Download presentation
Download article

The idea of our presentation is to expose challenges of behavior driven development (BDD) automation for mapping use cases written in narrative manner to unit tests. The BDD creates a kind of a communication framework that allows the developers to rediscover the customer context better in the process of software design and testing. Hence testing is considered here to be a sort of cooperative work: the customers propose user side-test scenarios, while the testers write stories and map them to the source code constructions. It contrasts with the test-first strategy that rather helps developers to communicate clearly with their teammates, not with the stakeholders.

As well as test-first techniques, the BDD approach is maturing, and there are still debates about definitions, application objects, characteristics, usage recommendations and effects. Particularly, we propose to discuss existing BDD-inspired toolkits aimed to facilitate using BDD approach. We noticed that even in cases where mapping rules are well defined, most test systems and BDD implementations lack the support of automatic transition from plain text stories to test classes and acceptance tests maintenance, and need deeper integration with developing environments.

One of the most serious drawbacks of BDD toolkits lies in their inability to convert natural language-based user stories into executable tests. Unfortunately, we have to accept this situation, since such a conversion requires translation of informal natural language constructions into formal statements of a programming language and thus can be considered a variation of programming, which is a task for a human expert.

Even the tools that are advertised as “supporting plain-text user stories” in fact set certain restrictions on the structures of stories. The latter observation makes us believe that the process of user stories conversion can be automated further, at least, to some extent. While automatic story conversion is infeasible, individual elements of computer-aided conversion are certainly doable with the help of modern natural language processing methods and simple heuristic procedures.

As a conclusion we propose requirements check list for further analysis of the BDD and IDEs integration solutions.

Evgeny Pyshkin

Evgeny Pyshkin

Ph.D., Associate Professor, Vice-dean (International Programs), Department of Computer Systems and Software Engineering, Saint-Petersburg State Polytechnical University

Evgeny Pyshkin teaches courses on programming and object-oriented analysis and design. He is author and instructor of intensive courses “Introduction to the Java Language and Technology” and “Microsoft.Net Framework Programming” in the Central Ostrobothnia University (Finland), “Object-Oriented Software Engineering” in the University of Aizu (Japan).
He is an author of several books on software technologies including the titles both in Russian and English and more than 40 research papers in the areas of design automation, software technologies and information retrieval.

Official Web-page: http://kspt.ftk.spbstu.ru/info/staff/pyshkin/en/

Maxim Mozgovoy

Maxim Mozgovoy

Associate professor, University of Aizu

Maxim Mozgovoy is an associate professor at the University of Aizu (Aizu-Wakamatsu, Japan), where he studies and develops practical game-oriented observation-based AI systems. The main purpose of his research is to demonstrate the advantages of machine learning and case-based reasoning over traditional approaches to game AI development that often require enormous handwork.

His other research interests are focused around natural language processing technologies. In particular, he is currently working on a “virtual language lab” that will combine natural language processing algorithms with a computer-assisted language learning environment.

He is an author of several books on programming and computer science.

Mikhail Glukhikh

Mikhail Glukhikh

Ph.D in Computer Science, Associate professor, Department of Computer Systems and Software Engineering, Saint-Petersburg State Polytechnical University

Mikhail Glukhikh is a lecturer of academic courses of programming, concepts of programming languages, systems and devices dependability.

He participates in research in the area of static code analysis. He is an author of Aegis defect detection tool. He is an author of 30 scientific papers, 10 of them are in the areas of software engineering. Other scientific interests include reliable systems synthesis and analysis and high-performance system programming.

Official Web-page: http://kspt.ftk.spbstu.ru/info/staff/glukhikh/en/