Re: Programming?

[awilliam@xxxxxxxxxxxxx]

> Increasingly, but in general it depends on how well a shop has kept up
> with and heeded developments in software engineering. Prototyping (for
> all but the most trivial app.'s) SHOULD be a standard approach to
> developing user
> interfaces. In current practice, it happens but spottily at best at
> most companies. I use it for everything except cycle, batch tasks.
> >  Isn't this a very expensive way of developing an application.
> Yes, BUT...  Not when compared to the (usual) cost of developing the
> WRONG application.  In most development projects there are missed
> specifications (things known but not mentioned and things not known)
> and mis-specified requirements - prototyping helps to reveal many of

Isn't this the truth!

> these. The most costly part of the life-cycle for any significant app.
> is its post-deployment, support phase. The retrofit of functionality

I'm an administrator/maintainer so I heartily agree with the above.

> missed during development costs hundreds of times more than getting it
> right "up front." As a rule, we used to see relatively low development
> costs and astronomically high support costs. We also witnessed an
> alarmingly high turnover in app. versions or replacements (or
> perpetual dissatisfaction with existing app.'s). Prototyping is one
> method which transfers costs to the development phase but results in
> large factor, multiplicative reductions in
> support-phase costs relative to those development increases.
> Prototyping also reveals flaws in the ease of use and logical flow of
> the app. in light of the actual business process which only the user
> is likely to detect. If the interface ain't "ergonomically" well
> organized, it will
> not be as well received,  maybe not even used. This translates to
> app.'s that look good on paper, get built, and nobody will touch (cuz
> they hate the front-end)...talk about expensive.
> Finally(still "re" expense), I'm prototyping in APL. I can prototype a
> front-end in hours that would takes days (maybe weeks, if I'm
> comparing to a CICS/COBOL alternative) in most environments. If I
> throw the whole thing

But are "modules" of the prototype ever used in the "real" system, or 
is it a requirement that all code be scrapped in order for it to be an 
official "prototype"?  I'm just curious of the precision of these 
terms.

> out, no big loss, especially in light of what we learn about WHY I
> should chuck it.

> "Software engineering" is finally becoming a meaningful phrase, and
> that requires that we move beyond a reliance on the "gifted artisan"
> programmer/analyst who's "know-how" is incredible but entirely
> personal (and
> God help us, if he slips in the bathtub!). It extends the capabilities
> of each of us, when we have better tools - prototyping is one such
> tool.