exist at
> different points in time makes no difference to this equality
> comparison. Thinking in terms of permanent 'entities' clouds this logic
> (which granted is unintuitive to everyday life), and hence the
> preference for thinking in terms of statements of fact alone.
> ...

Thanks for that, seems useful to me. I don't object to people sometimes wanting to model permanently fixed 'entities', just to them saying that the purpose of RT is to always reflect whatever they happen to think an entity is at the time.

p Received on Fri Jul 28 2006 - 21:25:32 CDT