Re: more closed-world chatter
From: Bob Badour <bbadour_at_pei.sympatico.ca>
Date: Thu, 03 May 2007 12:46:37 -0300
Message-ID: <463a03ae$0$4062$9a566e8b_at_news.aliant.net>
>>On May 2, 10:13 pm, paul c <toledobythe..._at_oohay.ac> wrote:
>>
>>
>>>If I have a domain of items consisting of the values "a" and "b" and a
>>>domain of item prices consisting of the values "1" and "2" and an empty
>>>relation {ItemID, Price} known by the relvar name "Items", I take it
>>>that the logical complement of "Items" has four tuples - loosely,
>>><a,1>,<a,2>,<b,1>,<b,2>.
>>>
>>>If I select from Items where Price = 3, should a dbms answer with an
>>>empty relation or should it take exception, such as "illegal question"?
>>
>>I would like to point out two comments triggered by your question.
>>
>>A TRDBMS should detect disjoint types are to be determined at compile
>>time. In my perspective, there should be no exception because there
>>should be no execution in the first place. That of course, in a
>>purely empirical perspective.
Date: Thu, 03 May 2007 12:46:37 -0300
Message-ID: <463a03ae$0$4062$9a566e8b_at_news.aliant.net>
David Cressey wrote:
> "Cimode" <cimode_at_hotmail.com> wrote in message > news:1178200219.598399.84110_at_o5g2000hsb.googlegroups.com... >
>>On May 2, 10:13 pm, paul c <toledobythe..._at_oohay.ac> wrote:
>>
>>
>>>If I have a domain of items consisting of the values "a" and "b" and a
>>>domain of item prices consisting of the values "1" and "2" and an empty
>>>relation {ItemID, Price} known by the relvar name "Items", I take it
>>>that the logical complement of "Items" has four tuples - loosely,
>>><a,1>,<a,2>,<b,1>,<b,2>.
>>>
>>>If I select from Items where Price = 3, should a dbms answer with an
>>>empty relation or should it take exception, such as "illegal question"?
>>
>>I would like to point out two comments triggered by your question.
>>
>>A TRDBMS should detect disjoint types are to be determined at compile
>>time. In my perspective, there should be no exception because there
>>should be no execution in the first place. That of course, in a
>>purely empirical perspective.
> > I am not convinced. Why should the binding happen at compile time. > Couldn't a TRDBMS be extended to allow dynamic typing?
In a way, that depends on what you mean by dynamic typing. As D&D point out, one would have to evaluate the comparison as a common supertype. If no other common supertype exists, one would have to evaluate the comparison as the universal supertype, which includes all values and no operations other than the equality comparison.
The result of the join would be empty and have an attribute with no operations defined other than the equality comparison. Of course, one could project away the attribute defined on the universal supertype, but one would still have at best an extended DUM. Received on Thu May 03 2007 - 17:46:37 CEST