The convergence of modeling and programming: facilitating the representation of attributes and associations in the umple model-oriented programming language

  • Authors:
  • Andrew Forward

  • Affiliations:
  • University of Ottawa (Canada)

  • Venue:
  • The convergence of modeling and programming: facilitating the representation of attributes and associations in the umple model-oriented programming language
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

This thesis investigates approaches to model-driven development (MDD) in which developers can keep using their familiar textual programming languages, but with additional model-oriented concepts. The added concepts include associations and attributes as found in the Unified Modeling Language (UML), as well as concepts from software patterns and other common programming abstractions. By keeping text at the forefront of development, we maintain all of the advantages of text, without having to sacrifice the benefits of diagrams. By allowing a model to be equally expressed in either diagrammatic or textual form, we enable what we have termed text-diagram duality, a duality that benefits programmers and modelers alike. We explore why software developers in some situations prefer diagrams, but tend to revert to textual means to write and maintain software systems. To explore the capabilities of modeling in code, we developed a model-oriented programming language called Umple. At its core, Umple is a family of object-oriented languages enhanced with additional abstractions. Umple supports both platform-independent models (PIM), as well as platform specific models (PSM). Umple currently integrates with Java, PHP and Ruby; referred to as base languages throughout this thesis. Our research focuses on investigating the opportunities and obstacles we discovered in the course of implementing and using UML-like associations and attributes in Umple. It is our hypothesis that current features available in object-oriented languages can be enhanced with a more model-oriented approach, providing a textual form for modeling concepts that have been primarily available diagrammatically. By providing modeling abstractions in a programming language, the complexity and size of the resulting systems, we argue, is reduced and more developers, particularly those who are used to writing code, will be more eager to adopt modeling practices. At the same time, our approach maintains the benefits of diagrammatic approaches to software development, since Umple can be rendered and edited as a UML-like diagram. Our primary contributions to the field of computer science are as follows: First, we provide an empirical investigation on the nature of modeling practices. Second, we present the design, implementation and analysis of a model-oriented language, Umple. The language is presented as enhancements to existing programming languages including Java, PHP and Ruby.