Re: Lucid statement of the MV vs RM position?

From: Jon Heggland <jon.heggland_at_idi.ntnu.no>
Date: Fri, 05 May 2006 14:42:00 +0200
Message-ID: <e3fh6m$rro$1@orkan.itea.ntnu.no>

paul c wrote:
> Jon Heggland wrote:

>> I can't parse the above, but as I read the GROUP definition, R GROUP ()
>> AS RVA is equivalent to EXTEND R ADD TABLE_DEE AS RVA.
>> ...

>
> Oh, maybe this is a question for the authors. I thought it was
> equivalent to R JOIN TABLE_DEE

No, R JOIN TABLE_DEE equals R. GROUP always adds an RVA; that's its purpose. I checked the GROUP/UNGROUP treatment in "Temporal Data and the Relational Model", and there they say explicitly that grouping on no attributes is equivalent to extending with TABLE_DEE.

Since GROUP is a unary operator, I don't think it makes sense to talk about identity with respect to it (like TABLE_DEE is identity with respect to JOIN or whatever the terminology is).

> and now that I think about it maybe I was
> wrong to say that group always implies a new FD.

No, I think that's right. In the resulting relation, the set of attributes that were not grouped, functionally determines the RVA. That follows from the definition of GROUP.

-- 
Jon

Received on Fri May 05 2006 - 07:42:00 CDT