Domain-driven software development - a world of transformations

Software development teams are faced with bridging the gap between the problem, as envisaged by the stakeholders and constrained by the environment, and a software solution, which is built upon the abstractions offered by current software technologies. Unfortunately, too often the abstractions offered are limited and disparate with respect to the problem space. Reducing this gap would facilitate more sophisticated problems to be tackled in software development projects, and it would comparatively reduce development costs and time-to-market, and remove errors caused by the disparity. In this talk, I will explore a number of techniques for improving current software development practice, which relate to the theme of domain-driven software development. Domain-driven software development is concerned with making use of languages that better capture the problem by using abstractions that are more familiar to experts in the domain. These domain-specific languages are made executable either directly (compilation or interpretation) or through tool-supported refinement/elaboration to computational models that can be executed, e.g., to a mainstream programming language where one can make use of existing frameworks, components, services, etc. In the later case, real value is added to software development only if we can automate as much as possible the transformation step(s). Automating these steps requires languages that can express such transformations in a concise and maintainable manner. The principles of abstraction, separation of concerns, and problem decomposition are essential in providing intuitive and manageable domain-specific languages. The practice of software modeling has become a significant way of applying these principles to software development. Over the last few years, the software development industry has gone through the process of standardizing visual modeling notations. The Unified Modeling Language (UML) is the product of this effort, and it unifies scores of notations that were proposed in the '80s and '90s. The language has gained significant industry support and became an object management group (OMG) standard in 1997. Nowadays, the majority of software modeling techniques and approaches use UML.