Re: ID field as logical address
Date: Fri, 29 May 2009 19:01:14 GMT
I don't know that doing so is some kind of theoretical error. Since they are used to define relations, then presumably a single tuple could be an attribute value. However, when we already have relations and a set of relation operators, doing so seems superfluous.
(I have often mused to myself about sets of tuples, constituting a bastard attribute that is unnamed, which may have uses, but it wouldn't be allowed in Codd's model.) Received on Fri May 29 2009 - 21:01:14 CEST