Re: Just one more anecdote

David Cressey wrote:
> "Mike" <mike@xxxxxxxxxxxxxxxxxxx> wrote in message
> news:1122468034.924778.194610@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
> > dawn wrote:
> >
> > [snip]
> > >As much as
> > >we might want to think otherwise, even the design of a conceptual data
> > >model is influenced by the designer's knowledge of the target dbms.
> > [snip]
> >
> > Speak for yourself, please.

Of course, I was talking about everyone except for me ;-)

> > A conceptual model describes how a piece the world works; you don't
> > even need a computer to do that. I translate conceptual models into
> > logical (relational) models to make sure I've accounted for all the
> > constraints I've found. My target platforms range from SQL dbms to
> > awk/m4 utilities controlled by shell scripts. I never consider the
> > target platform until I start accounting for all the constraints I've
> > uncovered.
> >
>
> Hear, hear!
>
> You've just outlined, in precise terms, the distinction between analysis and
> design.

What you wrote is true in theory, but I did a little look at what
passed under that category on various web pages and it was interesting
that when there what I might consider attributes of an entity (they
don't exist when the entity is gone), those are sometimes split out if
their value would be a list or a tuple already in the conceptual model.
I'll figure, then, that my idea of a conceptual model is closer to
yours.

Describing a piece of the world does entail design in that it is not
objective. How we perceive the world and zero in on what we consider to
be relevant is a creative, design task too, even if under the
"analysis" umbrella. If it were not, any two experienced, high
aptitude modelers would come up with the same model. That is rarely
the case, with deviations even greater in the logical model.

> Many times, we engage in unconscious design when we think that what we are
> doing is actually analysis. When we do that,
> we reframe the original problem. If we do so in a positive way, it's
> serendipity. If we do so in a negative way, it's "just one of those
> things". But it isn't. It's flawed analysis.
>
There is no such thing as complete objectivity when working with
language. All such analysis is flawed. But I will agree that some
analysis is better than others.

>
> Anyway, thanks for your comments! really!
Hey, you didn't say that when you replid to me, David. smiles. --dawn

.

Relevant Pages

  • Re: Just one more anecdote
    ... >>we might want to think otherwise, even the design of a conceptual data ... My target platforms range from SQL dbms to ... > target platform until I start accounting for all the constraints I've ...
    (comp.databases.theory)
  • Re: XP Requirement Analysis?
    ... That is "fawlty" because, again as I see things, all objectivity has a ... While Test Driven Design tests are created from pre-coding ... overall in one direction or another) by core abstraction ...
    (comp.object)