Re: NULLs: theoretical problems?
Date: Thu, 30 Aug 2007 20:06:23 +0200 (CEST)
JOG <jog_at_cs.nott.ac.uk> wrote in
>> A natural question arises what is the def construct ? Is it a
>> classical logic extension ?
> PMFJI but I'm struggling to see why you believe there is a diversion
> from classical logic. How does def differ in your mind from the
> existential quantifier?
The existential quantifier does not really promise that something exists. The quantifier is just an 'OR' shorthand. So you need to have either strong/"existential" equality or a definedness operator to make this weird two-valued partial logic that Jan's so fond of work. None of this trickery is needed with the three-valued tools but they have their problems too !
>When I ask
> for a subset from a set of propositions (via a select clause say),
> well, I need a DBMS to be able to fulfill that request. I can't ever
> imagine a situation where I'd want the system to be "unsure" of what
> goes into that subset. Surely by encoding accurately, not using
> metadata, etc that situation need never arise - isn't that
> instinctively more desirable?
The partial two vl and the three vl logics express roughly the same wrt. undefinedness so you can have a false sense of security with the former and be confused with the weak/strong equality, or have less verbosity but get confused with truth tables of the latter ! The problem is undefinedness, the logics are just a tool. Whether it's a good or bad tool is up to you to decide.
It is not. I've misunderstood from the very beginning what Jan has been talking about hoping to hear something new and exciting but in vain, alas. Received on Thu Aug 30 2007 - 20:06:23 CEST