G5 Artikkeliväitöskirja

Technical Debt in Software Development : Examining Premises and Overcoming Implementation for Efficient Management

Julkaisun tekijät: Holvitie Johannes

Kustantaja: University of Turku

Paikka: Turku

Julkaisuvuosi: 2017

ISBN: 978-952-12-3524-5

Verkko-osoite: http://urn.fi/URN:ISBN:978-952-12-3524-5


Software development is a unique field of engineering: all software constructs retain their modifiability — arguably, at least — until client release, no single project stakeholder has exhaustive knowledge about the project, and even this portion of the knowledge is generally acquired only at project completion. These characteristics imply that the field of software development is subject to design decisions that are known to be sub-optimal—either deliberately emphasizing interests of particular stakeholders or indeliberately harming the project due to lack of exhaustive knowledge. Technical debt is a concept that accounts for these decisions and their effects. The concept’s intention is to capture, track, and manage the decisions and their products: the affected software constructs.

Reviewing the previous, it is vital for software development projects to acknowledge technical debt both as an enabler and as a hindrance. This thesis looks into facilitating efficient technical debt management for varying software development projects. In the thesis, examination of technical debt’s role in software development produces the premises on to which a management implementation approach is introduced.

The thesis begins with a revision of motivations. Basing on prior research in the fields of technical debt management and software engineering in general, the five motivations establish the premises for technical debt in software development. These include notions of subjectivity in technical debt estimation, update frequency demands posed on technical debt information, and technical debt’s polymorphism. Three research questions are derived from the motivations. They ask for tooling support for technical debt management, capturing and modelling technical debt propagation, and characterizing software development environments and their technical debt instances. The questions imply consecutive completion as the first pursued tool would benefit from—possibly automatically assessable—propagation models, and finally the tool’s introduction to software development organizations could be assisted by tailoring it based on the software development environment and the technical debt instance characterizations.

The thesis has seven included publications. In introducing them, the thesis maps their backgrounds to the motivations and their outcomes to the research questions. Amongst the outcomes are the DebtFlag tool for technical debt management, the procedures for retrospectively capturing technical debt from software repositories, a procedure for technical debt propagation model creation from these retrospectives, and a multi-national survey characterizing software development environments and their technical debt instances.

The thesis concludes that the tooling support, the technical debt propagation modelling, and the software environment and technical debt instance characterization describe an implementation approach to further efficient technical debt management. Simultaneously, future work is implied as all previously described efforts need to be continued and extended. Challenges also remain in the introduced approach. An example of this is the combinatorial explosion of technology-development-context-combinations that technical debt propagation modelling needs to consider. All combinations have to be managed if exhaustive modelling is desired. There is, however, a great deal of motivation to pursue these efforts when one re-notes that technical debt is a permanent component of software development that, when correctly managed, is a development efficiency mechanism comparable to a financial loan investment.

Last updated on 2021-24-06 at 10:11