Re: Stored fields ordered left to right

From: Marshall Spight <mspight_at_dnai.com>
Date: Sat, 27 Dec 2003 15:32:05 GMT
Message-ID: <VBhHb.671183$Tr4.1678668@attbi_s03>

"Dawn M. Wolthuis" <dwolt_at_tincat-group.com> wrote in message news:bsira5$jlt$1_at_news.netins.net...

>

> I've been puzzled by this for quite a while, just figuring that relational
> theorists have this wrong. But the writings seem so sure of this. I had
> thought that the relations in relational database theory were mathematical
> relations, but I am beginning to think that might not be the case. My
> masters in mathematics was quite some time ago, so I hauled out some books
> and googled a bit and everything I find that is mathematics, rather than
> database theory, indicates what I thought about relations -- a relation is a
> set of ordered tuples -- right? What am I missing here?

I'll give you my perspective, but I don't know how much it'll help.

The question is, how does one distinguish the attributes in the relation? There are two choices: numerically/positionally, or by name. That is, one either has a mapping from 1, 2, ... n to attribute, or one has a mapping from name1, name2, ... namen to attribute. To me, it's not all that great a difference. It's just a question of what the application is.

If one is writing a page of equations, the convenience of using positional identification is high. One is likely working with a single relation at a time, and it is relatively simple to keep the ordering straight in the author's and the reader's head.

If one is part of a team building an enormous software system, then by-name is the better way to go, because of the mnemonic value of the names. There are likely a lot of relations with a lot of attributes.

It doesn't affect the semantics of relations or relational operators; it just affects how attributes are identified.

Marshall Received on Sat Dec 27 2003 - 09:32:05 CST