The Unified Modelling Language (UML) has become a global standard in software development. Now Version 2.0 is set for release this spring. Alan Raubenheimer, CEO of Embarcadero Africa, advises what we should expect.
Over the last five years UML has quietly become the industry standard for specifying, visualising, constructing and documenting the artefacts of software systems. However, it has tended to remain the preserve of computer scientists and theoreticians. The release of Version 2.0, which will be completed over the next 12 months, could see it enter the mainstream of corporate programming.
UML is a collection of best practices, as identified and documented by software specialists, which have helped model complex systems. Its value is well understood, but its broad takeup has been hampered by its lack of integration with development environments. Typically this has meant that to take advantage of UML, developers have had to stop writing code, and toggle between modelling and development environments. This is inconvenient and can interrupt a developer`s train of thought.
This fact has seen usage of UML limited to 5% of corporate development, according to Gartner.
Modelling as a discipline has been in and out of vogue. There are clear benefits to modelling, but developers often ignore it - especially when the heat is on. The demands and complexity of Web application development and the wide usage of Java have brought modelling to the forefront again, especially as UML complements Java.
Ultimate goal
The OMG has issued requests for proposals (RFPs) covering the critical areas that make up UML 2.0: so far 35 companies have indicated they will respond, with the final responses expected next year. There is no consensus yet as to what to expect, but it is helpful to interpret industry predictions.
The ultimate goal of integrating UML into the development environment will be to have some of the code generated directly out of the model.
This expectation has led to a degree of confusion and misunderstanding regarding UML 2.0, with some analysts inferring that it will evolve to become a development environment; that it will lead to code being generated automatically, and as a consequence it will remove the need for programmers and their associated coding.
This is not surprising. The ultimate, surely, is for well-defined and accurate models to generate their own executables, bypassing coding, testing, compiling and the errors associated with any human process. Unfortunately, this is unlikely. Apart from the sheer scale of such an initiative, key members of the OMG, such as Joseph Rumbaugh, have made it clear that UML will remain a modelling rather than a runtime environment.
What is more likely is that UML and associated tools will evolve to automate some of the boring, mundane, repetitive tasks which take up the time of all programmers without adding value. This will free up programmers to focus on the creative aspects of their job. (Creativity cannot be automated, which is why we won`t see UML assuming a broader role in programmers` lives.) The consequence of all this effort should be better architected applications, and the lower cost of ownership savings typically associated with modelling.
Share
Editorial contacts