Object-Oriented Programming on the Network

Object-oriented programming techniques have been used with great success for some time. But the techniques of object-oriented programming have been largely confined to the single address space, and have not been applicable to distributed systems. Recent advances in language technology have allowed a change in the way distributed systems are constructed that does allow real object-oriented programming on the network. But these advances also change some of our most basic conceptions about the relationship between processor and code, and what it is that constitutes a computer. We will argue that a new computing architecture, based around the ideas of the network and full objectorientation, will soon become the dominant computing architecture, allowing us to tie together large numbers of devices but requiring that we think and design in entirely new ways. 1 Object-Oriented Systems Over the past fifteen years, object-oriented programming has moved from the research lab to the mainstream. The use of object-oriented techniques is rarely questioned these days; organizing a program as a group of interacting components, each of which is treated as an object, is now the default way of designing any program of any size. To do otherwise is sometimes necessary, but that necessity must be justified; the default way of organizing a program or system of any size is to use object-oriented techniques. Of course, different people have different notions of what it is for a program or sysem to be constructed along object-oriented lines. To program in an object-oriented fashion is more than using a language that labels itself as an object-oriented language. All of us have seen many C programs with little or no object-orientation written in either C++ or Java. Some of the better object-oriented programs I have had the pleasure of reading have been written in languages that are not themselves considered object-oriented. Real object-orientation is a design philosophy, not something that is automatically gained by choosing a language. Given the wide variety of what is meant by „object-oriented programming“ , perhaps I should start by heeding the advice of Humpty Dumpty in Through the Looking Glass[1] and say what I mean when I use the term. The object-oriented approach to programming rests, I believe, on a small number of principle.: The first of these is the principle of abstractionthat the actual representation of an object is an accidental property of the object, with the essence of the object being shown by the operations that can be performed on that object.