Traditional software metrics have been static metrics which measure various aspects of the software at compile time. In our research we have examined a very new field of software engineering research, that is, dynamic or runtime metrics that measure object oriented software quality in terms of cohesion and complexity when the code is executing. We define a suite of 2 dynamic complexity metrics and 4 dynamic cohesion metrics that were collected at runtime, compare them to Bug Data, Human Data, static metrics and (for cohesion) to another dynamic cohesion metric to validate them. Our results indicate that several of our dynamic metrics are actually measuring complexity or cohesion, as appropriate, and also shows there is some similarity between various dynamic metrics. Since the Dynamic metric research area is still in its infancy, we note the absence of established validation criteria in comparing Dynamic metrics to other static and dynamic metrics.
展开▼