Borland International Inc has started styling itself the leader in object-oriented programming – is there anything more behind this claim than hype? Matthew Price, Borland UK’s language product manager certainly thinks so, since, he claims that the Scotts Valley, California company has sold more object-oriented tools than any other company and that Borland was the first company to bring object-oriented Pascal to the MS-DOS environment in 1988. The company’s move to object-oriented methods for internal development has also enabled it to start getting its products out a lot quicker than hitherto. However, Price admits that the object-oriented concept is a difficult one to get across as it is really a design methodology rather than a language technique.
Passing fad
The main factor that has alerted the market to the idea that there may be more to object-oriented technology than a passing fad is the arrival of Windows 3.0. Suddenly application developers are being asked to develop Windows applications and are finding that traditional procedural tools are just not right for the job. The big cultural breakthrough from Borland’s point of view came with the release of TurboVision for Pascal, which is an object-oriented application framework for MS-DOS that developers use in conjunction with Turbo C++. Until then, Price thinks that object-oriented programming was seen as an academic topic, which required the learning of a totally new language such as SmallTalk. Of course within the object-oriented community there is a great deal of debate as to whether C++ is often successfully used as an object-oriented language. Price acknowledges these doubts, but says that the language’s designer Bjarne Stroustrup has decided that C++ is a separate language from C and should be viewed as a tool to implement a different design methodology. So, while Borland appreciates that the developer’s learning curve is driven by her ability to learn a new methodology, it believes that the argument still stands that a C programmer will not also have to learn the basic syntax of a new language if she opts for C++. Of course, SmallTalk enthusiasts say their language is much richer as it has its own development environment. Price, however, thinks that users pay a performance penalty for that environment and that SmallTalk applications are inherently slower. He thinks that users want to take an executable file without a runtime. Price argues that the other advantage of C++ is that it has good control of machine architecture, a quality bought over from C. Consequently, when a developer implements class libraries, the base classes have machine access and read-write access. Anyway C++ has taken off more quickly as the key object-oriented development language, while SmallTalk is used more as a prototyping object system. There are benefits from both and, whisper it low, projects exist at IBM Corp to try and interface the two.
By Katy Ring
Borland sees incredible opportunities for its object-oriented programming languages for Windows 3.0 development as Microsoft Corp has signally failed, in Borland’s opinion, to provide Windows developers with object-oriented tools. Instead, Microsoft offers Quick C for Windows and the Software Development Kit – both of which follow procedural programming methods. The difference between developing a Windows 3.0 application with a procedural or an object-oriented technique has to be seen to be believed. With procedural methods, programmers have to learn more than 550 Windows Applications Programming Interface routines. With Borland’s ObjectWindows most of those programming interface calls are encapsulated as high-level functional objects. So, basically, a developer working with ObjectWindows has a jumpstart in the form of a generic Windows application, which she can customise with windows, menus, dialogues, list boxes, buttons, edit fields, icons and so on. ObjectWindows is the result of joint development work with Evanston, Illinois-based Whitewater Group and is a specification for a language independent classs library for ap
plication developers writing for windowing environments and has been submitted to the Object Management Group’s Object Services Task Force. It has been implemented in several languges such as Actor, Borland C++ and Turbo Pascal but has not yet reached the market in Borland products. Indeed, Borland and Whitewater Group are evidently bent on seizing the market with their class libraries even if they don’t win in the Object Management Group standardisation process. Whitewater has already developed Object Graphics, which is a specification for a language- and environment-independent class library that extends ObjectWindows to include graphic objects for use in applications. And Borland says that markets move faster than standards bodies and that it has the largest user base for object-oriented tools. As a tools vendor for the desktop, where does Borland see future growth – with IBM-Apple and Pink? Or with Microsoft’s New Technology? Price thinks it is a mistake to write off either of these operating systems and adds the Novell Inc-Digital Research DR DOS progeny to the list of likely successful operating systems for the future. Quite simply, he does not believe that one dominant standard operating system will emerge for the desktop in the 1990s because with the arrival of object-oriented technology application developers can afford to support several major systems. This is because application frameworks will protect developers from the complexity of different operating systems, while client-server technology will evolve to handle the communications problems involved in supporting a variety of desktop systems. It is this perspective – that object-oriented technology will be key to the survival of application developers – that is driving Borland’s product vision.
Spreadsheet objects
Borland itself is interested in providing off-the-shelf class hierarchies as application components. In the long-term, Borland believes it will, for example, be able to provide self-contained word processor or spreadsheet objects that would be able to print themselves, know how to work with different operating systems and so on. Object-oriented technology will also help Borland make sense of its acquisition of long-time rival Ashton Tate. The idea is that an object layer will be created above the file format in the Paradox engine. This will mean that users can enter data via either Paradox or Object dBase and either can then take advantage of the object layer in the Paradox engine. So anybody that is looking carefully at what Borland is up to in various strategic areas comes away convinced that object-oriented technology is crucial to the future of both Borland and its customers.
