Traditionally, end-to-end testing of mobile apps is either performed manually or automated with test scripts. However, manual GUI testing is expensive and slow, and test scripts are fragile for GUI changes, resulting in high maintenance costs. Scriptless testing attempts to address the costs associated with GUI testing. Existing scriptless approaches for mobile testing do not seem to fit the requirements of the industry, specifically those of the ING. This study presents an extension to open source TESTAR tool to support scriptless GUI testing of Android and iOS applications. We present an initial validation of the tool on an industrial setting at the ING. From the validation, we determine that the extended TESTAR outperforms two other state-of-the-art scriptless testing tools for Android in terms of code coverage, and achieves similar performance as the scripted test automation already in use at the ING. Moreover, we see that the scriptless approach covers parts of the application under test that the existing test scripts did not cover, showing the complementarity of the approaches, providing more value for the testers.

The sound identification of refactoring opportunities is still an open problem in software engineering. Recent studies have shown the effectiveness of machine learning models in recommending methods that should undergo different refactoring operations. In this work, we experiment with such approaches to identify methods that should undergo an Extract Method refactoring, in the context of ING, a large financial organization. More specifically, we (i) compare the code metrics distributions, which are used as features by the models, between open-source and ING systems, (ii) measure the accuracy of different machine learning models in recommending Extract Method refactorings, (iii) compare the recommendations given by the models with the opinions of ING experts. Our results show that the feature distributions of ING systems and open-source systems are somewhat different, that machine learning models can recommend Extract Method refactorings with high accuracy, and that experts tend to agree with most of the recommendations of the model.

Developing complex (sub)systems is a multi-disciplinary activity resulting in several, complementary models, possibly on different abstraction levels. The relations between all these models are usually loosely defined in terms of informal documents. It is not uncommon that only till the moment of integration at implementation level, shortcomings or misunderstanding between the different disciplines is revealed. In order to keep models consistent and to reason about multiple models, the relations between models have to be formalized. MultiDisciplinary System Engineering (MDSE) ecosystems provide a means for this. These ecosystems formalize the domain of interest using Domain Specific Languages (DSLs), and formalize the relations between models by means of automated model transformations. This enables consistency checking between domain and aspect models and facilitates multi-disciplinary analysis of the single (sub)system at hand. MDSE ecosystems provide the means to analyze a single (sub)system model. A set of models of different (sub)systems can be analyzed to derive best modeling practices and modeling patterns, and to measure whether a MDSE ecosystem fulfills its needs. The MDSE ecosystem itself can be instrumented to analyze how the MDSE ecosystem is used in practice. The evolution of models, DSLs and complete MDSE ecosystems is studied to identify and develop means that support evolution at minimal costs while maintaining high quality. In this paper, we present the anatomy of MDSE ecosystems with industrial examples, the ongoing work to enable the various types of analysis, each with their dedicated purpose. We conclude with a number of future research directions.

The increasing number of languages used to engineer complex systems causes challenges to the development and maintenance processes of these languages. In this paper, we reflect on our experience in developing real life complex cyber-physical systems by using MDE techniques and DSLs. Firstly, we discuss a number of industrial challenges in the modeling software engineering domain. To address these challenges, we propose the concept of language architecture as an organizational principle for designing, reusing and maintaining DSLs and their infrastructure. Based on this, a metamodel for a DSL is designed and a tool support (LanArchi) is developed. Finally the possible future directions are given.

Recently modeling techniques are introduced to support safety assessment. Goal Structural Notation is one of these modeling techniques, which can be used to facilitate the development of safety argumentation and create reusable safety argumentation models. Consequently, GSN-based safety cases are widely used to demonstrate the safety of systems in safety-critical domains. Due to the amount of manual work, constructing a safety case is usually time-consuming. Moreover, the re-usability of GSN-based safety cases is limited. To address this, safety case patterns are introduced to support safety case reuse. As more and more GSN-based safety cases and patterns are designed with different goals in different contexts, it becomes hard to identify a reusable safety case or pattern. In this paper, we carried out a study on the categorization of existing GSN-based safety cases and patterns. As a result, a number of high cited publications are selected and studied. Finally a categorizatio n of GSN-based safety cases is proposed. A clear categorization of GSN-based safety cases can be used to identify similar safety cases or patterns and facilitate safety case reuse.