Citation Context

...using the EMFtoCSP tool. As further work, we plan to extend our to approach to cover other transformation languages like QVT. We would also like to study complexity metrics like cyclomatic complexity =-=[18]-=- to establish a limit on the number of test cases that need to be created, something that can be specially useful when testing large transformations. Finally, ATLTest is a first step in the developmen...

Citation Context

...enerate test input models out of ATL model transformations. Our goal is to optimize the generation of the tests by maximizing the coverage of the internal transformation structure. We have chosen ATL =-=[14]-=- as target transformation language due to its popularity (both in academia and industry). However, many of the ideas presented herein could be applied to other ha l-0 07 11 81 9,sv er sio ns1s- 2 9sJu...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

...ome a matter of utmost importance. Among the possible strategies to improve the quality of model transformations, several testing techniques for model transformations have been recently proposed (see =-=[4]-=- for a recent survey). So far, most of the techniques follow a black-box approach (i.e. transformations are regarded as a black-box so the generation of test input models does not take into account th...

Citation Context

...roaches facing this challenge are based on black-box techniques [11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]. As far as we know only two white-box approaches for transformation testing have been proposed =-=[9, 15]-=-. Both address the identification of the relevant parts of the input metamodel to be exercised by the tests: by looking at the transformation definition they detect the subset of the metamodel (and po...

Citation Context

...s essentially the same as for software testing and, therefore the same conclusions can be applied. However, compared to program testing, model transformation testing must face an additional challenge =-=[3]-=-: the complex nature of model transformation inputs and outputs. Models can be large structures and must conform to a meta-model (possibly extended with OCL well-formedness rules) thus making even har...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

... to the source metamodel and satisfying the constraints for the test case. This computation can be performed using any of the SAT-based or CSP-based solvers available. In particular, we use EMFtoCSP1 =-=[12]-=- to generate the input test models. EMFtoCSP is an Eclipse2-integrated tool for the automatic verification of UML models and EMF models annotated with OCL constraints by means of reexpressing them as ...

Citation Context

...CL expressions is not exclusive of model transformation testing, on the contrary, it is also an important method for the verification and validation of UML/OCL specifications. Regarding this, [7] and =-=[2]-=- propose approaches to generate test data from OCL specifications, based on the utilisation of Higher-Order Logic and constraint solving techniques, respectively. Another approach based on the utilisa...

Citation Context

... to find the errors (whose presence is assumed) in a program or piece of software, of course, with the intent of establishing some degree of confidence that the program does what it is expected to do =-=[19]-=-. A common methodology to test a piece of software generally comprises a number of well known steps, namely the creation of input test cases, running the software with the test cases, and finally, usi...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

... = {(Gc(V )[last], 1)} ∪Gc(E) 9 c→ excludes(o) V = {c→ excludes(o)} ∪Gc(V ), E = {(Gc(V )[last], 1)} ∪Gc(E) 10 c→ includesAll(cl) V = {c→ includesAll(cl)} ∪Gc(V ) ∪Gcl(V ), E = {(Gc(V )[last], Gcl(V )=-=[1]-=-), (Gcl(V )[last], 1)}∪ Gc(E) ∪Gcl(E) 11 c→ excludesAll(cl) V = {c→ excludesAll(cl)} ∪Gc(V ) ∪Gcl(V ) E = {(Gc(V )[last], Gcl(V )[1]), (Gcl(V )[last], 1)}∪ Gc(E) ∪Gcl(E) 12 c→ union(cl) V = Gc(V ) ∪Gc...

Citation Context

...anches in the transformation, this way maximizing their effectiveness. White-box techniques can also be used in coverage analysis, to measure the quality of the generated test models. Regarding this, =-=[17]-=- proposes a number of white-box coverage measures for ATL transformations, namely rule coverage, instruction coverage and decision coverage, that are used to check how a number of test cases cover ATL...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

... on the utilisation of constraint solving techniques is proposed in [1] to generate test cases out of UML specifications, although in this case only a limited subset of the OCL is supported. Finally, =-=[23]-=- proposes an approach to evaluate the quality of test cases generated from OCL expressions based on the utilization of several coverage criteria. 7 Conclusions and Future Work We have presented ATLTes...

Citation Context

...out of OCL expressions is not exclusive of model transformation testing, on the contrary, it is also an important method for the verification and validation of UML/OCL specifications. Regarding this, =-=[7]-=- and [2] propose approaches to generate test data from OCL specifications, based on the utilisation of Higher-Order Logic and constraint solving techniques, respectively. Another approach based on the...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

...ortant tasks when testing a model transformation is the creation of an adequate set of test input models. Currently, the majority of approaches facing this challenge are based on black-box techniques =-=[11, 9, 10, 16, 21, 22, 3, 6, 20, 8, 13]-=-. As far as we know only two white-box approaches for transformation testing have been proposed [9, 15]. Both address the identification of the relevant parts of the input metamodel to be exercised by...

Citation Context

...results yielded to determine whether errors came up or not. An oracle is any program, process or body of data that specifies the expected outcome for a set of test cases as applied to a tested object =-=[5]-=- and it can be as simple as a manual inspection or as complex as a separate piece of software. It is generally accepted that the more input tests are created and the more time is spent running the sof...