Detail

Agile methodologies and test driven development: report on two empirical studies

The Agile movement is having a major impact on theory and practice of software development.
However, its applicability and effectiveness in industrial context is still controversial, with very limited empirical evidence.
The outline of the talk is the following:
-an overview of agile methodologies and eXtreme Programming in particular.

-a critical analysis of current empirical evidence on agile methodologies, both in academic and industrial environment

-a report on an empirical study about test Driven Development (one of the XP practices) performed with students at
Politecnico di Torino. Test-Driven Development (TDD) is based on formalizing a piece of functionality as a test, implementing it such that the test passes, and iterating the process, gradually adding functionality. We performed a controlled experiment for empirically evaluating TDD conducted with undergraduate students. While the experiment group applied TDD, the control group applied a more conventional programming technique by writing tests after the implementation. Both groups followed an incremental process, adding new features one at a time and regression testing them. We found that TDD students wrote more tests, and in turn, students who wrote more tests were more productive. We observed that the minimum quality increased linearly with the number of tests: the quality scores of the students who wrote an equal number of tests remained above a certain minimum level, and this minimum level was higher for students who wrote more tests, regardless of the technique applied.

-a report on the introduction of some agile practices at Motorola Global Software Group (GSG) Italy.
GSG Italy is currently at CMM level 3, and its process is an instance of a process framework capable of supporting CMM level 5. In this context we have introduced some agile practices (notably Test Driven Development) and performed a case study.
We found that it is feasible to merge a traditional, structured process (like the CMM oriented one used at Motorola GSG) and agile practices.
We found also that the productivity of the modified process is slightly lower but quality (in the sense of defect density) is higher.