Re: Multiplicity, Change and MV

From: David Portas <>
Date: 12 Apr 2006 13:01:28 -0700
Message-ID: <>

Neo wrote:
> Assume at initial design time, you are told by the customer to model a
> thing (ie thing1) which has an attribute (attrib1) which has one value
> (val1). Please post the actual RM script for this situation and
> populate db with thing1. Also include the query to find thing1's
> attrib1 value.

What is an "RM script"?

> Next assume you are now told by the customer that attrib1 can have any
> number of values (ie NULL, val1, val2, val3 ...). Please post the RM
> script, which when executed against the already populated db (populated
> with thing1), allows thing1 to have val1, val2 and val3. Or post the RM
> script which creates an empty db with new schema and transfers data in
> old db to new. Verify if the original query still works, if not,
> include the new query to find thing1's attrib1 values.

Why "create an empty database and transfer the data"? That sounds very like a description of a physical implementation, not the model itself. At the logical level this is as simple as variable assignment. A new relvar is assigned the result of some relational expression on the old one. You are right - today's databases often make this hard because of insufficient physical data independence. It's not a model problem however.

David Portas
Received on Wed Apr 12 2006 - 22:01:28 CEST

Original text of this message