Re: 1GB Tables as Classes, or Tables as Types, and all that refuted

From: Jan Hidders <jan.hidders_at_REMOVETHIS.pandora.be>
Date: Wed, 24 Nov 2004 21:06:40 GMT
Message-ID: <AJ6pd.34550$0d1.1769598@phobos.telenet-ops.be>

Alfredo Novoa wrote:
> On Wed, 24 Nov 2004 19:52:30 +0100, "Ja Lar" <ingen_at_mail.her> wrote:
>
>

>>So the set of values specified by the type is (in this case) a set of 
>>relations.

>
>
> Indeed.
>
>

>>But the set of values that a relation is ("contains") cannot be mapped to 
>>the set of values that a type is ("contains")

>
>
> They can be mapped but then you are not mapping a relation to a type,
> you are mapping the values contained in the tuples of a relation to
> the values of a type. This is a value to value mapping.
>
>

>>, ie. "a relation cannot be a 
>>type", if I understand you right ? 

>
>
> This is true, but this is not the same as you said above.

Just out of curiosity, Leandro, are you seriously contending that DBMSs that treat relational variables as types are making a fundamental mistake because by definition relational variables and types are distinct in the sense that something cannot be both at the same time?

I simply cannot believe that you do not see what is problematic about this type of argument. Seriously.

• Jan Hidders

Received on Wed Nov 24 2004 - 15:06:40 CST