Database myths and legends (Part 6) This is a great example of a myth/legend that is both true and false; it all depends on how you define relational.
Ted Codd produced one of the first attempts to define exactly what the term relational Database Management System (DBMS) means. Since Dr. Edgar Codd is regarded as ‘the Father of the Relational Database’, most people assume he knew what he was talking about. He published two articles in Computerworld (14th and 21st October 1985) and in the first he wrote:
“In this paper I supply a set of rules with which a DBMS should comply if it is claimed to be fully relational.”
These rules are often referred to as Codd’s 12 rules, despite the fact that there are 13 of them. Codd numbered them from zero and many people didn’t read the rules, they simply looked at the highest number and assumed that there were 12 - apparently even including Codd, who didn't count Rule 0 and mostly referred to his "12 rules" himself.
The first rule makes one thing perfectly clear:
‘Rule 0: For any system that is advertised as, or claimed to be, a relational data base management system, that system must be able to manage data bases entirely through its relational capabilities.’
In other words, to be considered a relational database engine, the product in question has to obey all of the rules. He then goes on to list what he considers the essential requirements. Not only are the rules relatively stringent, it turned out that one was impossible to fulfil. Rule 6 sounds simple initially:
‘All views that are theoretically updatable are also updatable by the system.’
However, Codd’s expansion of that rule: “Note that a view is theoretically updatable if there exists a time-independent algorithm for unambiguously determining a single series of changes to the base relations that will have as their effect precisely the requested changes in the view” is a logical killer. In1988, H. W. Buff published a paper entitled “Why Codd’s Rule No. 6 Must Be Reformulated” which proved that it is impossible to define such an algorithm precisely.
So, to be relational, an engine has to meet all of the rules and one of those rules has been proved to be impossible to fulfill. Since it is now clear that no database engine can ever be relational, it follows that Access is not relational. QED.
I agree that this is simply an exercise in logic chopping (although I have heard precisely this argument used); however it is also true that some people who work very seriously in this field really do think that none of the mainstream engines are truly relational. Chris Date (Ted Codd’s closest co-worker) has stated in an interview he gave to Tony Williams here that the existing mainstream DBMSs are “all deeply flawed”.