E-Assessment mit Graja – ein Vergleich zu Anforderungen an Softwaretestwerkzeuge

Die automatisierte Bewertung studentischer Übungsabgaben in Programmieren-Lehrveranstaltungen weist Parallelen zum automatisierten Test in der professionellen Softwareentwicklung auf. Allerdings muss ein Autobewerter (Grader), um lernförderlich zu sein, andere Zielsetzungen erfüllen als üblicherweise im professionellen Softwaretest eingesetzte Analyse- und Testwerkzeuge. Dieser Beitrag identifiziert wesentliche Unterschiede und beschreibt, wie sich diese Unterschiede in dem an der Hochschule Hannover entwickelten und seit mehreren Jahren im Einsatz befindlichen Autobewerter "Graja" niederschlagen.

Instant Storyboarding

This thesis aims at empowering software customers with a tool to build software tests them selves, based on a gradual refinement of natural language scenarios into executable visual test models. The process is divided in five steps: 1. First, a natural language parser is used to extract a graph of grammatical relations from the textual scenario descriptions. 2. The resulting graph is transformed into an informal story pattern by interpreting structurization rules based on Fujaba Story Diagrams. 3. While the informal story pattern can already be used by humans the diagram still lacks technical details, especially type information. To add them, a recommender based framework uses web sites and other resources to generate formalization rules. 4. As a preparation for the code generation the classes derived for formal story patterns are aligned across all story steps, substituting a class diagram. 5. Finally, a headless version of Fujaba is used to generate an executable JUnit test. The graph transformations used in the browser application are specified in a textual domain specific language and visualized as story pattern. Last but not least, only the heavyweight parsing (step 1) and code generation (step 5) are executed on the server side. All graph transformation steps (2, 3 and 4) are executed in the browser by an interpreter written in JavaScript/GWT. This result paves the way for online collaboration between global teams of software customers, IT business analysts and software developers.

Geração de casos de teste a partir de especificações B

With the increasing complexity of software systems, there is also an increased concern about its faults. These faults can cause financial losses and even loss of life. Therefore, we propose in this paper the minimization of faults in software by using formally specified tests. The combination of testing and formal specifications is gaining strength in searches mainly through the MBT (Model-Based Testing). The development of software from formal specifications, when the whole process of refinement is done rigorously, ensures that what is specified in the application will be implemented. Thus, the implementation generated from these specifications would accurately depict what was specified. But not always the specification is refined to the level of implementation and code generation, and in these cases the tests generated from the specification tend to find fault. Additionally, the generation of so-called "invalid tests", ie tests that exercise the application scenarios that were not addressed in the specification, complements more significantly the formal development process. Therefore, this paper proposes a method for generating tests from B formal specifications. This method was structured in pseudo-code. The method is based on the systematization of the techniques of black box testing of boundary value analysis, equivalence partitioning, as well as the technique of orthogonal pairs. The method was applied to a B specification and B test machines that generate test cases independent of implementation language were generated. Aiming to validate the method, test cases were transformed manually in JUnit test cases and the application, created from the B specification and developed in Java, was tested. Faults were found with the execution of the JUnit test cases

Uma abordagem para a verificação do comportamento excepcional a partir de regras de designe e testes

Checking the conformity between implementation and design rules in a system is an important activity to try to ensure that no degradation occurs between architectural patterns defined for the system and what is actually implemented in the source code. Especially in the case of systems which require a high level of reliability is important to define specific design rules for exceptional behavior. Such rules describe how exceptions should flow through the system by defining what elements are responsible for catching exceptions thrown by other system elements. However, current approaches to automatically check design rules do not provide suitable mechanisms to define and verify design rules related to the exception handling policy of applications. This paper proposes a practical approach to preserve the exceptional behavior of an application or family of applications, based on the definition and runtime automatic checking of design rules for exception handling of systems developed in Java or AspectJ. To support this approach was developed, in the context of this work, a tool called VITTAE (Verification and Information Tool to Analyze Exceptions) that extends the JUnit framework and allows automating test activities to exceptional design rules. We conducted a case study with the primary objective of evaluating the effectiveness of the proposed approach on a software product line. Besides this, an experiment was conducted that aimed to realize a comparative analysis between the proposed approach and an approach based on a tool called JUnitE, which also proposes to test the exception handling code using JUnit tests. The results showed how the exception handling design rules evolve along different versions of a system and that VITTAE can aid in the detection of defects in exception handling code

An Experience-Driven Pedagogy for the Instruction of Software Testing in Computer Science

