Baris Güldali

Towards model-driven unit testing

The Model-Driven Architecture (MDA) approach for constructing software systems advocates a stepwise refinement and transformation process starting from high-level models to concrete program code. In contrast to numerous research efforts that try to generate executable function code from models, we propose a novel approach termed model-driven monitoring. On the model level the behavior of an operation is specified with a pair of UML composite structure diagrams (visual contract), a visual notation for pre- and post-conditions. The specified behavior is implemented by a programmer manually. An automatic translation from our visual contracts to JML assertions allows for monitoring the hand-coded programs during their execution. In this paper we present how we extend our approach to allow for model-driven unit testing, where we utilize the generated JML assertions as test oracles. Further, we present an idea how to generate sufficient test cases from our visual contracts with the help of model-checking techniques.

Assuring Consistency of Business Process Models and Web Services Using Visual Contracts

Business process models describe workflows by a set of actions together with their ordering. When implementing business processes within a service-oriented architecture, these actions are mapped to existing IT (web) services, which are then to be executed in the order specified by the business process. However, the execution of a web service can require certain preconditions to be fulfilled. These might not hold at the time of execution specified in the business process model: it can be inconsistent with the web service specification. In this paper we propose a technique for checking consistency of process models with web service specifications. To this end, both are equipped with a formal semantics (in terms of graph transformations). We show how to use an existing model checker for graph transformation systems to carry out the consistency check.

From design models to test models by means of test ideas

Model-Based Testing is slowly becoming the next level of software testing. It promises higher quality, better coverage and efficient change management. MBT shows two main problems of modeling the test behavior. While modeling test cases test designers rewrite most of the system specification. Further, the number of test cases generated by modern tools is often not feasible. In practice, both problems are not solved. Assuming that the functional design is based on models, we show how to use them for software testing. With so-called test ideas, we propose a way to manually select and automatically transform the relevant parts of the design model into a basic test model that can be used for test case generation. We give an example and discuss the potentials for tool support.

Software Testing via Model Checking

Testing is a necessary, but costly process for user-centric quality control. Moreover, testing is not comprehensive enough to completely detect faults. Many formal meth ods have been pro posed to avoid the drawbacks of testing, e.g., model checking that can be automatically carried out. This paper presents an approach that (i) generates test cases from the specification and (ii) transfers the specification-oriented testing process to model checking. Thus, the approach combines the advantages of testing and model checking assuming the availability of (i) a model that specifies the ex pected, desirable system behavior as required by the user and (ii) a second model that describes the system behavior as observed. The first model is complemented in also specifying the undesirable system properties. The approach analyzes both these specification models to generate test cases that are then converted into temporal logic formulae to be model checked on the second model.

Model-Based System Testing Using Visual Contracts

In system testing the system under test (SUT) is tested against high-level requirements which are captured at early phases of the development process. Logical test cases developed from these requirements must be translated to executable test cases by augmenting them with implementation details. If manually done these activities are error-prone and tedious. In this paper we introduce a model-based approach for system testing where we generate first logical test cases from use case diagrams which are partially formalized by visual contracts, and then we transform these to executable test cases using model transformation. We derive model transformation rules from the design decisions of developers.

Semi-automated Test Planning for e-ID Systems by Using Requirements Clustering

In acceptance testing, customer requirements as specified in system specifications have to be tested for their successful implementation. This is a time-consuming task due to inherent system complexity and thus a large number of requirements. In order to reduce efforts in acceptance testing, we introduce a novel approach that exploits redundancies and implicit relations in requirements specifications, which are based on multi-viewpoint techniques, in our case the reference model for open distributed processing (RM-ODP). It deploys requirements clustering and linguistic analysis techniques for reducing the total number of test cases. We report on concrete experiences with this approach within joint R&D work of the Software Quality Lab (s-lab) of the University of Paderborn and HJP Consulting, an international consulting company, specialized in planning, procurement and acceptance testing of national electronic identification (e-ID) systems. The paper is concluded with an overview on the current tool support especially for automated detection of the redundancies and implicit relations in requirements. Also the future work on the tool support for the overall test specification process is discussed.

A holistic approach to test-driven model checking

Testing is the most common validation method in the software industry. It entails the execution of the software system in the real environment. Nevertheless, combination of formal methods and test methods has been widely advocated. Model checking be longs to the promising candidates for this marriage. The present paper modifies and extends the existing approaches in that, after the test case generation, a model checking step supports the manual test process. Based on the holistic approach to specification-based construction of test suites, this paper proposes to generate test cases to cover both the specification model and its complement. This helps also to clearly differentiate the correct system outputs from the faulty ones as the test cases based on the specification are to succeed the test, and the ones based on the complement of the specification are to fail. Thus, the approach handles the oracle problem in an effective manner.

