What is the definition of a Volatile Class?
in [OOP]
|
Prev: Call for Papers Reminder (submission deadline extended): International MultiConference of Engineers and Computer Scientists IMECS 2010
Next: Last Call for Papers Reminder (extended): IAENG International Conference on Artificial Intelligence and Applications (ICAIA'10)
From: S Perryman on 1 Jan 2010 12:49 MikeB wrote: > the reason I questioned what looked like an assertion was that I've been > exposed to codebases where both V and I are high. Those systems > invariably exhibited poor design. By "invariably" , I would expect to be shown a strong objective statistical correlation between V(E) , I(E) , and other canonical 'quality' measures (coupling, cohesion etc) . > Taking this further, I'd venture that for a given volatility, impact is > a purely function of the quality of the design. When E has a high value for : 1. I(E) , we are informed about the potential effects of change throughout a system. 2. V(E) , we are informed about the amount of change occuring for E. #1 allows us to ask what will happen (rework, re-testing etc) if E changes, and the justification thereof. #2 allows us to ask why is E changing so often. The reasons for either are often nothing to do with poor design (good design is actually often the cause) . For example, something that is very useful will have a very high I(E) value. Think of things like I/O, container functionality etc. > I wonder whether I misunderstood you and you meant that "in a good > design, V and I are inversely proportional". Over time, entities with high V(E) will exhibit I(E) values tending to the relationship (high impact entities tend to become less volatile - they 'stabilise' as their desired/essential purpose becomes clear etc) . Regards, Steven Perryman
From: S Perryman on 1 Jan 2010 12:57
Daniel Pitts wrote: > MikeB wrote: >> I wonder whether I misunderstood you and you meant that "in a good >> design, V and I are inversely proportional". > It could be that Steven meant that there is a lower bounds on V*I, but > that is only my guess, I have nothing to back that up. V(E) and I(E) are in the range [0,1] . The product can be 1 at worst (impacts everything, changes continually) . And tends to 0 at best (inversely proportional, the implications of one property is tempered by that of the other) . Regards, Steven Perryman |