Architectural Mismatch: Problems and Approaches

The papers:

[GAO95]: Architectural Mismatch

[Sha95a]: Architectural Issues in Software Reuse

[Ock95]: Exploiting Structured Data in Wide-Area Information Systems

[SK96]: Experience Assessing an Architectural Approach to Large-Scale Systematic Reuse

Hints:

We turn now to a class of implementation problems that arise when different components make different assumptions about system organization and the types of interactions that will be used. These problems have always plagued system designers, but they are easier to recognize and discuss now that you have a set of architectural concepts and matching vocabulary. Garlan, Allen, and Ockerbloom set out to use several reusable components; they write about the surprises they encountered. Shaw enumerates and classifies many of the techniques used to patch up problems like these. Ockerbloom describes a tool for bridging data-format mismatches. Sullivan and Knight describe their experience integrating OLE components and mediators.

Questions:

1. Briefly explain what Garlan, Allen, and Ockerbloom mean by the term "architectural mismatch".
2. Why don't traditional program development tools (compilers/linkers and the like) solve the problem of detecting mismatches?
3. List two of the specific architectural mismatches that occurred in the development of the Aesop system.
4. Name at least two architectural compromises that Sullivan and Knight had to make in order to construct their system

Email the Current Maintainer

Modified: 04/22/97