Metrics to identify where object-oriented program comprehension benefits from the runtime structure

To evolve object-oriented code, developers often need to understand both the code structure in terms of classes and packages, as well as the runtime structure in terms of abstractions of objects. Recent empirical studies have shown that for some code modifications tasks, developers do benefit from having access to information about the runtime structure. However, there is no good sense of when object-oriented program comprehension clearly depends on information about the runtime structure. We propose using metrics to identify cases in object-oriented program comprehension that benefit from information about the runtime structure. The metrics relate properties observed on a statically extracted hierarchical object graph to the type structures declared in the code and highlight key differences between the runtime structure and the code structure.