In the realm of computer programming, the experience of writing a program is used to reinforce concepts and evaluate ability. This research uses three case studies to evaluate the introduction of testing through Kolb's Experiential Learning Model (ELM). We then analyze the impact of those testing experiences to determine methods for improving future courses. The first testing experience that students encounter are unit test reports in their early courses. This course demonstrates that automating and improving feedback can provide more ELM iterations. The JUnit Generation (JUG) tool also provided a positive experience for the instructor by reducing the overall workload. Later, undergraduate and graduate students have the opportunity to work together in a multi-role Human-Computer Interaction (HCI) course. The interactions use usability analysis techniques with graduate students as usability experts and undergraduate students as design engineers. Students get experience testing the user experience of their product prototypes using methods varying from heuristic analysis to user testing. From this course, we learned the importance of the instructors role in the ELM. As more roles were added to the HCI course, a desire arose to provide more complete, quality assured software. This inspired the addition of unit testing experiences to the course. However, we learned that significant preparations must be made to apply the ELM when students are resistant. The research presented through these courses was driven by the recognition of a need for testing in a Computer Science curriculum. Our understanding of the ELM suggests the need for student experience when being introduced to testing concepts. We learned that experiential learning, when appropriately implemented, can provide benefits to the Computer Science classroom. When examined together, these course-based research projects provided insight into building strong testing practices into a curriculum.

Using Dynamic Analysis for API Migration

When changing the API of a framework, we need to migrate its clients. This is best done automatically. In this paper, we focus on API migration where the mechanism for inversion of control changes. We propose to use dynamic analysis for such API migration since structural refactorings alone are often not sufficient. We consider JExample as a case-study. JExample extends JUnit with first-class dependencies and fixture injection. We investigate how dynamically collected information about test coverage and about instances under test can be used to detect dependency injection candidates.

JExample: Exploiting Dependencies Between Tests to Improve Defect Localization

To quickly localize defects, we want our attention to be focussed on relevant failing tests. We propose to improve defect localization by exploiting dependencies between tests, using a JUnit extension called JExample. In a case study, a monolithic white-box test suite for a complex algorithm is refactored into two traditional JUnit style tests and to JExample. Of the three refactorings, JExample reports five times fewer defect locations and slightly better performance (-8-12\%), while having similar maintenance characteristics. Compared to the original implementation, JExample greatly improves maintainability due the improved factorization following the accepted test quality guidelines. As such, JExample combines the benefits of test chains with test quality aspects of JUnit style testing.

Tool support for executable documentation of Java class hierarchies

While object-oriented programming offers great solutions for today's software developers, this success has created difficult problems in class documentation and testing. In Java, two tools provide assistance: Javadoc allows class interface documentation to be embedded as code comments and JUnit supports unit testing by providing assert constructs and a test framework. This paper describes JUnitDoc, an integration of Javadoc and JUnit, which provides better support for class documentation and testing. With JUnitDoc, test cases are embedded in Javadoc comments and used as both examples for documentation and test cases for quality assurance. JUnitDoc extracts the test cases for use in HTML files serving as class documentation and in JUnit drivers for class testing. To address the difficult problem of testing inheritance hierarchies, JUnitDoc provides a novel solution in the form of a parallel test hierarchy. A small controlled experiment compares the readability of JUnitDoc documentation to formal documentation written in Object-Z. Copyright (c) 2005 John Wiley & Sons, Ltd.

Bewertungsaspekte und Tests in Java-Programmieraufgaben für Graja im ProFormA-Aufgabenformat

Ein Autobewerter für von Studierenden eingereichte Programme führt die im ProFormA-Aufgabenformat sequentiell spezifizierten "Tests" aus, um die Einreichung zu prüfen. Bzgl. der Interpretation und Darstellung der Testausführungsergebnisse gibt es derzeit keinen graderübergreifenden Standard. Wir beschreiben eine Erweiterung des ProFormA-Aufgabenformats um eine Hierarchie von Bewertungsaspekten, die nach didaktischen Aspekten gruppiert ist und Referenzen auf die Testausführungen besitzt. Die Erweiterung wurde in Graja umgesetzt, einem Autobewerter für Java-Programme. Je nach gewünschter Detailaufschlüsselung der Bewertungsaspekte müssen in der Konsequenz Testausführungen in Teilausführungen aufgebrochen werden. Wir illustrieren unseren Vorschlag unter Einsatz der Testwerkzeuge Compiler, dynamischer Softwaretest, statische Analyse sowie unter Einsatz menschlicher Bewerter.

Graja - Autobewerter für Java-Programme

In diesem Bericht wird der Autobewerter Graja für Java-Programme vorgestellt. Wir geben einen Überblick über die unterstützten Bewertungsmethoden sowie die beteiligten Nutzerrollen. Wir gehen auf technische Einzelheiten und Randbedingungen der in Graja eingesetzten Bewertungsmethoden ein und zeigen die Einbindung von Graja in eine technische Gesamtarchitektur. An einem durchgehenden Beispiel stellen wir die Struktur einer Programmieraufgabe sowie die von Graja unterstützten Feedback-Möglichkeiten dar. Informationen zum bisherigen Einsatz des Graders runden den Bericht ab.