Second Life, Second Chance for CodeGear

When Borland Software Corp. spun off its Developer Tools Group in November, the move signaled a positive turn of events for its development community -- or what's left of it -- after years of changing strategies at the fabled company with roots in Scotts Valley, Calif.

Borland's history of false starts, ill-fated acquisitions, mismanagement, and constant change in the executive ranks as it chased a broader market in enterprise and database applications has distanced many developers -- people who lived in Turbo Pascal in the '80s and learned rapid application development (RAD) in the Delphi Windows IDE in the late '90s.

The winds of change at Borland haven't died down: 2007 ushered in a new marketing campaign for the company's current focus, application lifecycle management (ALM) products, and a move out of the Golden State after 24 years, to new headquarters in Austin, Texas.

As Borland continues to find its footing, the split between the parent ALM company and its wholly-owned tools subsidiary CodeGear, which remains in Scotts Valley, may bode well for the developer community.

Our focus right now is creating an optimized organization for a software development tool company, says Jim Douglas, CodeGear's chief executive, who replaced interim head Ben Smith in April. It has been part of Borland for a long time, one of the things that we've got to do is establish all of the operational excellence that is associated with this type of business, and that's really what the driver was to separate the entities.

Commitment to Development
Borland had put its tools division up for sale in February 2006. After eight months, during which no buyers emerged, the company announced plans for a wholly owned subsidiary.

Now the publicly traded firm is broken into an Enterprise segment, which includes the ALM and deployment tools, and CodeGear. For the first quarter of 2007, the Enterprise group posted a $10.6 million loss on revenues of $56.9 million, whereas CodeGear took in $14.1 million in revenues and listed $1.9 million in profit, according to a company filing.

Jim Douglas Operationally, Borland and CodeGear are different: ALM is a high-touch, lower-volume business, and tooling is a low-touch, higher-volume business, says CodeGear CEO Jim Douglas.

Operationally, the two businesses are very different: ALM is a high-touch, lower-volume business, and tooling is a low-touch, higher-volume business, says Douglas. The game plan is to eventually create a separate tools company.

That separation may be what's needed. After several years of lackluster releases, CodeGear is re-establishing the company's commitment to its developer tools. In December, CodeGear shipped JBuilder 2007 for Eclipse. Delphi 2007 for Win32 and Delphi for PHP, pitched as the first RAD IDE for PHP, followed in February. Last month, CodeGear shipped C++Builder 2007 and announced a commercial Ruby on Rails IDE slated for release later this year. As CodeGear moves ahead, the plan is to refocus on core strengths such as rapid application development for native Windows applications.

So far, developers have warmed up to the new CodeGear. The tools group has been much more responsive and much more in touch with their customer base, says Charles Vinal, president and CEO at Euclid Technology LLC in Bethesda, Md. I can't say exactly what was happening inside of Borland but the Delphi community felt that the tools division was being milked to support acquisitions in the ALM division for years and that really hurt the tools.

Return to Core Customers
Two years ago, Vinal and his development team faced a tough decision: Port the Win32 client for the company's flagship association -- management software ClearVantage to Visual Studio, or continue to develop the app, which was written in Object Pascal in Delphi.

During the time of our evaluation we found that Borland was making some significant progress in re-establishing the quality of its tools, he says. After testing and prototyping the ClearVantage client, and showing it to customers, the company determined that if they recreated the app in Visual Studio, it would involve reproducing 450,000 lines of code; code that would be untested, and likely not include any new functionality. We would be getting no new features and most likely fewer features, Vinal recalls. We actually created prototypes and ran it by our customers and they said, 'We don't care what this is written in, we want the functionality that we've always had.'

The company decided to stick with Delphi, in part because Borland showed signs that it was refocusing on its tools. Recently, Euclid Technology ported the ClearVantage client to Delphi 2007 for Win32. There were multiple reasons why we wanted to port it to the latest platform, explains Vinal. A primary one was Vista compatibility. The company had also built a lot of code in Delphi 5 and Delphi 7, and porting it to Delphi 2007 made it a lot easier to build and consume Web services through the IDE's wizards and other features.

