Re: WWW/Internet 2009: 2nd CFP until 21 September
Date: Fri, 14 Aug 2009 15:11:32 -0400
"paul c" <toledobythesea_at_oohay.ac> wrote in message
> Mr. Scott wrote:
>> The closed world assumption doesn't mean that unknown values are false.
>> What it means is that when there isn't a row in a table, the fact
>> represented is false. A row with a null is still a row, so the fact
>> represented is true.
> The CWA doesn't depend on 'rows', only whether a fact is recorded (true)
> or not recorded (false). The fact that an order has a particular address
> is a different fact, with a different predicate, from the fact that an
> order's address is unknown.
I think you're confused about what a predicate is. The predicate states the fact that an order has one address from the domain of all addresses. Whether that address is at present known or unknown doesn't change that fact. That an order has a stated address implies that it has one address from the domain of all addresses; that an order's address is unknown also implies that it has one address from the domain of all addresses. The predicate is the same. Under the CWA, the lack of either a stated address or a statement that the address is unknown would imply that there isn't an address, which is clearly wrong.
> Trying to express that in an SQL row (maybe I should say table) means the
> row must have more than one predicate. It is unclear to me when SQL uses
> one of those predicates and when it uses the other.
> (Apparently the SQL standard addressed this at one point, maybe still
> does, by mentioning in some places that nulls are not values and in others
> that they are values, link below.):
I'm willing to stipulate that SQL sucks. Received on Fri Aug 14 2009 - 21:11:32 CEST