Carleton University
Technical Report TR-254
October 1994
Object-Oriented Methodologies: A Matter of Perception
Abstract
Object-oriented analysis and design is a rapidly expanding field. There are about 20 methods claiming to support full development lifecycle. Several methods cannot be judged easily, because they are not well documented, are quite immature, are a single short publication, or are not intended for the open market. The existing methodologies change dramatically from year to year. Now, we reached a crucial point in methods evolution. New methods claim to represent a second-generation. Which methodology is better? Is choosing a method independent from the problem or not? How to choose an appropriate methodology? Are the secondgeneration of methods better than the first-generation? Are we on the right track of evolution? Do we need more methods? What should we look for next? Does the perfect methodology exist or there is always a missing piece of the puzzle? If there is something missing in a method, where is it? The answers of these questions are debatable, and it is always a matter of perception!
In this paper, we try to answer these questions for a subset of the methodologies. We compare among three different first-generation methodologies, which are OMT by Rumbaugh, OOD by Booch and RDD by WirfsBrock, and the first second-generation method which is Fusion, by solving the same problem using each methodology.We also include a study for the four methodologies in details. Then we try to combine the best of the methodologies in a coherent way.