Re: Gary's weird test case

Michael Kifer wrote:
> 
> 
> On Fri, 26 Sep 2008 16:00:21 -0400
> Jos de Bruijn <debruijn@inf.unibz.it> wrote:
> 
>> It was my test case, and you missed the p(?x) in the body, so the rule
>> is not inconsistent.
> 
> OK. But then we actually derive
> 
> Forall ~p(?X) \/ isInt(?X) \/ isStr(?X)
> 
> which is not a condition formula.

Sure.
But we also derive isInt(a) \/ isStr(a)

because p(a) is in the premise.


Best, Jos

> 
> m
> 
>>
>> Jos
>>
>> Michael Kifer wrote:
>>> This is regarding the rule that kept bugging me:
>>>
>>> 1=2 :- AND(p(?X) isNotInt(?X) isNotString(?X)).
>>>
>>> It seems this is inconsistent in the presence of other data types, such as 
>>> xsd:date. This is because this implies
>>>
>>>  Forall ?X OR(isInt(?X) isString(?X))
>>>
>>> Since there are things that are disjoint from ints and strs (e.g., dates,
>>> time), it follows that the above rule is inconsistent.
>>>
>>> To make the above stick, all data types must be mentioned in the body.
>>>
>>>
>>> 	--michael  
>>>
> 

-- 
                         debruijn@inf.unibz.it

Jos de Bruijn,        http://www.debruijn.net/
----------------------------------------------
One man that has a mind and knows it can
always beat ten men who haven't and don't.
  -- George Bernard Shaw