Unresolved Issues

In a single day, one cannot realistically explore the value of a model; at best, one can determine there is some utility. A number of obvious issues remain:

• Does a transformational model apply to other generic architecture systems?

• More detailed models exist (c.f. [Baxter92], covering explicit performance constraints and providing considerable flexibility in the explicit statement of control knowledge). How well do they hold up, and what, if any, extra insights might they provide?

• How does a transformational view explain architecture of components for systems?

• How does it explain architectures of the generated systems?

• How does the composition mechanism affect the components of generic architecture?

• Are there alternative hypotheses that explain the commonalities better?

The main value of the transformational perspective is to allow comparison of features of the mechanisms used. Such a perspective can be used to determine the robustness of the composition process, how complete the specification language is for both problem and components, and to determine what mechanisms a particular generic architecture system may need.