Organisations often find themselves trapped in a costly cycle of replacing legacy systems. Jidoka is a software development methodology that aims to break this cycle and for organisations to mature into a mode of continuous modernisation. Jidoka, which loosely translates to automation with a human touch, offers several strategies for developers to modernise legacy systems onto their preferred technology stack, ensuring that quality, security, and other compliance is met. The paper presents state-of-practice information on how the process emerged through the use of Model-Driven Engineering (MDE), DevOps, and lessons learned using other software development methodologies across 100+ projects.

To be adopted by architects, modelling approaches must provide a means to leverage the software patterns and architectural styles that are relevant to development practice, instead of those proscribed by black-box CASE tools. Architecture-Centric Model-Driven Software Development (AC-MDSD) is a modelling approach that provides architectural control of the generated application. However, AC-MDSD primarily focuses on generating infrastructure code. We apply AC-MDSD to web engineering and contribute a technique to define and generate system behaviour that goes beyond the create/read/update/delete infrastructure functionality. We use UML profiles augmented with OCL to specify the behaviour. We provide an example to illustrate the approach and outcomes.

Much of the research into testing model-driven systems has focused on checking the models used to drive code generation, and testing the transformations that are used. This is predicated upon the use of full code generation. However, much of the model-driven system development used in practice is based on the use of partial code generation, where the developer implements sections of the system by modifying or adding to the generated code. In these situations, it is important to test for problems in or caused by this manually modified code. In this paper, we present a pragmatic approach to testing model-driven systems based on partial code generation. Our approach uses model-based testing techniques based on reuse of development models to drive test case generation.