Subject: Analysis Independent of Implementation


From: Mark Collins
Date: Fri, 14 Apr 1995 08:43:03 -0600

There was a quote from your position statement at WOOD '94 in the comp.software-eng newsgroup, so I read your complete statement. You seem to be arguing that analysis should be tied to the implementation at the same time that you are proposing more flexibility in design/implementation.

Is your point that a more flexible implementation will eliminate the need to re-implement a design in a new environment? and by extension it will no longer matter if the analysis is tied to the implementation because the implementation will not change. Current practice is that analysis concerns itself with the problem and not the solution and that design takes into account how the solution will be constructed in a given implementation environment.


From: ward (Ward Cunningham)
Date: Sat, 15 Apr 1995 10:29:50 -0700 (PDT)

No. Though the idea is interesting. What do you call the activity of matching capabilities (say core competencies or even reusable frameworks) with desires (like improve customer service)? That's what I meant by analysis. I see development teams co-evolving with their business sponsors. Flexibility simply enables the evolution.

So, what was the general conclusion of the discussion participants? When I've argued these points with people steeped in the traditional roles they are quick to point out that analysis, design and implementation can be done by the same person and that that person may need to pop back and forth between the activities. I say, when you can get a whole team to pop around like this without making lots of extra work, then you're flexible.


ward@c2.com