Re: Does Codd's view of a relational database differ from that of Date & Darwin? [M.Gittens]

NULLs in the real world? Where, I don't see any? I have never used the expression NULL in anything other than a technical discussion about databases.

If my Manager points to a NULL in a report, I don't say 'That is a NULL.' I say something like 'That customer is retired, and does not have a work address' or 'The user left that field blank.' That is if I am familiar with the data. If I am unfamiliar then the NULL tells me nothing, I can only shrug my shoulders and say 'Sorry, I don't know.

Date does recognise the problem of missing information, which NULLs attempt to address, RM Very Strong Suggestion number 8 states: "D should provide some kind of special values mechanism for dealing with "missing information."

This is the candidate method for dealing with missing information.

Special Values could be considered a superset of the use of NULL. For example, a special value NULL could be defined and given the same behaviour as the existing null.

Howver, it would also allow much more useful special values. NOT_APPLICABLE and NOT_YET_PROVIDED , for example.

Another proposed method can be found here: "How to Handle Missing Information without Using Nulls" (free pdf from Hugh Darwen using vertical decomposition)

And here: http://www.dbdebunk.com/page/page/1396241.htm (charged pdf from Fabian Pascal using meta data.)

While special values are clearly superior to NULLs, I think they are still searching for the final answer to the problem of missing data. This explains why they are a very strong suggestion rather than a presciption. Received on Wed Jun 08 2005 - 14:30:09 CEST