Re: Eliminating Nulls
Date: 4 Jul 2005 13:55:20 -0700
I believe that the comment refers to the fact that a great deal of the discussion in the field of statistics deals with how to handle missing data. Since relational database theory was first described using relational algebra, which Codd invented just for that purpose, if follows that the thinking in one related scientific discipline may well apply to rdbms theory.
Personally I try to stay as far from math related topics as I can but I like the example as many data items will not have a known value till some unknown future point in time. You do not wait to define a data element till it has a value. If a single row has a value then every row in the table will have the attribute but will not have a value. Some designers argue that you can assign a default value but then how do you distinguish that the value present represents the actual value or is the default. You would end up having to hardcode the defaults into your applications or perhaps creating a default value table and joining it all over the place. Nulls are a natural default.
HTH -- Mark D Powell -- Received on Mon Jul 04 2005 - 22:55:20 CEST