Re: Define "flatten database" ?

"David Cressey" <david.cressey_at_earthlink.net> wrote in message news:Rc6Ld.3992$Ix.764_at_newsread3.news.atl.earthlink.net...
>
> "Alan" <not.me_at_rcn.com> wrote in message
> news:36483jF4tcn38U1_at_individual.net...
>
> > I now understand why you understand it the way you do, but flattening is
> not
> > normalizing. Flattening a relational database is denormalizing. Not all
> > denormalizing is flattening, but all flattening is denormalizing- unless
> > someone can prove this to be wrong.
>
> I have not wanted to offer a definition of "flattening" because I've never
> seen a formal definition of "flat data". I've seen database data
contrasted
> with "flat files" enough times to have an inkling of what the writers
> intended by the term "flat files". And not all writers mean exactly the
> same thing. Let me come back to this.
>
> First, a problem with this discussion is that "normalized",
"denormalized",
> and "flattened" are all being discussed as though they were points (or
> directions) along a scale with only one dimension. I don't think so.
There
> is plenty of data that is neither normalized nor flat, nor anywhere in
> between.
>
> When you are faced with complex data rendered in a normalized form, and
you
> are asked how to flatten it, you might say "denormalize". When you are
> asked how to get from New York to San Francisco, you might say "cross the
> Hudson". Both are true, but don't really deal with the subject.
>
> Now back to what I think the people who have written about data in "flat
> files" meant. Most of the time, the writer was referring to a stream of
> records, where all the records were of the same type. Some writers wanted
> the records themselves to be "flat records", and some writers meant
> different things by that.
>
> By "flat records" some writers meant that no field of a flat record could
be
> a record. This turns out not to be a very useful distinction.
>
> Other writers meant that a "flat record" could not contain an array of
data,
> otherwise known as a repeating group. Hence, Dawn's comments.
>
> If there is an actual definition of "flattened", it would be nice to add
it
> to the glossary. If not, we should just accept that IT argot, like
that
> of other crafts, has a lot of terminolongy that just came into being
without
> formal analysis.
>
>
>

A very reasonable discussion/analysis of the points. Thank you. Received on Sun Jan 30 2005 - 19:26:06 CET