Re: does a table always need a PK?

Heikki Tuuri wrote:
>
> "Bob Badour" <bbadour_at_golden.net> kirjoitti viestissä
> news:ZIN2b.963$8o2.85881998_at_mantis.golden.net...

>> 
>> Are you aware of any sound theory purporting to deal with the problem of
>> missing information?

3-valued logic does not have a higher complexity then 2-valued logic. The problem is that such n-valued logics make a clean and straightforward interpretation of NULL values impossible. For example, if you think a NULL means that "there should be value here but I don't know what it is" then the condition "X = 1 OR X <> 1" should return TRUE, even if X is unknown, but the 3-valued logic will probably say something like UNKOWN. As you can probably guess computing the truth value of propositions with AND, OR and NOT under the given interpration of NULL values is already NP-hard if you have simple equalities and becomes undecidable if you allow simple arithmetic.

So the bottom-line is that n-valued logics make a logical and theoretically sound interpreation of NULL values impossible, but keep things computationally tractable and simple enough to be understandable for most users.

• Jan Hidders