Localizing Views

Concern Oriented Programming seeks to decrease the cost of anticipated areas of change by isolating them from the rest of the code base. Although this reduces the cost of making anticipated changes, unanticipated changes may result in higher costs of program understanding during the reengineering process. Preserving the “chain of evidence,” a software design record semantically linked with annotated code, may help reduce the cost of program understanding and evolution. The semantic facets we emphasize deal with “mechanical” program properties such as concurrency, aliasing, effects, uses, and the like. Answers to questions about these facets enable many program transformations and restructurings. We use the chain of evidence to create “localizing views,” which are dynamically created views addressing a particular concern. These allow a developer to view only the parts of the software artifact that will help in the task at hand. Our vision is that changes can then be made in any of the software lifecycle elements and eventually, with tool assistance, the changes will be propagated through the system.