Collecting Feedback during Software Engineering Experiments
Empirical Software Engineering
A unit test approach for database schema evolution
Information and Software Technology
Hi-index | 0.00 |
Research in schema evolution has been driven by theneed for more effective software development andmaintenance. Finding impacts of schema changes on theapplications and presenting them in an appropriate wayare particularly challenging. We have developed a toolthat finds impacts of schema changes on applications inobject-oriented systems. This tool displays components(packages, classes, interfaces, methods and fields) of adatabase application system as a graph. Componentspotentially affected by a change are indicated bychanging the shape of the boxes representing thosecomponents. Two versions of the tool are available. Oneversion identifies affected parts of applications at thegranularity of packages, classes, and interfaces, whereasthe other version identifies affected parts at the finergranularity of fields and methods. This paper presents thedesign and results of a controlled student experimenttesting these two granularity levels with respect toproductivity and user satisfaction. There are indicationsthat identifying impacts at the finer granularity canreduce the time needed to conduct schema changes andreduce the number of errors. Our results also show thatthe subjects of the experiment appreciated the idea ofvisualizing the impacts of schema changes.