Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> comp.databases.theory -> Re: How to find Brothers and Sisters?

Re: How to find Brothers and Sisters?

From: paul c <toledobythesea_at_oohay.ac>
Date: Tue, 05 Dec 2006 00:18:53 GMT
Message-ID: <NV2dh.424620$R63.236212@pd7urf1no>


JOG wrote:
...
> I would think this is more a design flaw than the table not formally
> being in 1NF. No repeating groups. check. ...

I would like to know where Codd spelled out just what a repeating group is (not saying he didn't just that I haven't seen it). At the time he wrote his first papers, there was a construction supported by the IBM access methods as well as IBM's IMS, known by various names which typically included the term "variable length" that encouraged a special case of what we equate with relation-valued attributes today (even though these days we might equate that stuff with the "physical" layer).   This stuff was rampant and encouraged something that he was battling, application-specific data that could not be shared by other applications unless one app distorted itself in favour of the other. (I've heard that his first stroke was encouraged by antipathy within IBM towards his stirring of the waters.)

Today, as we well know, most products confuse representation with expression and tend to store the straightforward (and admittedly useful)   for the purposes of manipulation and so-called "persistent" memory.

Also today, Darwen and Date more or less insist that there are certain relations that can't be represented without 'artifice', eg., additional attributes that aren't inherent in their original data. My impression is that Codd was not interested in these cases, preferring to require that users supply such attributes. I think that is fair enough. But I think he was also very focussed on having a very straightforward representation for users to look at. It is equally fair to try to imagine a logical representation that is not part of the "physical" layer but which allows rva-based expressions that are compatible with the straightforward one. In other words, I think such relations could be expressed without their extensions necessarily being materialized.

p Received on Mon Dec 04 2006 - 18:18:53 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US