Re: sql views for denomalizing

From: Lauri <lauri.pietarinen_at_a>
Date: Mon, 01 Aug 2005 14:32:15 GMT
Message-ID: <PnqHe.256$C13.184_at_read3.inet.fi>


dawn wrote:
> Lauri wrote:

>>
>>Let me see if I have understood this correctly:
>>
>>1) first the "DBA" creates the tables (or whatever they are called in
>>Pick), more or less as he/she would if using an SQL DBMS

>
>
> Quite close, however, it is often a systems analyst who would execute a
> CREATE FILE (like CREATE TABLE) command, and often without including
> any attributes. This would be done during a prototyping phase, with
> attributes described and redescribed, possibly generating some
> prototype update screens until a system design is established. Unlike
> what is commonly considered good engineering of applications, I suspect
> it is very common in pick to work first on a prototype and then migrate
> that exact prototype into a final product, with the code and the data
> descriptions being complete at the same time.
>
> [Aside on methodology, skip below if not interested. Using some of the
> metaphors for software engineering, where SQL-DBMS app development
> might look more like a team producing a film, PICK app development
> might look more like a team climbing a mountain (or perhaps producing
> an indie film on a tight budget). While either methodology could be
> used in either environment, I would guess a pick shop is more likely to
> employ a methodology leaning toward "agile" (even if never having heard
> of XP) and a SQL-DBMS more likely to have heard of the CMM.

I agree that prototyping is a good approach for system development.
>

>>It would be interesting if you could provide the complete DDL (or
>>whatever it is called in Pick) for your mini-example together with the
>>queries.

>
>
> Since I don't have a pick environment handy, before I go through the
> trouble of loading one and making an example, there is a high level
> summary intro in a little "triology" that I prepared on business-card
> stock a few years ago. It can be read at
> http://tincatgroup.com/mv/trilogy.html . You can likely zip through
> the data and commands slides quickly. Then ask questions (or give
> opinions). Unlike what I write here, the trilogy has gone through a QA
> process.

I went throu the slides. I still think it would be advandageous for the discussion in hand if you could just create a 3-file minisystem and give a couple of queries. It does not have to be syntactically correct and you can even use pseudo code if you want. I think that here 20 rows of code easily equals 1000 words of explanation.

regards,
Lauri Received on Mon Aug 01 2005 - 16:32:15 CEST

Original text of this message