Re: We claim that delete anomality is due to table not being in 3NF, but...

On Wed, 29 Oct 2008 18:32:13 -0700 (PDT), Srubys_at_gmail.com wrote:

(snip)
>So in essence, we don't consider just the loss of existence an
>anomaly?

Hi Srubys,

If the model was designed to explicitly store existence, then loss of existence would be considered an anomaly. If the model doesn't store existence, then it can't be lost.

Consider this (simplified) model for a savings account:

  ACCOUNT (AccountNo, Balance)

If I make a withdrawal taking my balance from $1300.75 to $1200.75, and at that time no other account happens to have a balance of $1300.75, would you consider that loss of existence? And would you worry about the database no longer explicitly representing $1300.75 as a possible balance?

(snip)
>I understand ( to a point ) what you are saying here, but still, even
>though the existence is just implied in a relation
>
>SUBJECT_TEACHER( SUBJECT_ID, SUB_NAME, TEACHER_NAME, CLASSROOM )
>
>would it be so wrong to say that due to transitive dependency we have
>a delete anomaly, which may cause a lost of relatioship ( thus lost of
>data ) and also a loss of existence ( I realize lost of existence is
>not considered an anomaly )?

You can't lose what you don't store. If you want to persist existence, you need to add a relation for it in your model. If you don't have such a relation, then you're obviously not interested in storing existence, so you shouldn't worry about losing it.

Best, Hugo Received on Fri Oct 31 2008 - 00:34:11 CET