Printer Friendly

Why are objects important?

At Comdex last week, Borland chairman Philippe Kahn gave a keynote speech in which he predicted that object-based technologies will soon create a post-GUI "third wave" of applications development. Anything new here?

Philippe zeroed in on a fundamental but often-ignored question about object-based technology: Why does it matter? Or, to put the question another way, why should customers give a hoot about a developer's choice of programming tools?

His answer--which we find compelling--is that object-based technologies offer virtually the only way to cope with the escalating complexity of next-generation applications. "By the year 2000, we will sell you a database manager or spreadsheet which will have ten million lines of code," Kahn says. "And to build a state of the art application of ten million lines of code, it will take 35 years--if we keep on using the same development methodologies."

Ten million lines of code? What makes Kahn think new applications are going to get so big?

To help answer that question, Kahn brought Intel chairman Andy Grove onto the stage to talk about processor trends. Today, says Grove, a standard 486 processor incorporates 1.2 million transistors. By 1995, Intel's new Pentium 586 processor, with over three million transistors, will probably be the standard PC configuration. And by the year 2000, says Grove, users will be able to buy a $5,000 desktop machine with a 100-million-transistor processor running at 100 megahertz, with 12 megabytes of RAM and 200 megabytes of hard drive. To remain competitive, says Grove, developers will have to write applications that take full advantage of this kind of computing power.

Why not just keep evolving existing applications to keep pace with the hardware environment?

The basic problem, says Kahn, is that most software today starts with a "monolithic" architecture that is difficult to extend. "What happens in software is the following: The release 1.0 of a product has a fairly good balance of features. By the time release 4.0 comes out, you have a real high rise of things that have been added and you end up with so many features that they are difficult to get to. They've been added to a system that wasn't conceived to do that. So what's the point in building all this software, and giving you all this power, if most people won't access this power?"

Rather than trying to salvage monolithic code bases, says Kahn, developers need to "fundamentally change" the way they design products. "Essentially, we need to take a factory approach to software development," he argues--that is, to create products out of reusable, scalable components that are assembled quickly in new and interesting ways. Some of these components, he adds, "could be full scale products like word processing packages or a simple text engine component," some of which will be created by independent software labs "more like the semiconductor business."

Nice theory, but do customers really care how a product--whether it's a car, a PC, or software--is assembled? Do they care how many programmers worked on the product, or how many lines of code they produce per day?

Not directly. Kahn's argument is that object technologies are likely to yield products that are more competitive. As hardware becomes more powerful and users begin requesting new features, object-based products can be upgraded faster and more successfully than their monolithic counterparts. In short, the real payoff from using object-based tools is time to market.

Any drawbacks?

Kahn didn't talk about the cost of the "fundamental change" he proposes, but Borland's own experience suggests that rewriting major applications from scratch is an enormously difficult, risky proposition. Kahn admits privately that he underestimated the time and effort necessary to build object-based tools and create Windows versions of Quattro Pro, dbase, and Paradox. If Borland can now get leading-edge features into these applications faster than Lotus or Microsoft, the investment should pay off. But for the moment, it's ironic that object technology has only delayed Borland's ability to bring new features to market.

Philippe Kahn, chairman, Borland International, 1800 Green Hills Rd., Scotts Valley, Calif. 95067; 408/438-8400.
COPYRIGHT 1992 Soft-letter
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Borland International Chmn Philippe Kahn and Intel Chmn Andrew Grove discuss future trends in software programming at Comdex
Date:Nov 30, 1992
Previous Article:Great Plains president Doug Burgum.
Next Article:Market trends: a mixed message.

Related Articles
The second biggest software company.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters