Few flat-earthers on a new horizon

From being a curiosity within the field of simulation, object oriented technology has rapidly thrust itself to the centre stage of computing: at a time when attendances at academic computing conferences are declining drastically any minor workshop on the topic can guarantee a healthy number of attendees; computing academics whose careers have been built on swinging Tarzan-like from trendy subdiscipline to trendy sub-discipline all seem to occupy the object oriented liana; and object oriented technology books have elbowed many respectable computing books from academic book shop shelves.

Object oriented technology is the term used to describe a particular form of software development based on programming languages such as Smalltalk, C++ and Eiffel where the data properties of the language contain a wealth of features not found in other programming languages. These features encourage massive reuse. A programmer who employs an object oriented programming language often accesses libraries of data structures which encourage a form of software development that involves collecting together prewritten components and adding a small amount of new code. Potentially this small amount of new code can be a few per cent of the final size of a system. Since software costs now dominate the cost of modern computer systems, this fact alone makes object-oriented technology a powerful solution to what academic computer scientists - usually in grant applications - call the software crisis.

I was pleasantly surprised by the journal for two reasons. First, burgeoning areas in computing tend to attract more than their fair share of three card tricksters, flat-earthers, quick-change artists, soi-disant gurus, messiahs and scissors grinders. Happily the first two issues of this journal are relatively free of these.

Second, I was impressed by where the editors seem to be positioning the journal. Computer journals are roughly bounded by the admirable Software Practice and Experience and Formal Aspects of Computing - a journal looking more like Laurie Taylor's Journal of Inconsequential Results with every issue. Object Oriented Systems lies close to the ground occupied by Software Practice and Experience. The papers represent work in the middle ground which British computer scientists often seem bashful about entering: the application of theory to large problems whose solution is vitally important for the computer industry.

In the first two issues are articles on how to adapt object oriented technology to parallel hardware architectures, the restructuring of object-oriented libraries, the use of animation to design object oriented systems and the use of error recovery in fault tolerant computing.

The final paper exemplifies the encouraging bias towards solving engineering problems. The authors choose the programming language C++ as the subject of their research. Now, C++ is a programming language which is the work of the devil: it is untidy, unsafe, and makes too many compromises towards its ancestral language C.

However, it will be the main object oriented language in industrial use for at least for the next decade. Most academics who are involved in the area of programming language research equate C++ to something you tread in rather than carry out research on; it is a welcoming sign that a journal is prepared to consider a paper such as this one.

One of the interesting facets of the journal is that in the second issue there are two reviews of the same book. I suspect that this was not planned and that two copies of the same book were sent to two separate reviewers and that the editors, quite rightly, felt that they had to publish both. I would encourage them to make this a policy. Too many reviews of books on object oriented topics are blighted by the reviewer's personal preferences, distaste for certain programming languages and his or her doctrinal concerns.

The only blight on the first two issues is a paper describing a calculus for performance prediction. It is typical of much that has been published by European computer scientists over the past 15 years: acres of mathematics, no purposive dimension and little validation. This is the sort of paper that has given computing a bad name.

A warm welcome then to Object-Oriented Systems. Provided that the editors try and encourage contributions which provide solutions to important real problems, it will be a success. However, they will have to be careful that they do not turn it into a graveyard for mathematical posturings.

Darrel Ince is professor of computing science, Open University.