TORC: test plan optimization by requirements clustering

Acceptance testing is a time-consuming task for complex software systems that have to fulfill a large number of requirements. To reduce this effort, we have developed a widely automated method for deriving test plans from requirements that are expressed in natural language. It consists of three stages: annotation, clustering, and test plan specification. The general idea is to exploit redundancies and implicit relationships in requirements specifications. Multi-viewpoint techniques based on RM-ODP (Reference Model for Open Distributed Processing) are employed for specifying the requirements. We then use linguistic analysis techniques, requirements clustering algorithms, and pattern-based requirements collection to reduce the total effort of testing against the requirements specification. In particular, we use linguistic analysis for extracting and annotating the actor, process and object of a requirements statement. During clustering, a similarity function is computed as a measure for the overlap of requirements. In the test plan specification stage, our approach provides capabilities for semi-automatically deriving test plans and acceptance criteria from the clustered informal textual requirements. Two patterns are applied to compute a suitable order of test activities. The generated test plans consist of a sequence of test steps and asserts that are executed or checked in the given order. We also present the supporting prototype tool TORC, which is available open source. For the evaluation of the approach, we have conducted a case study in the field of acceptance testing of a national electronic identification system. In summary, we report on lessons learned how linguistic analysis and clustering techniques can help testers in understanding the relations between requirements and for improving test planning.

Leichtgewichtige Traceability im agilen Entwicklungsprozess am Beispiel von Scrum

Motivation Einer der wichtigsten Erfolgsfaktoren der agilen Softwareentwicklung ist die schnelle und unkomplizierte Verteilung von Informationen. Dabei reicht das Spektrum der Informationsverteilung von einfachen Dokumenten uber Wikis, Videos und Telefonaten bis hin zum “Face-to-Face” Gesprach. In der traditionellen Softwareentwicklung lag der Fokus typischerweise auf einer dokument-basierten Erfassung und Verteilung von Informationen. Explizite Dokumentation beruht auf der Annahme, dass sich die festgehaltenen Informationen nur in bestimmtem Mase andern. Da dieser Umstand insbesondere in der Softwareentwicklung nicht automatisch gegeben ist, hat man dies im Kontext der agilen Vorgehensweisen als ein Kernproblem von schwergewichtigen Prozessen identifiziert. Als Konsequenz dazu wird in der agilen SoftwareEntwicklung der Umfang der Dokumentation minimiert. Dies spiegelt sich im Manifest fur Agile Software-Entwicklung wider (s. agilemanifesto.org): Obwohl die umfassende Dokumentation als wichtig erachtet wird, wird der Wert funktionierender Software hoher eingeschatzt. Nach den zwolf inizialen Prinzipien hinter dem Agilen Manifest gilt die direkte personliche Ubermittlung als effizientestes und effektivstes Verfahren. Funktioniert diese Kompensation besonders gut solange die Entwickler moglichst nah zusammen arbeiten, so ergeben sich jedoch Herausforderungen beim Versuch, agile Entwicklung in mehreren Teams und Standorten zu realisieren. Insbesondere fehlt es in der aktuellen Praxis an Techniken, welche die Dokumentation und Traceability der Entwicklungsartefakte optimal unterstutzen. Somit kann es in der Praxis vorkommen, dass Anforderungen oder Entwurfsentscheidungen dem Rest des Teams nicht mitgeteilt werden und diese Wissenslucke die weiteren Implementierungsund Testaktivitaten negativ beeinflusst.

Sichern der Zukunftsfähigkeit bei der Migration von Legacy-Systemen durch modellgetriebene Softwareentwicklung

Softwareunternehmen stehen zunehmend vor der Herausforderung, grose betriebliche Informationssysteme, die mit Softwaretechnologien wie den 4. Generationssprachen (4GL) und -werkzeugen aus den 90er Jahren entwickelt wurden, durch moderne Systeme mit mehrschichtigen Architekturen abzulosen, die auf fortgeschrittenen Softwaretechnologien aufsetzen. Fur sie gibt es leistungsfahige Entwicklungsplattformen, und die neuen Technologien und Architekturen erlauben es, die Systeme einfacher zu warten und sie leichter an sich andernde Anforderungen anzupassen. Das erhoht die Zukunftsfahigkeit der neuen Systeme. Gleichzeitig sind die Softwareunternehmen bestrebt, die mit viel Aufwand entwickelte Funktionalitat der Legacy-Systeme wieder zu verwenden.