Re: How would you approach this?

"Anthony W. Youngman" <thewolery_at_nospam.demon.co.uk> wrote in message news:<k+WamtIpyvp$Ew7J_at_thewolery.demon.co.uk>...
> In article <64ea97cf.0310261908.51048256_at_posting.google.com>, Bruce
> <brennie_at_dcsi.net.au> writes
[]
>
> As is often the case, here we have the SQL guys thinking they live in an
> ideal world, thinking that their clients actually KNOW and CAN EXPLAIN
> what they want. The real world ain't like that!

This is getting into Software Engineering issues rather than DB design issues.

Actually there needs to be an analyst that talks to the client and get the needs and requirements. Whether that analyst is a separate person or, in the case of a small contract assignment, the engineer takes that role.
>
> It's just so *easy* in MV to knock up a demo, and because MV relates to
> the real world, not ivory-towered theory, very often it doesn't take
> much to turn what started out as a demo system into rock-solid real-
> world production system. Maybe it's not hard with SQL, but it's a
> pushover for us to design a system based on partial requirements, show
> the customer and get them to explain further, and then repeat the loop.

Prototypes are not a function of the language so much. SQL would be no harder than other DB languages. Usually it is the User Interface that needs a rapid prototype. Dummy data can be got from files, quick layout tables, or even generated on the fly sometimes. Work flows and mockup screens are more useful than a DB at this stage of development.

>
> Oh - and in case you're thinking that putting into production something
> that started as a demo system is daft - well the big problem is always
> getting the file layout (or as SQL would put it, the table layout)
> right, so I think you would do the same as us - demo file layouts until
> you get it right and the final demo goes into production ... except we
> can probably demo a lot more cheaply :-)
>
> Cheers,
> Wol

No, the "file layout" is easy. Getting the workflow right can be very hard. Prototypes are good for their purpose, and I use them when appropriate.

However, I, for one, am of the school that you "Throw away the prototype."
WHY? all the quick little compromises along the way (some that you don't even remember you made) are the stuff of maintenance nightmares. It's appropriate for finding the new workflow, and that's about it.

Engineering is all about using tools properly to solve problems. Putting a prototype in production is short term savings but long term expensive. Long term maintenance costs of many production systems outstrip their initial development costs. Doing a prototype and then a clean production version can have real long term benefits.

 Where do you want to invest your money?

This is off databases, so I'll stop here.   Ed Received on Tue Nov 04 2003 - 20:50:00 CET