Object languages

Ten years after it first appeared on the scene, object-oriented technology is going from strength to strength. Its success is evident in almost all areas of computing. Several international congresses are devoted to it every year in Europe, the United States and the Pacific, and magazines are multiplying, as are more or less noisy commercial meetings. One often hears talk of object technologies, in the plural. This success is accompanied by considerable confusion, as specialists in different fields do not give the same technical meaning to the word "object", a magic word already heavily loaded with meaning in everyday language. The barbarism orienté objets, a direct translation of object-oriented, doesn't help to clarify matters. We've replaced it with terminology that's more in line with French grammar: programmation par objets, langages à objets.

To avoid misunderstandings, we need to make a distinction between the two fields. Like all programming techniques, object-oriented programming itself can be practiced in any language, but preferably in object-oriented languages. It can be used just about anywhere, but is particularly useful when it comes to writing complex systems. It then meets the concerns of software engineering and gives rise to object-based analysis and design which, through the multiple methods proposed on the market (such as OOA, OODLA, HOOD, OMT, etc.), poses profound problems of knowledge acquisition and joins a rapidly developing branch of artificial intelligence. As the object point of view is the antithesis of the relational model, object databases have appeared on the market . Object-based operating systems will come on stream any day now, and distributed applications based on competing objects will multiply. The objects used in these latter fields have to cope with difficult living conditions (persistence, distribution), and are no longer exactly those of classic object languages (cf. the CORBA standard promoted by the OMG organization). In the field of artificial intelligence, knowledge representation also makes use of objects, which are in fact derived from frames and differ from the ones we're dealing with here both in their intent (e.g. classification) and in their structure (absence of the procedural component).

In this article, we restrict ourselves to object-based (programming) languages from the point of view of their use in software production (and not, for example, from the point of view of artificial...