Re: Proposal: 6NF

> I don't agree. Judicious use of NULL can improve performance, response
> time, and concurrency.
Give me an example of *judicious* use of NULLS.

> In addition, when used correctly, no subjectivity is
*used correctly* ? By whom? Can you seriously guarantee that a developper will certainly not make a mistake during the entire life cycle in one of the queries conditions? Do you think that developpers have in mind all the consequences of not putting all IS NULL/IS NOT NULL? Besides, at what price is such *correct usage* of NULLS (say how many extra hours of development). Imagine you have 60 stored procedures refering to one table Table1 with 2 NULL values. Imagine that each stored procedure has 2 NULL conditions. That makes 60 * 2 extra conditions and SCANS on stored procedure. I do not see HOW that can improve response time OR concurrency. But I am positive you will provide an example.

> introduced. For every database schema with NULLable attributes, there is
> always an equivalent schema without. Any subjectivity would therefore be
> independent of the database schema.
The fact that there is a way to get rid of NULLS does not say anything about subjectivity or objectivity.

A normalized schema allows systems to make inferences about data according to a predefined rules. Such rule do not change every time a user reads false query results. Use of NULLS imposes subjectivity on every read done on false results.

> On the other hand, arbitrary or
> incorrect use of NULL can cause all of the problems you describe.
Use of NULL is necessary arbiratry no matter what you do. If you find a NULL value in a field, is there a systematic way to say anything about on all tables all schemas: NO.

Besides, even if you apply some discipline onto a systematic treatment of NULLS, how can you guarantee that none of the developer will not make any mistake now and after you leave the company to get more money?

Don't you realize that you are stating that incorrect use of something cause numerous problems therefore it is a good idea of using it. Is that your way of proving that NULL are useful? I say they create problems, don't use them. Period. Received on Mon Oct 02 2006 - 18:12:10 CEST