Ovum Ltd has published its Object Technology Sourcebook, which is the result of a collaborative effort with the Object Management Group, and offers a practical, down to earth guide to the concepts, products and vendors to be found within the object-oriented technology area. The point of the research, author Judith Jeffcoate explains, was to bring together as much information as possible about object-oriented technology and its markets. The book has been designed and written for marketing managers in the computer vendor fraternity that have to prepare new products for market and want to learn about leading edge technologies.
Applies to everyone
And make no mistake about it, object-oriented technology applies to everyone in the industry. Ms Jeffcoate had written a report on object-oriented technology for Ovum in 1988 and was pleased to find the progress that had been made in the area during the 18 months that had elapsed between the first report and the research for the Sourcebook. For example large companies are starting to use the technology such as BP Canada, American Airlines, and Brooklyn Union Gas. Second-ly, she has been able to follow the progress of one or two early products that are now on the market and selling to real users. Here she gives the example of System 9, a geographic information system developed by Wild Leitz Canada Ltd and written in Objective C. The product is now developed and marketed by Prime Computervision which also markets the Analytical Tool Box, described as an object-oriented fourth generation language, for third-party software vendors. However, there have been disappointments in the way the market for object-oriented technology is developing. For example, the concept of re-usable software has not taken off – that is Stepstone’s founder Brad Cox’s idea that class libraries with compilers would be sold off the shelf and ready to use like integrated circuits. People are simply not going out to buy classes separately, they want them as part of a development environment. Of course, a lot of vendors and users alike in the industry are sceptical about the significance of object-oriented technology. For a start they think that it is too complicated, and they simply do not understand the way that the technology demands that information is organised. Ms Jeffcoate agrees that this is currently a problem but believes that by 1995 all computer science graduates should be familiar with the technology. In the meantime, there is more of a perceptual problem in the commercial sector than within science and engineering companies that are used to thinking in object terms and arranging hierarchies. In the last analysis it is a question of choosing the best technology for any given application. The perception that object-oriented technology is complicated has largely arisen because of the strong academic following it has on conference circuits such as OOPSLA. However there are now a number of other conferences and exhibitions that deal with the real world and existing systems such as the Object Management Group’s Object World event being held in San Francisco at the beginning of June. Ms Jeffcoate thinks that there are very few niche areas for stand-alone purist object-oriented products, and that Object World is your oyster if you favour a pragmatic approach. Object-oriented technology encompasses the whole industry and it is a fallacy to believe it only concerns front-end windowing systems. These are visible applications for object-oriented technology, and although the idea of co-operative processing is not visible, it is central to the industry and will involve object technology.
By Katy Ring
Some companies that have already made a name for themselves in client-server computing may get by without dirtying their hands with object-oriented technology so long as they offer interfaces that enable their objects to be encapsulated within a wider multi-vendor environment – the sort of idea being promoted by the Object Group via its Object Request Broker. Some vendors offer the argument that it is too early
to tell what future systems will look like and therefore there is no point in getting involved with object-oriented technology, as it may prove to be a dead duck. In Ms Jeffcoate’s opinion this is a silly argument because investment determines future technology, and lots of money is being spent on object-oriented technology by large companies. For example, DEC, Hewlett-Packard and IBM have put a lot of money into building object-oriented products and therefore other companies will have to take an interest. She believes that object-oriented technology will start to make a serious commercial impact by the end of 1997. There will be a permanent overlap between procedural programming and object-oriented programming, as there will between object-oriented databases and extended relational databases. By 1996, Ms Jeffcoate estimates that object-oriented databases will have between 8% and 10% of the total database market where the extended relational database will probably take around 60% because companies like Oracle and Ingres have such huge installed user bases. An object-oriented database is only really desirable if the user wants compound documents or has large design requirements for computer-aided design or software engineering. This is why the relational vendors have never made a killing in the computer-aided design market. But they are heavily involved with software engineering environments. And that is why they will have to build object-oriented extensions to the relational model. Even IBM’s DB2 will have its semantic chunk on top that some in IBM are arguing should be object-oriented. The problem with this approach is that it could lead to performance problems because of the need to access the relational database underneath. Altair, a company based in Rocquencourt, France is developing an object-oriented database access language that Ms Jeffcoate thinks may go some way to alleviating the problem of having to move from an object-oriented language to a procedural language to access the database. At the moment object-oriented software engineering tools are proliferating as a glance at the Ovum Sourcebook reveals. Object-oriented fourth generation languages for the general commercial user will come from the extended relational vendors, argues Jeffcoate, citing the example of the Ingres Windows 4GL product. She thinks that Oracle and Informix will probably take the same approach. Some specific 4GLs will probably surface in the geographical information system and design markets as part of application frameworks. In terms of higher-level languages, C++ currently dominates the market, and the only serious challenge it might have as the stand-ard object-oriented language is from object Cobol when it emerges.
Son of SAA
This is because the combination of this with a code generator such as Telon will offer an attractive way forward for the traditional data processing manager. IBM has shown a predilection for SmallTalk, using it for internal development, but Ms Jeffcoate cannot see it being used for building large applications. She thinks that users should not wait for standards to emerge but should go with what is available today try parts of Software AG’s Entire strategy for a taster (CI No 1,649) – as vendors will have to achieve compatibility with standards. Of course as with all industry standardisation processes vendors like DEC and IBM will build supersets of the standard that will be upward compatible and not downward compatible to rival offerings, but that is a general fact of life in the computer industry. Among the big name US vendors, Ms Jeffcoate thinks that Hewlett-Packard has the most focused product strategy for object-oriented technology, DEC is coming up strong from the database side and IBM is, well, making a lot of marketing agreements with small companies. But then IBM will inevitably pop up with some concept the moment the market looks like taking off – at this very moment Earl Wheeler is probably working on OOSAA, son of SAA. Indeed, AD/Cycle Business Partner Index Technology is currently hard at it b
uilding its next generation software engineering tool around Ontologic’s Ontos database…
