One of the innovations of the new ISE Eiffel 5 environment is a set of metric facilities enabling developers and managers to obtain quantitative information about software systems and the process of producing them.
Find out basic
quantitative properties of a system, such as the number of classes,
the number of features, how many times the system has been compiled
and many others.
Define new properties as mathematical combinations of the basic ones, and compute the resulting values, to get the answers to such questions as what percentage of my system’s routines have precondition clauses? or what is the average number of features, lines, or invariant clauses per class?
Vary the scope of such measurements, so that the measured result may cover a single feature, a class, a cluster, an entire system or an archive of another system to make comparisons with other projects.
Compare the result of the same metric over several of these scopes, to see for example how a project differs from those on record, or whether some parts of a project depart from the norms applied in others.
The environment will support such measurements through a simple graphical interface, closely integrated with the rest of the Eiffel Explorer. Users of the environment can start applying these facilities to their systems without learning any theory.
Metrics, however, are notoriously subject to abuse; we must be wary of the temptation to revere numbers just because they are numbers (lies, damn lies and metrics). This note presents a simple approach to software metrics designed to establish a clear and sound basis, so that we know what we are measuring, why we are measuring it, and how much value we may attach to the results.