RE: Re[2]: Is 'testability' a useful architectural constraint?

CL> MR> Clean architecture aids testability.  A good architecture need to
CL> MR> address what occurs when it is improperly used as well as 
CL> when it is
CL> MR> correctly used.  If something is designed with 
CL> improper/limited criteria
CL> MR> for the error cases, then dealing with these cases will be hard to
CL> MR> implement or even test.
CL> Well I elieve that statement too, which implies that there is a
CL> feedback loop there  - improving architecture improves testability
CL> which improves architecture of other specs that join on to that one.
CL> Sort of like ripples in a pond.

I'll agree with the feedback thought, but would offer crystals as
the analogy -- short range interactions creating long range order.

It should also be noted that feedback can be both positive and negative.
While neither one is inherently good or bad, most examples of
useful feedback are negative feedback... which would imply limiting
a particular specification's scope, size, or amount of detail.

Positive feedback brings to mind an image of 'exploding specifications.' 

Received on Monday, 7 January 2002 12:04:12 UTC