Improvements on the Online Testing with T-UppAal: Coverage Measurement and Re-runs
Authors
Hall, Gunnar ; Kordy, Piotr ; Vitkauskaite, Dalia
Term
10. Term
Education
Publication year
2005
Abstract
T-UppAal er et online testværktøj, der tester, om indlejrede realtidssystemer følger deres modeller, ved hjælp af modelbaseret black-box overensstemmelsestest. Med andre ord kontrollerer det, om en tidskritisk enhed opfører sig som specificeret, uden at kigge i den interne kode. Hidtil var det ikke undersøgt, hvordan man bedst gentager tests (re-runs) og måler, hvad testene faktisk dækker (coverage) i T-UppAal. Denne afhandling beskriver klare kriterier for at vælge, hvilke tests der bør gentages, analyserer hvor godt kriterierne fungerer i praksis i T-UppAal, og foreslår dækningsmål til at vurdere testkvalitet. Vi evaluerer udvalgte re-run-strategier og dækningsmål i et industrielt casestudie. For at støtte brugere udviklede vi værktøjet Butler, som hjælper med at opsætte re-runs og beregne dækning i T-UppAal.
T-UppAal is an online tool that tests whether embedded real-time systems follow their models, using model-based, black-box conformance testing. In simple terms, it checks if a time-critical device behaves as specified without looking at its internal code. Until now, T-UppAal had no systematic study of how to repeat tests (re-runs) and how to measure what the tests actually exercised (coverage). This thesis defines clear criteria for choosing which tests to re-run, analyzes how well these criteria work in practice within T-UppAal, and proposes coverage measures to assess test quality. We evaluate selected re-run strategies and coverage measures in an industrial case study. To support practitioners, we developed Butler, a companion tool that helps set up re-runs and compute coverage for T-UppAal.
[This abstract was generated with the help of AI]
Documents