An agile model-driven method for involving end-users in DSL development

Abstract

Domain-specific languages (DSLs) are considered to be a powerful tool for enhancing the efficiency of software developers and bring software development closer to end-users from complex domains. However, the successful development of a DSL for a complex domain is a challenge from the technical point of view and because end-user acceptance is key. Despite this fact, the relevant role of end-users during DSL development has traditionally been neglected. Normally, end-users participate at the beginning to communicate their preferences but they do not participate again until the DSL is completely implemented. As a consequence, if the language to develop reaches a complex domain, the chances that errors appear in the DSL are higher and solving them could involve large modifications that could have been avoided. As a solution, in this PhD thesis, we propose an agile, model-driven method to involve end-users in DSL development. This thesis researches if the combination of best practices from the model-driven development (MDD) discipline and best practices from agile methods is a suitable approach to involve end-users in the DSL development process. In order to validate the proposal, we have selected a highly complex domain such as the genetic analysis domain and we have collaborated with geneticists from three organizations. The proposed method has been used to involve these geneticists in the development of a DSL for the creation of genetic analysis pipelines. Simultaneously, we have carried out an empirical experiment to validate whether end-users and developers were satisfied with the proposal.