Re: The Theoretical Foundations of the Relational Model

On 25 Jun 2002 21:42:16 +0200, hidders_at_hcoss.uia.ac.be (Jan.Hidders) wrote:
>>How about just ordering views aka relations?
>
>In the external schema you can order as much as you like.
>
>> When it comes time to implement, don't I perhaps want to "bias"
>> towards a specific schema?
>
>Yes, that's what you do in the internal schema, but *not* in the conceptual
>schema. I assume you still know what the purpose of having the three layers
>was.
>
>> Anyway, if I define different views with different sorts, I'm not sure
>> there's any bias. Even "order on a table" can just be a view plus a sort,
>> "select * from mytable order by akey", rather than a physical ordering.
>
>Again, in the external schema you can choose your favourite data model to
>represent your data and order all you like.

Actually, no, I am constitutionally unable to "know" what the three layer model is supposed to do for anybody, and this is a perfect example why. It seems to make any discussion of particulars three times more difficult for all involved with absolutely no compensating benefits, quite the opposite.

I always go right to the physical schema, and do not consider views to be logically separate.

My question to you is why actual database engines and standards should not allow ordering as part of a view. I am not advocating any revision of the laws of first order logic or anything else, and nothing like that would be required to implement ordering in views.

Joshua Stern Received on Wed Jun 26 2002 - 00:56:07 CEST