Evolution of Build Artifacts in Continuous Delivery

Nowadays software gets more and more complex and, as such, the process that brings the software from developers to the users gets more complex as well. Artifacts play an important role in this process. They are files that are created during the process and are of importance for it to work correctly. In the end, the whole process is about creating, transforming and promoting artifacts. As performing these steps is a tedious task and therefore likely to cause problems when done by humans, it should be automated. This is what continuous delivery is about. In order to maintain high quality software, reduce problems during this process and get the software to the user as fast as possible continuous delivery aims to automate as much of it as possible. The automation is captured in a so-called deployment pipeline that performs the actions on different artifacts.

Current literature does not provide a structured approach or model that helps to build and maintain such an automated process in form of a deployment pipeline and, in addition, does not discuss artifacts explicitly. In this thesis we will develop a model that helps to build and maintain a pipeline in a structured way.

Therefore, we will look at the deployment pipeline and the corresponding process from a software engineering perspective and determine requirements, building blocks for design and evolution of the pipeline as well as quality aspects of the pipeline with an explicit look at the artifacts involved. Furthermore, a reference pipeline for an example application will be built and discussed.