Re: 4 the FAQ: Are Commercial DBMS Truly Relational?

Kenneth Downs <firstinit.lastname_at_lastnameplusfam.net> wrote:

>Gene Wirchenko wrote:

[snip]

>> That single, unique column is also a constrained (PK). What
>> happens when a key for that column is entered wrong? Why would it be
>> any different for another key for another unique column, say SSN?

>The key fact about the Employee_ID column is that by definition it has no
>meaning, it is correct if and only if it is unique. It's only purpose is
>to be unique.

     An SSN is correct iff it is unique, too.

>Practically, the use of such a meaningless column solves a lot of practical
>problems which I won't belabor here. This technique is ubiquitous.
>
>Theoretically, this practice of introducing a meaningless unique column also
>just so happens to make my database "truly relational" at least insofar as
>all tables are now sets and not bags. A transaction id does the same for
>transaction tables.

     It has the danger of avoiding thinking the problem through. There may well be a natural key that would serve.

>My original question in this subthread is: what theory explains the benefits
>of the meaningless unique key? Do we claim that the RDM saved the day
>once we found a technique to use sets instead of bags? Or is the use of
>the key suggested by and supported by some other body of theory?

     Key choice, AIUI, is not part of the RDM.

     I would go with this as at least a part-answer.

     Not that I am aware of.

Sincerely,

Gene Wirchenko

Computerese Irregular Verb Conjugation:

     I have preferences.
     You have biases.
     He/She has prejudices.