Evaluation of the QVT Merge Language Proposal
Grønmo, Roy; Aagedal, Jan Øyvind; Solberg, Arnor; Belaunde, Mariano; Rosenthal, Peter; Faugere, Madeleine; Ritter, Tom; Born, Marc
Research report
Permanent lenke
http://hdl.handle.net/11250/2388147Utgivelsesdato
2005Metadata
Vis full innførselSamlinger
- Publikasjoner fra CRIStin - SINTEF AS [6009]
- SINTEF Digital [2569]
Originalversjon
SINTEF Rapport A312, 17 p. SINTEF , 2005Sammendrag
STF90 A05045This report has identified 29 weighted evaluation criteria representing desired properties of a model to model transformation language. These criteria have been used to evaluate the current QVT Merge specification. We have so far only been able to evaluate 21 of these criteria, mainly due to missing tool support. Some of the criteria are considered absolute in the sense that missing to fulfil such a criterion is considered a failure. The 21 evaluated criteria give a score of 59 out of a maximum possible score of 68 (language-based + example-based testing). We have also compared the QVT-Merge submission with the QVT-Compuware/Sun submission and at the time being the QVT-Merge seems to be the preferred one due to more support on the absolute criteria and better easy-to-use score. Eight transformation examples for solving six different transformation tasks have given a lot of insight on the ease of use criteria for both simple and complex transformations. When defining transformations using QVT Merge we believe that a lot of effort may be required in order to define the source and target metamodels. The evaluation in this report could be improved by using the reference examples with alternative approaches published in the literature. An available QVT-Merge tool is necessary in order to provide evaluations of all the suggested criteria. In order to further investigate the usability of the graphical notation, we need to define more of the transformation examples graphically. Only one of the examples has been specified graphically in this version. The current evaluation has been done by a single evaluator who has only reviewed the transformation code that was written by somebody else. The evaluation will be further improved by incorporating input from other evaluators as well as evaluation from those who wrote the transformation code.
Oppdragsgiver: EU Commission
Beskrivelse
-