Re: Question on duplicates

Of course those "tables" without primary keys don't even meet the criterea of 1NF, much less meet the definition of a relation. Thus, in terms of the relational model for databases, there is no guarantee that relational operators (including those of SQL), nor the built-in mechanisms for integrity and consistency will perform as expected, or as needed. At least one candidate key is required for a "tuple-level addressing mechanism".

Daniel Guntermann

"Jerry Gitomer" <jgitomer_at_erols.com> wrote in message news:3C819F1A.2A99874F_at_erols.com...
> Jayne Heger wrote:
> >
> > Hiya all,
> >
> > I have a question on duplicate rows in a database.
> >
> > What I am having trouble understanding is, if for example, you have a
> > primary key value in every relation/table, surely you will have
eliminated
> > the problem of duplicates? so why the argument?
> >
> > Or am I missing something?
> >
> > I just want to make sure I am covering all aspects, and can answer
> > questions for when I do my seminar on Wednesday.
> >
> > Thanks
> >
> > Jayne
> >
> > Oh, and another thanks for the people who helped me earlier.
>
> Because some RDBMS allow you to create tables that do not have
> primary indexes. In fact it is possible to create tables with
> no indexes.
>
> Although this is contrary to theory indexes are often omitted
> because lookup performance for very small tables (those which
> will fit within a single OS physical block) is faster without an
> index than with an index.
Received on Mon Mar 04 2002 - 07:02:22 CET