Quick xing ATL transformations with speculative analysis

Model transformations are central compo- nents of most model-based software projects. While en- suring their correctness is vital to guarantee the quality of the solution, current transformation tools provide lim- ited support to statically detect and x errors. In this way, the identi cation of errors and their correction are nowadays mostly manual activities which incur in high costs. The aim of this work is to improve this situation. Recently, we developed a static analyser that com- bines program analysis and constraint solving to iden- tify errors in ATL model transformations. In this paper, we present a novel method and system that uses our analyser to propose suitable quick xes for ATL transfor- mation errors, notably some non-trivial, transformation- speci c ones. Our approach supports speculative analy- sis to help developers select the most appropriate x by creating a dynamic ranking of xes, reporting on the consequences of applying a quick x, and providing a previsualization of each quick x application. The approach integrates seamlessly with the ATL ed- itor. Moreover, we provide an evaluation based on exist- ing faulty transformations built by a third party, and on automatically generated transformation mutants, which are then corrected with the quick xes of our catalogue
Work supported by the Spanish Ministry of Economyand Competitivity (TIN2014-52129-R), the R&D programme of the Madrid Region (S2013/ICE-3006), and the EU commission (FP7-ICT-2013-10, #611125)