We were able to take our Delphi application and make it more of a multi-tier, service-oriented architecture application, says Vinal. The development team used the RemObjects software development kit (SDK) for Delphi from RemObjects Software to create a smart client for applications and services in a cross-platform infrastructure. What we were able to do is to take a lot of our business logic and wrap that up into Web services. And what's nice about that is, because Delphi is already object-oriented, all we had to do is create wrappers around objects so that they can be called from various applications.

While Delphi 2007 allowed the company to be the first in its industry to market with a Vista-compatible application, Euclid Technology still uses Visual Studio for ASP.NET development. We are not anti-Microsoft, says Vinal. It is just the tools that CodeGear has provided us have allowed us to excel in our industry.

Native Strengths
After conducting surveys and evaluating the needs of its customers, CodeGear refocused its efforts on native tooling and released Delphi 2007 for Win32 and C++Builder before C#Builder and Delphi for .NET, which are expected in future releases.

Delphi 2006, dubbed Borland Developer Studio, included C#Builder, .NET, C++Builder and Win32 in a single IDE.

Many developers argue that Delphi's tooling for native code, despite no support for Unicode and 64-bit compilers, still surpasses what's offered in Visual Studio. We do use Visual Studio for .NET, but it's really not up to what Delphi can do for native code, says Craig Stintz, a software developer with Vertex Systems Corp., a company that builds case-management software for non-profit organizations. The products link to CodeGear's back-end database server InterBase. Vertex Systems uses Visual Studio for .NET 2.0, which Delphi does not yet support, and for the compact framework.

Visual Studio to this day is not a very good tool for developing native Win32 applications, asserts Stintz. The major difference in his view is the component library: MFC just comes up short compared to the Delphi VCL.

Microsoft is working on its C++ tooling in Visual Studio, and a major revision is expected sometime after Visual Studio 2008 ships. They generally are putting effort into their C++ tools but they have not come up with a compelling client library for it, notes Stintz.

Profiting from Open Source
While some longstanding customers find the recent developments hopeful, CodeGear's pursuit of new revenue streams with dev platforms aimed at Ruby and PHP is getting a mixed reaction. There are more audiences that CodeGear is being asked to serve, observes Stintz.

Personally, I can't see a reason why to use their tools for Microsoft development, but their Ruby stuff is amazing, says Corillian Corp.'s Chief Architect Scott Hanselman, a former Delphi user, who is now a Microsoft MVP, via e-mail. They might just be the Rails IDE if they play it right.

Battles Beyond Borland
  • Revenue and growth as a standalone tools vendor
  • A disgruntled installed base
  • Attracting new users
  • On the coattails of Microsoft, which controls Windows and .NET
  • Adding value to open source IDEs and tooling
  • R&D for multi-platforms
  • No Linux tooling

In light of Borland's recent struggles, and Microsoft's obvious advantages in the Windows tooling market, some bloggers have suggested that CodeGear's best strategy is to open source their development tools. Borland actually tried this tack when it released InterBase 6.0 as an open source distribution. New management went back to a commercial model with the next release, but some developers still use the open source version, notes Stintz.

It seems to me that the people who make money off of open source are the people who take an already open sourced product and support it, he observes. CodeGear is doing that with Eclipse. It seems far more common when companies open source a product that was formerly close sourced that it's a way of abandoning it. He points to the older Turbo Power components for Delphi and SAP DB.

The cost of development tools is such a small portion of what it costs to build a product, says Euclid's Vinal. What I mean is -- spending a thousand dollars on development tools when you're paying a developer $100,000 a year or whatever it is.

You want the developer to be as productive as possible, so a zero-cost development tool does nothing from my standpoint. Delphi made our developers -- as well as a lot of other developers -- more productive than their counterparts, he adds.

As a commercial -- tools vendor in a commoditized market, CodeGear faces numerous challenges. Tools really are now a component of a larger platform, says Greg DeMichillie, an analyst with Directions on Microsoft. With Microsoft developing the platforms and relevant tooling, it doesn't make much sense to use a standalone tool, he says.

Part of CodeGear's challenge is to show people that they're not Borland, says Stintz. From a business point of view I'm certainly happy that they're now more free than ever to concentrate on a developer market. It's up to CodeGear to show that they're independent from Borland and that they're going to make decisions based on what's best for the developer market and not to serve their principal shareholder.