Re: Database design

From: mAsterdam <mAsterdam_at_vrijdag.org>
Date: Wed, 22 Feb 2006 00:23:08 +0100
Message-ID: <43fba0c5$0$11061$e4fe514c@news.xs4all.nl>

Mark Johnson wrote:
> mAsterdam wrote:
>

>>An entity is a thing of interest (old ISO definition).

>

>>Some prefer to talk about relations (or relational variables), 
>>attributes and tuples.
>>Others about tables, columns and rows.
>>Though the concepts are different

>
> Which is to say, non-parenthetically:
>
> A relation is NOT a table
> An attribute is NOT a column
> A row is simply NOT a tuple, to the latter of course, I'd agree.

Not to the first two?

> A relation is a set, which need not be written as a table, to be sure.
> But in a world of constraints, situation and conditions - in a world a
> databases and a ng devoted to an aspect of same - I also agree it
> would be reasonable to speak of relations as tables, attributes as
> columns, and rows as not . . . entities.
>

>>Yet, neither tables nor relations map to entity types.

>
>
> The single row is then called - what?

In the context of tables, just that: a row. In the context of relations: a tuple.

>>One relation may have attributese from several entitiy types,

>
> Then an entity type is seen as some unique attribute domain?

No. Why?

>>and one entity may have data spread across several relations

>
>
> Then an entity type is not some unique attribute domain?

Indeed not. Why should it? Received on Tue Feb 21 2006 - 17:23:08 CST