Assessment of OO

In May 1990 the Object Interest Group (OIG) was formed to understand and assess object oriented technology, to get hard evidence to support the assessment and to enable member companies to position the technology in their IT strategy. In 1991 they reached the shared conclusion that object orientation was potentially one of the most powerful technologies ever to become available to the IT industry and its users. As such it would demand high calibre management. It was not a panacea but a high-power tool, dangerous if misused but capable of great things.

Here is a summary of the rest of their findings from their initial assessment of OO ...

The aim of the technology is to enable software production to mature to a state analogous to the production line in engineering manufacturing. This has evolved through the concept of standard components from which the main assembly is built, without the need to be concerned with their internal working. This allows for the continuous improvement in component design and the production of different but conformant (e.g. variations in car model - but still a car!) products from the line. A parallel evolution has never occurred in software because IT models the business with programs separated from data. There is no unit of software equivalent to the engineering component. The nearest, the program, is not effective. If a business process changes, both the program and data modelling it usually change. The data change then affects other programs not related to this business process. The effect of change is widespread throughout the system.

Object orientation defines a new unit of software 'the object'. Each object models a real entity in the business by encapsulating the data that describes the entity together with the logical processes that define the services provided by the object. The object is now equivalent to the engineering component and its services are equivalent to component behaviour. This enables a more natural model of the business. Objects provide 'firewalls' preventing the effects of business changes spreading uncontrollably. Over time an ever growing model of the whole business evolves in the form of an inventory of objects that can be reused in many different applications. Eventually most of the new applications' software needs will be assembled from the inventory, making the IT operation far more responsive.

Competitive edge in the 90s will depend on the ability of the IT operation to reach out to wherever the business wants to go and to integrate rapidly existing and new functions to meet the local needs of the workplace. The need is also to make employees more effective by widening the range of activities they can undertake. This requires a shift to interfaces at which users can easily manipulate familiar business objects that correspond to their own perception of their business world. Current interfaces despite the sophistication of graphics, multiple displays etc are 'applications oriented' and require the user to know and think 'applications' which are not business concepts. Today's concept of a first major delivery followed by years of maintenance must disappear so that all work is seen as part of a continuous, evolving, asset building, production process in which the major part of any new application can be met by the assembly of reusable components.

An important independent pressure is coming from the hardware cost revolution (one megabyte of RAM now costs less than two lines of source code). Every few years the performance/price ratio of hardware doubles encouraging demands for software far beyond our ability to deliver, particularly if real computing horsepower is to be obtained from all this equipment. Object orientation in our view is the only technology today that has the potential to address these issues.

It is clear that these benefits are unlikely to be achieved without experienced assistance and investment in training. New management problems appear. Changes in culture and the need to reorganise are common in most organisations. Management of software as an asset (especially at the analysis and design level) will be a novel experience for most enterprises. There are no methodology, CASE or database products sufficiently mature to justify immediate migration of any large scale IT function. From a commercial systems viewpoint the available products and techniques can currently be used for small to medium applications, for front end work (new user interfaces and integration of large existing systems to meet the needs of specific workplaces) and for enterprise modelling. Nevertheless, there have been several recent examples of its use for large scale enterprise critical applications to either gain competitive edge or because a competitor has adopted the technology.

The way forward is to explore the technology and its management on pilot applications, front end work and enterprise modelling. On a broader front a proposal for a further years work by the OIG is circulating for approval. The work is focused on helping members to get started, on developing method, technique and organisation based on sharing real project experience and on influencing suppliers to respond to our needs and priorities.