Re: Advice on SQL and records

From: David Cressey <david.cressey_at_earthlink.net>
Date: Wed, 24 Aug 2005 13:25:01 GMT
Message-ID: <Ny_Oe.1158$9i4.3_at_newsread2.news.atl.earthlink.net>


"Marshall Spight" <marshall.spight_at_gmail.com> wrote in message news:1124814513.160195.185860_at_o13g2000cwo.googlegroups.com...
> Jon Heggland wrote:
> > marshall.spight_at_gmail.com says...
> >
> > > You're right that "row" is a less than ideal term, though. The
> > > best term I can think of would be "member" from set theory, but
> > > I don't think that term would work out too well.
> >
> > What's wrong with "tuple"?
>
> Nothing! But the term "tuple" doesn't say anything
> about this datum existing within a larger structure,
> the set, whereas "member" does. Of course, "tuple"
> says what the metastructure of the datum is, and
> member says nothing about the structure.
>
> Sigh.
>
>
> Marshall
>

I prefer to use different parlance for different levels of abstraction. Here's what I'm accustomed to, after twenty years worth of relational database work:

Traditional parlance: file, record, field, data element.

SQL parlance: table, row, column, domain.

Relational parlance: relation, tuple, attribute, domain.

These are not quite interchangeable concepts.

For example, elsewhere in the discussion, someone said that the RM requires rows in a table to be unique.

Not quite. The RM, by definition, makes the tuples in a relation distinct. An SQL table can be designed that will allow duplicate rows. I've seen it done, in production systems. It's bad design, to be sure, (except in very unusual circumstances) but it isn't ruled out.

Elsewhere in the discussion, someone asked why we are so obsessive about terminology. I think that, to the extent that I obsess about the terms given above, it's because they are nearly interchangeable, but not quite. The danger is that someone that knows one of the sets of concepts will think they know the other set automatically, and they don't. Received on Wed Aug 24 2005 - 15:25:01 CEST

Original text of this message