[Research Paper] Fine-Grained Model Slicing for Rebel

Eilers, Renate; Hage, Jurriaan; Prasetya, I. S. W. B.; Bosman, Joost

doi:10.1109/scam.2018.00035

Cited by 1 publication

(1 citation statement)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are many opportunities for further research including 1) extend REBEL2 to support deadlock detection, since this is now an impediment to model checking; 2) implement different slicing algorithms (e.g., [35]) and assess the impact in terms of performance and soundness; 3) experiment with learning mocks by incorporating the assumption learning techniques of Cobleigh et al used in assume-guarantee reasoning [34] and 4) provide empirical corroboration of the mocking hypothesis.…”

Section: Discussionmentioning

confidence: 99%

Modeling with Mocking

Stoel

Storm

Vinju

2021

2021 14th IEEE Conference on Software Testing, Verification and Validation (ICST)

View full text Add to dashboard Cite

Writing formal specifications often requires users to abstract from the original problem. Especially when verification techniques such as model checking are used. Without applying abstraction the search space the model checker need to traverse tends to grow quickly beyond the scope of what can be checked within reasonable time.The downside of this need to omit details is that it increases the distance to the implementation. Ideally, the created specifications could be used to generate software from (either manually or automatically). But having an incomplete description of the desired system is not enough for this purpose.In this work we introduce the REBEL2 specification language. REBEL2 lets the user write full system specifications in the form of state machines with data without the need to apply abstraction while still preserving the ability to verify non-trivial properties. This is done by allowing the user to forget and mock specifications when running the model checker. The original specifications are untouched by these techniques.We compare the expressiveness of REBEL2 and the effectiveness of mock and forget by implementing two case studies: one from the automotive domain and one from the banking domain. We find that REBEL2 is expressive enough to implement both case studies in a concise manner. Next to that, when performing checks in isolation, mocking can speed up model checking significantly.

show abstract

Section: Discussionmentioning

confidence: 99%