Generation of task-specific architecture documentation for developers

  • Authors:
  • Dominik Rost

  • Affiliations:
  • Fraunhofer Institue for Experimental Software Engineering, Kaiserslautern, Germany

  • Venue:
  • Proceedings of the 17th international doctoral symposium on Components and Architecture
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

To utilize the potential of software architecture it is essential to make it explicit. Architects produce architecture documentation when designing systems and provide it to developers who use it as the basis for their implementation tasks. However, the views of architects and developers on the system diverge: architects look at the system as a whole, developers have a more local perspective, focusing on the elements and context they are responsible for. The comprehensive architecture documentation produced by architects does not take this divergence into account. Although all needed information may be included, the information overhead and suboptimal representation render such generic architecture documentation inadequate to be used without adaptation for all implementation tasks. To overcome this deficiency, the doctoral thesis presented in this paper proposes an approach for the creation of architecture documentation that is tailored to specific implementation tasks. Such documentation shall contain all relevant, but only a minimum of overhead information and present it in a way that allows developers to perform their implementation tasks efficiently, with high-quality results. By using automated generation techniques that derive the task-specific architecture documentation from the general documentation, it is possible to largely avoid impacts on architects' normal way of working. To achieve applicability of the approach in practice, ideas for tool support, integrated into an existing modeling environment are highlighted. Finally, ideas for the validation of the expected benefits, the current status of the doctoral thesis, as well as next steps and possible future extensions are presented.