Re: candidate keys in abstract parent relations

"David Cressey" <dcressey_at_verizon.net> wrote in message news:5_4Bf.10306$mj3.8749_at_trndny06...
> Your post reminds me of the "database contents as opinion" concept.

I'd rather think of it as "contents as provisional assertion". But broadly yes, that is what I think it is.

> Second, there are two ways in which value can be measured in the
> information industry: content and conveyance.
>
> Considering the database as pure conveyer, what you say above is true.
But
> if the database is valued for content,

I think we are in danger of using the word "database" to mean two different things. I don't care about and don't wish to comment on the contents (one sense of the word). The contents belong to the business, and indeed there are some businesses (e.g. national security) where the likes of me are scrupulously denied sight of the contents. But I do care about the conveyance (the other sense of the word). So I guess you are saying you think I am right about the bit I care about, so I am happy.

> not merely conveyance, then the above
> comment is merely finger pointing.

Pointing the finger to falsely avert blame is wrong of course. But accepting blame that cannot be one's own is just as wrong.

> Consider a dictionary, as a database of words and definitions. If a word
is
> mispelled in the dictionary, the guy who runs the printing press can
claim
> that it's not his fault. But somebody in the production of the
dictionary
> is responsible.

The DBMS has no way to gather or authenticate its contents any more than a dictionary does. If there is an error in a dictionary, someone (i.e. a person, a "user") did make it.

> Surrogate keys are a hammer. Not every design case is a nail.

Yes.

Roy Received on Mon Jan 23 2006 - 15:47:09 CET