Lou's Views: Barbarians in TuxedoesJan 31, 2001, 15:30 (38 Talkback[s])
(Other stories by Lou Grinzo)
I recently wrote about a conversation I had with Zend's Jim Jagielski about his company's approach to blending the needs of a for-profit company and open source in a creative and particularly enlightened way. What I couldn't say in that article was that when I spoke with Jim I'd already had a strikingly similar conversation with another company. I was precluded from talking openly about that company and their product plans until the official launch. Since that's happening at LinuxWorld Expo as this is being posted, I'm free to spill the beans.
The company is Borland, a company that became one of the original big names in the desktop PC business, and earned their "barbarians at the gate" tag, long before Linux was so much as a twinkle in Mr. T's eye. (Everyone reading this who bought a copy of Turbo Pascal from Borland via their ad in Byte, as I did, please raise your hand.) It's been public knowledge for some time that Borland was porting their Delphi and C++ products to Linux, under the collective name of Kylix. What hasn't been known was how Borland planned to deal with the for-profit/open source dichotomy. After speaking with Michael Swindell of Borland about this at some length, I can summarize it by saying: the good news is that it's all good news, and the barbarians are about to prove all over again why they've long been one of the smartest and most aggressive competitors in the industry. (Honestly--do you think a small company could possibly endure 15 years of butting heads with Microsoft and not become battle hardened?)
The first available version of Kylix will be Delphi, which is basically Borland's (considerably) evolved, object-oriented version of Pascal, while the C++ version of Kylix is being actively worked on and will be released at a later date. But the real news is the versions of Kylix that will be available. As is typical of commercial development tools, Kylix will come in three flavors, traditionally called "enterprise", "professional", and "entry level". Borland is calling the first two "server" and "desktop", which are somewhat better names since they more clearly indicate the focus of the product. These versions will include Borland's CLX ("clicks") framework, which has the sub-components "VisualCLX", "DataCLX", and "BaseCLX" for GUI, database, and, well, basic aspects of programming. The server version will also include facilities to help professional and corporate Apache software developers. The DataCLX components will also include Oracle, DB2, and MySQL connectivity support.
What about that third variant? That's where things get the most interesting. The "open" version will be identical to the professional version, including the full GUI RAD environment, except it will be free (beer). You'll be able to download from Borland's web site or buy it in a shrink-wrapped box, likely with a printed manual, for approximately $99. (Upon hearing this I advised Borland to begin upgrading their server capacity immediately.) Like the other two versions the open flavor will include the complete source for the CLX framework, but covered by the GPL only, so you'll only be able to create free (speech) software with it. The server and professional versions will have dual-licensed copies of CLX, allowing you to create open or closed software.
When I spoke with Michael Swindell about this, he was very frank about their rationale for creating the open vesion. In particular, he said that Borland "doesn't expect to put a Porsche Boxster in every employee's driveway" by selling Kylix to the hard-core Linux developers. I thought this was a very diplomatic way of stating something everyone in the Linux area knows but seldom says publicly: You can't make a living selling software to a group of people who think it's their right to get your products for free. Borland chose to take a more enlightened approach; instead of giving up on the whole idea of porting their tools to Linux, they saw an opportunity to leverage the hard core developers. Their goal is to build the biggest, most active user community for Kylix as possible, and use that to help sell the commercial copies to professional developers who want to use Kylix to create closed software.
I'm sure this approach still won't please some factions, and a few of the more rabid people hereabouts will howl like wounded animals that some Big Evil Company is exploiting their pure and pristine Linux to make profits, yadda, yadda, yadda. I think such reactions don't matter one iota. Every major technical change in the PC programming world, whether it was the move from DOS to Windows, the move from Windows to Linux, or this move from the more traditional Unix-style tools to a true RAD development environment has always split the developer base. There are those who resist the change only to be left behind, effectively selecting themselves for obsolescence, while others embrace the change and look for ways to exploit the new tools to serve their own ends, whether that's writing closed-source programs they intend to sell or creating free software they can give away simply because they want to help others. That's the beauty of giving users, including programmers, truly free choice--each person can take the most comfortable path without being forced into anything. Which brings up another one of those nasty details people like to avoid, the fact that until now there's been precious little real choice for Linux programmers. The IDE's available for Linux have, as a group, been less than spectacular. Kylix changes that by bringing not just a solid IDE, but a RAD environment to Linux that delivers the usability of Visual Basic, build speed you have to see to believe, and a real programming language intended for large-scale, professional work. Frankly, the traditional Linux developers won't know what hit them.
Lest I forget, there are two other items that deserve mentioning: cross-platform capability and third-party add-ons. One of the truly impressive features of Kylix is the level of portability it provides for moving applications between Windows and Linux. This is a very big deal to Linux, thanks to the large and dedicated group of Delphi programmers, many of whom are anxious to port their programs from Windows to Linux. Applications that don't directly call the Windows API (something that there's very little, if any, real need for in Delphi) or use Windows-specific features can be ported to Linux in literally minutes. A few "uses" clauses (the equivalent of "#include" in C) are changed, and the project is recompiled. Instant Linux app.
Portability is yet another area where Borland showed an uncommon amount of common sense. When they spoke with corporate developers and decision makers, they heard loud and clear that those potential customers were concerned about Linux's long-term prospects, and the dangers of having large-scale projects marooned on a dead platform. Sure, you and I know Linux has a bright future and is here to stay, but these people aren't nearly as confident, if only because many of them are still healing career wounds after betting on OS/2. Borland realized that not only was it vital to provide the lowest possible barrier to entry for people moving into Linux development, they also saw the importance of guaranteeing their corporate customers a painless exit route, too. This might not sound important to the average open source developer who would give up computing before writing so much as a "hello, world" program for Windows, but for the corporate set it's critical. And it's good news for everyone that Borland was smart enough to shape their product accordingly, just as they did to accommodate the open sourcers by including the "open" version of Kylix.
A critical part of that community building/infrastructure thing for commercial development tools has always been getting third-party support for add-on libraries and packages. Borland's tools have long had strong support in this area, and several of these companies will be in Borland's LWE booth, demonstrating copies of their tools ported to Linux. Like the portability issue, this is more immediately interesting to the corporate coders than it is to open sourcers, but it's also indirectly important to us in the same way: It makes it easier for Borland to succeed with Kylix financially, which in turn makes it easier for them to "embrace the open source community", to use everyone's favorite cliché of late. If I were a marketing guy, I'd call it a "win-win-win proposition", but I'm not so I won't.
The bottom line is that the barbarians are coming to town, and they're bringing the whole load. Not only are they armed with vast technical and marketing experience, but, like Zend, they're also showing impressive levels of flexibility and creativity. I certainly hope this is the beginning of a trend among both the companies and the individuals involved with Linux.