Re: NULLs

In article <5jqnqr$dh3@cssun.mathcs.emory.edu>, thabib@bob.zentropy.soft aware.com writes >You could also point to Codd's Relational Model, which explains that the >purpose of a null field is represent the absence of data, because it is >either unknown or unecessary. Replacing null with default values, would >presume the presence of data that is valid, thus ruining the model of the >databases as it was created orginally. Hope this helps. > Correct but there should be no unknown data in the database. i.e. if you don't know then find out before trying to enter it into MY database. :-> I've heard the argument that if you allow nulls then you're design is flawed.... -- David Williams i agree, if you define a relation schema where an attribute has the potential of being "null", you should consider a decomposition of the orignal relation schema. the attribute in the original schema that had the "null" value potential is now in a relation schema with at least one other attribute (primary key). make sure to test your decomposition to ensure it is a lossless join decomposition and it is dependency preserving. edward