Quality Engineering: Leveraging Heterogeneous Information - (Invited Talk)

In this paper we present a flexible framework for fine tun- ing the quality of program analysis based on variations, generalizations, and pragmatic extensions of Plotkin's Structured Operational Semantics (SOS). Key to these variations is the idea of Property-Oriented Expan- sion, here the non-standard use of the data component in SOS configura- tions, which ranges from simple abstract interpretations, over arbitrary data flow information, to e.g., temporal constraints. In its most general form, which is characterized by the notion of unifying models, this re- sults in a framework not only for fine-tuning program analysis according to an aspect (quality) of choice, but also for synthesizing orchestrations for service-oriented applications based on loose temporal specifications. From an engineering perspective, the simple interface pattern underly- ing the unifying models approach was key for realizing our experimental platform. Our experimental results, in particular concerning the state ex- plosion problem, indicate that, in practice, limiting the expansion to the previously determined areas of impact suffices to keep the code growth quite moderate.