Software Devlopment Cycle [ new header, old topic ]

Mats Wichmann mats at dual.UUCP
Tue May 8 01:29:53 AEST 1984


Dick Dunn makes a good point in that building a house is not `state-of-the-art'.
And it also evolved iteratively. 

I am going to put my foot in it and claim that a useful piece of major 
commercial software is not going to be right the first time. Sometimes I
think we could all benefit from some of those hardware designers who may
redo a board four times, totally throwing out the previous desgin each
time. We tend to hack up the code in attempts to `fix' it, often (mind
you, not ALWAYS) in the mistaken belief that there is something worth
salvaging....

I know plenty of people who have very carefully laid out their plans for
a project, and followed these steps, only to find that they are six months
over schedule and only thought of a third (it they were lucky) of what
they needed to...this is because most of us are learning as we develop.

This sort of thing is *going* to happen, and I could argue about HOW the 
revisions should be made (modify, rewrite, patch, etc.); the point is that 
most of us are driven by market pressure (hey, we have to eat. And the payment
on the Rolls is due). Somewhere along the line you have to balance the need to 
run through yet another iteration to get it right, with the need to go to 
market. This call is obviously not easy, judging from the number who have made 
the wrong choice.

There was a recent attack on Fortune's software - I think they fall into
the category of someone who released things too early becuase of pressure.
They raised a lot of capital with a lot of very high promises, and (apparently)
their chance of a good market share depended on being early. Turns out
(my opinion only) that they were wrong - the IBM PC cleaned up from below
in a market Fortune thought they could dominate, especially when Fortune
was hurt by their bad early reuptation (although it might not have made
any difference). From what I can gather, there is a world of difference
between those early releases and what Fortune has now (but their stock
is still in the toilet).  Who knows what kind of impact Fortune might
have made in the UNIX marketplace (as opposed to the `business' marketplace)
if they introduced their machine a year ago, instead of 2 1/2 (or whatever
the date was). But it seems they were forced into an early release, before 
things had gone through enough iterations to be particularly useful....

[ These opinions are to be considered not mine, but actually investment  ]
[ advice of Dean Witter Reynolds, suggesting that you buy as much stock  ]
[ of any company developing state-of-the-art software under pressure  	 ]
[ from impatient venture capitalists as you possibly can    		 ]

[ From another displaced oarsman... (Dean Witter also rowed for the      ]
[ U.C. Berkeley Crew before becoming rich and famous - must be an omen)  ]

	    Mats Wichmann
	    Dual Systems Corp.
	    ...{ucbvax,amd70,ihnp4,cbosgd,decwrl,fortune}!dual!mats



More information about the Comp.lang.c mailing list