Re: dbdebunk 'Quote of Week' comment
Date: Mon, 29 Aug 2005 12:11:01 +0300
Message-ID: <deujf8$imv$1_at_domitilla.aioe.org>
"dawn" <dawnwolthuis_at_gmail.com> wrote in message news:1125079806.115887.87770_at_g47g2000cwa.googlegroups.com...
> and there have been several prior discussions on pointers that I think
> got most of us to the point of understanding that the pointers that the
> relational model were eliminating were related to memory locations.
> When talking about data that serve as references to other data at the
> logical level, there is nothing in the relational model that prohibits
> or even discourages such.
One goal of the relational model is making all data explicit.
Example:
E1(p1, id1, ...) with p1 primary key, id1 candidate key
E2(p2, id2,...) with p2 primary key, id2 candidate key
R(p1,p2,...) with (p1,p2) primary key
What is the reason for introducing p1 and p2 with no meaning for the
end-user ?
Are p1 and p2 references to id1 and id2 ?
E1(id1,...)
E2(id2,...)
R(id1,id2,...)
is a better or an worse model ?
> It is interesting to me that the relational model that some say is
> intended as a logical model for data had as one of its goals a physical
> issue. What do you make of that? --dawn
One goal of a logical model is to be (a) logical (model). Received on Mon Aug 29 2005 - 11:11:01 CEST