Re: object algebra

> > Unless Date was talking about a db with one two-column (one of which
> > one is auto supplied) table, Date is wrong.
>
> Read the chapter again. He wasn't talking about one 2-column "table", and
> yet he's still not wrong. Read what he did say.

I re-read Chapter 20 of 6th Ed. It does not indicate that NULLs aren't an integral part of RDM? Since he spends an entire chapter on the source of NULLs, the consequences of NULLs and how to deal with NULLs by replacing them with "disciplined default values", I would conclude that NULLs are an integral part of RDM and masking the problem with "disciplined default values" is the only practical/partial solution. If NULLs weren't an integral part of RDM, he could simply drop Chapter 20.

The very definition of a relation which optimizes RDM for a limited scope of all that is possible under relational algebra, is also the cause of NULLs when applications fall beyond RDM's scope. In extreme cases, the NULLs can be only be avoided by resorting to the impractical solution of a one 2-col table. The inability to implement Ex076 in RDM without partial generic modelling should make one realize this. Received on Sat Feb 28 2004 - 21:04:20 CET