Books for developers


Borland, Java, Eclipse, .Net futures

This piece is some thinking aloud. It is attempt to focus on why Eclipse.org is at the centre of a sea-change in the Java world; why it is such a threat to Borland; and why the company is unlikely to succeed if it pins all its hopes on Java and Borland Enterprise Server. There's a few ideas about .Net as well.

Consider it work in progress, please add your comments, and maybe I'll give it a revision or two as the thinking evolves.

Editors: Want to reprint this article? Rights may be available; contact me for details.


  There is a huge shake-up in the Java tools market happening right now. It's called Eclipse.



Why Borland should get behind .Net

It was some years ago, following years of alternating success and frustration trying to outsmart Microsoft, that Borland put together a strategy for a new future built primarily on Java. While not abandoning Windows tools such as Delphi and C++ Builder, Borland made it clear that cross-platform standards were more important. The company acquired VisiBroker and poured immense energy into building a business on its application server. Borland also saw that dependence on tools alone is a killer when it comes to long-term growth. With Borland Application Server it could enter the more lucrative world of services and deployment licences. By bundling the application server with Enterprise versions of all its tools, from Delphi to JBuilder, the company could promote its excellent technology to the influential development community. Clearly the main casualty of this strategy would be the Windows products, and a slide at a Borland Developer Conference in London last year duly showed Delphi and C++ Builder under the dreaded legacy heading, pointing the way to a new era where all the real action is in Java.

Running into the sand

I believe we are now seeing this strategy running into the sand. Here's why.

First, Borland has not captured a large slice of the application server market. According to Giga, the market shares in 2001 saw IBM WebSphere and BEA Weblogics with 34%, Sun's iPlanet (now Sun ONE) at 7%, Oracle at 6%, Sybase EASserver at 4%, HP at 3%, leaving Borland along with others like Iona and Allaire (now Macromedia) to scrabble around for the remainder. Can Borland grow its market share? It will be immensely difficult. Most analysts see the enterprise application server market consolidating. The main beneficiaries are likely to be the largest vendors for whom the application server is tightly integrated with their platform - IBM, Oracle, and maybe Sun. On this reading, BEA will struggle to maintain its share, and the minority players will more likely lose share. The platform vendors can always undercut those who primarily sell software; the application server becomes part of the bundle.

As if that weren't enough, the application server is now commoditised, with free options including JBoss, HP-AS, and a version of Sun ONE.

Second, there is a huge shake-up in the Java tools market happening right now. It's called Eclipse. This is the open-source, IBM-sponsored, extensible IDE built using Java and the SWT (Standard Widget Toolkit). SWT is a thin wrapper over the native code GUI of the host operating system, most often Windows although others such as Linux and Solaris are also supported. Because of this, Eclipse performs more briskly than Swing-based tools like JBuilder. That's actually the least of Borland's problems. Eclipse is run by a consortium, of which Borland is ironically a member. Developers love the tool, and with cross-industry support behind it, Eclipse has every prospect of becoming the standard Java IDE. It makes huge sense. Java vendors and developers contribute to a common IDE, and promote their particular application servers or other middleware by creating extensions for it. Well, it makes huge sense for most vendors, but not Sun, sidelined by SWT which is not standard Java, or Borland, trying to sell a rival product commercially, when Eclipse is a free download. Suddenly JBuilder looks dated, weighed down by Swing, and too monolithic in its architecture (despite its open tools API).

Of course the big beneficiary from an ascendant Eclipse project is IBM, whose Visual Age line is already Eclipse-based.

Where next for Borland?

I hope you will see that the above is no way critical of Borland's technology. I believe Borland Enterprise Server is an excellent product, and I have a high regard for Delphi, C++ Builder, JBuilder, and even Kylix. I think the company makes strategic mistakes from time to time, like launching JBuilder 7.0 without any replacement for the JBuilder 6.0 Professional Edition. I think also that Borland has become too much a follower of the market, rather than trying to get ahead of the game. In hindsight, the evolution of JBuilder has been conservative, where perhaps a little more vision might have seen the inevitability of Eclipse, or something like it, and done more to pre-empt it. JBuilder needs a complete makeover, rather than the frequent, incremental updates it has received.

However, I believe the best opportunity for Borland lies in embracing Microsoft's .Net platform, alongside its continuing Java effort. The industry has seen Java as the great open alternative to a Windows-centric world; but the picture looks less clear now that we see the same old big vendors using Java standards as a marketing tick while they grab even more market share. I believe that a successful .Net platform is essential to the health of the industry, providing competitive impetus to improvements in both Java and .Net.

I have nothing against Java; it is wonderful technology and continues to evolve and improve. However I do think there is room for other approaches, and not only because of the dullness of a single-language world. In particular, the immense complexity of J2EE is an opportunity for vendors who can achieve a proportion of J2EE's scalability and resilience, but with a simpler application model. Microsoft's .Net can do this (of course you can also build this on top of Java, with tools like Macromedia's ColdFusion MX). It is early days for .Net, and there is a lot of COM legacy which needs to unwind before it can achieve its full potential, but the momentum is there. A key milestone will be Microsoft's .Net server, if the company can persuade the world that it is able to produce a secure server operating system and thus open the door to greater adoption of IIS and ASP.Net.

Prospering in the .Net world

I digress. The point is that Borland is uniquely placed to prosper in the .Net world. First, it has a deep understanding of Windows and of RAD tools; undoubtedly the company has the resources to create great .Net development tools. Second, and unlike Microsoft, Borland has cross-platform expertise. Microsoft will never do cross-platform with any enthusiasm, because it is schizophrenic about promoting operating systems other than Windows (although I believe the company may yet port .Net to Mac OS X). Borland however can do cross-platform; it can do wrapper classes that can target either .Net or Win32 or Linux; it can make great contributions to the Mono project that aims to replicate the .Net Framework on Linux.

However, and it is a big hesitation, to prosper with .Net Borland needs to do more than simply build a Delphi for .Net at its own rather leisurely pace. To succeed the company needs to capture and pursue a vision of what .Net can do; RAD for the Enterprise, .Net beyond Windows; or whatever. Borland could do a great job with this; I hope it rises to the challenge.

Copyright Tim Anderson July 2002. All rights reserved.

Sign up to be notified of future articles

Latest books
On .Net, on JBuilder and Other Web development topics

Latest books
On .Net, on JBuilder and Other Web development topics