IBM Corp is really beginning to talk up the merits of its various object-oriented software development efforts. At last week’s Object Expo Europe ’93 event in London, Cliff Reeves, the firm’s worldwide director of object product technology, outlined IBM’s plan for the Taligent operating system – derived from the Apple Computer Inc-IBM object joint venture of the same name and its own Distributed System Object Model toolkit. He also told developers about the company’s plans to restructure its software development process. Object technology will become as revolutionary as the discovery of sub-routines were in the 1970s, Reeve claims, because objects go one step further and enable users to extend and modify code in line with their specifications. Traditionally, sub-routine technology has existed on the premise of pre-definition and it isn’t flexible or powerful enough to deal with new technologies such as the Object Management Group’s Common Object Request Broker Architecture specification and the Open Software Foundation’s Distributed Computing and Distributed Management Environments. In the past, software suppliers made money by providing shrink-wrapped packages, which in essence were complete solutions for single machines. But this isn’t a viable option for developers any more, because users are demanding more complex, scalable and distributed products… and object technology is one way of delivering extensible and pluggable components, he says. To mirror market trends, IBM has restructured its software strategy to take advantage of distributed computing at every level of systems software development. Future application development will be structured using a modular or component-based framework, which will enable applications to be smeared across networks at both the client and server level.

Taligent

Reeves claims that this will make application development and deployment more flexible and dynamic. End-users will be able to adapt software in accordance with their requirements. IBM currently has three major object projects: the Distributed System Object Model, Taligent and OpenDoc, the last being Apple’s answer to Microsoft Corp’s Object Linking and Embedding specification. The DSOM Distributed System Object Model, introduced last month, is a cross-development tool kit that enables objects to work together, irrespective of hardware or language in which they were originally written. In its simplest form, DSOM provides a standard way of defining an object interface by using the Common Object Request Broker’s interface language definition. Generic binaries are also provided, which enable the toolkit to work with any implementation of ANSI C, C++ or Cobol – support for Digitalk Inc’s SmallTalk language has been added this month. The Distributed System Object Model is available now for OS/2 and versions for Macintosh and AIX/6000 will be added by the end of the year. Last week, IBM also announced a co-development with Hewlett-Packard Co to do a version of Distributed System Object Model to Precision Architecture RISC systems and jointly to develop enhanced distributed features for the tool kit. The key to Taligent, meanwhile, is its technological structure, says Reeves. Unlike any other operating system apart from NeXTstep, it is based on an object-oriented microkernel framework, which categorises a series of co-defined objects and patterns into a set of class libraries. According to Reeves, Taligent engineers have already defined compound documents, multimedia, internationalisation and graphical user interface libraries, which will enable users to generate more than 60% of applications before manual coding begins. This means independent software vendors will be able to cut programming schedules in half and reduce the cost of development. Prices to the end user should also decrease in the long term. The other benefit Taligent brings, Reeves claims, is the ability to design and deploy more complex applications across computing environments. Developers kits for the first set of class libraries should be released by th

e end of the year.

By Alison Hawkings

Taligent has been designed to understand the complexities of networks, application interoperability and development. It will be Distributed Computing and Distributed Management Environment-, Common Object Request Broker Architecture- and Common Open Software Environment-compliant, which means that it is an operating system designed for the next generation of computers and software, he says. In sync with its Taligent effort, IBM is also revamping its overall Unix strategy. The plan is to move AIX/6000, AIX Lite, AIX/ESA and other IBM operating systems such as Workplace Shell and OS/2 over to a microkernel base in the next two years. (Microkernels essentially refine operating system functions to the irreducible hardware-dependent minimum, enabling the systems run atop them to run a lot faster). Sitting on top of the microkernel is a series of personality layers. Some cater for different operating systems and graphical user interfaces; and others, called personality-neutral services, will include features that operate the same way for a variety of different architectures such as databases, communications and print services. Eventually the system will turn into a complete Taligent implementation. IBM positions Taligent and OpenDoc against Microsoft Corp’s Object Linking & Embedding 2.0 specification, which will eventually turn into Cairo, and NeXT Computer Inc’s NeXTstep environment. OLE 2.0 is designed to embed compound document components in different applications and across different operating systems. But it is primarily concerned with desktop interaction, and it does a good job as far as it goes, says Reeves. For example a user in Microsoft Excel can request graphics from Lotus Development Corp’s 1-2-3 and import it into Excel. The graph can be manipulated, changed and updated in either application, everything a desktop user requires. But as a distributed technology it falls short. In Reeves’ words, OLE is like having a conversation between members of a family in the family’s house, but as soon as you go out of the room or introduce anybody new there are problems with communication. The Taligent approach is different. Taligent is like a telephone system, Reeve suggests. Any object or communication path can be established across a distributed object environment and applications can be deployed in the same way. To prepare users for Taligent, IBM, Apple and a host of other software suppliers including Novell Inc, WordPerfect Corp and Borland International Inc launched OpenDoc, an alternative to Microsoft’s Object Linking & Embedding specification, at PC Expo in New York few weeks ago. OpenDoc, like OLE, is intended to provide document interoperability between different applications and systems by passing objects across networks.

OpenDoc

It is based primarily on technology from Apple and IBM. Apple provides compound document and open scripting, which defines links and embed rules for events between applications. Apple’s Bento technology provides data structure definitions, which translates different application data types. The other parts of OpenDoc are based on the Distributed System Object Model, IBM says. OpenDOC is available now for OS/2, Windows, and AIX, and support for NetWare, OS/400, MVS and System 7 will follow later in the year. Microsoft is currently developing Dynamic Data Exchange extensions for OLE 2.0 to solve some of its distributed networking problems. The other player in the market is NeXT Computer Inc. Like Taligent, NeXTstep is based on an object framework, and was the first object-oriented operating system to be developed for the commercial market, in 1989. NeXTstep was a phenomenal programming environment for its time. It was fast and a very rich environment for application development, but it didn’t go far enough to solve issues like memory management or object distribution. Taligent is one rev on from that, he says. In Reeves’ view, NeXT’s problem wasn’t its technology but its business strategy. It designed NeXTstep specifically to run on proprietary

hardware, a weakness that was then compounded by high end-user prices and no commitment to develop versions for any other systems.