Most methodologies are short on “method” and long on “ology.” When writing this article about the UML, I remembered this quote from the good old Yourdon days of structured methods. With UML, we now have an approach that is very short on method and long only on notation. This article discusses the advantages, disadvantages, experience, joy, and anger of and with the (still incomplete) method wars between data modelers and function modelers. For a long time, people believed in either data flow diagrams or entity-relationship diagrams, but seldom both. Finally, toward the end of the ’80s, people began talking to each other. Even the method gurus started to agree that using data models and function models and behavior models (or so-called real-time extensions) was a good idea for most projects. I call this approach “integrated structured methods,” and it was (and is) good.
But just when we seemed to reach a consensus on the analysis and design approaches, some of the old and many new gurus started to preach the gospel of OO methods. The first reaction of the market and the CASE vendors was puzzlement, followed by chaos. It was not clear whom to follow. Dozens of notations, dozens of different explanations of what an object is and isn’t, no stability, no trends visible.
It was only when Grady Booch and Jim Rumbaugh joined forces in late 1994 and announced a “unified method” that the average user saw light at the end of the OO tunnel. This unification effort is why I praise UML. The time is right to define strict semantics for the basic OO concepts and agree on definitions and notations. OO is no longer in its pioneer days. We have tried out basic principles long enough to have a deep understanding. A couple of years ago it was fine to experiment with different notations and different concepts, but now we need to settle on a few of them.
But just when we seemed to reach a consensus on the analysis and design approaches, some of the old and many new gurus started to preach the gospel of OO methods. The first reaction of the market and the CASE vendors was puzzlement, followed by chaos. It was not clear whom to follow. Dozens of notations, dozens of different explanations of what an object is and isn’t, no stability, no trends visible.
It was only when Grady Booch and Jim Rumbaugh joined forces in late 1994 and announced a “unified method” that the average user saw light at the end of the OO tunnel. This unification effort is why I praise UML. The time is right to define strict semantics for the basic OO concepts and agree on definitions and notations. OO is no longer in its pioneer days. We have tried out basic principles long enough to have a deep understanding. A couple of years ago it was fine to experiment with different notations and different concepts, but now we need to settle on a few of them.
Read More/Download