Open Source

Everywhere you look there are so-called corporate, product, and technology wars. There's Linux versus Windows, AMD versus Intel, cable versus DSL, Microsoft versus the universe, and so on. One of the most interesting wars is the one between the two popular, open source, graphical desktop environments, KDE and GNOME. If I had to bet on the outcome of that particular battle, I'd put my money on KDE. Now I happen to prefer KDE to GNOME even more than ever now that KDE 2.1 is out the door. But that has little to do with my forecast. What I'm really betting on is the Trolltech GUI programming toolkit, Qt, upon which KDE is based. (GNOME is based upon GTK, aka the Gimp Toolkit).

I could provide a number of technical arguments in favor of Qt over GTK, though I've also heard some good technical arguments that favor GNOME over KDE. Regardless of which side you take, neither argument really matters very much as far as my prognosis is concerned. History demonstrates clearly that technology is usually only a minor ingredient in the recipe for success.

Qt has three things going for it that are more important than pure technology: a foot in the door of major universities, a promising start in embedded systems, and good portability (you can compile Qt applications to run on Windows and there are plans for portability to other systems).

Addressing the first point, Trolltech is about to announce a program called Qt University. Right now universities have to pay license fees to teach Qt on Windows. The only way a computer science department could give students free hands-on experience with Qt was to use Linux, because there was no free edition of Qt for Windows. Now, through the Qt University education program, Trolltech is planning to give away free site licenses for the Windows version of Qt to computer science departments.

Obviously, this is a win-win arrangement for Trolltech and the universities that choose to participate. Universities will get freebie toolkits for the Windows platform and Trolltech will get thousands more developers that will doubtlessly know and love Qt.

Geeks love Qt, but if Qt were overly geekish, the plan might not work. Fortunately, one reason Qt is likely to take off in computer science departments is that there are friendly professional-quality development tools for Qt and KDE. Right now Kdevelop (see Resources for a link) is one of my favorite development tools for Qt and KDE, but KDE Studio (see Resources for a link) is quickly wooing me away. There is a version of KDE Studio that is free for the download, but you can also purchase the more comprehensive KDE Studio Gold for between $20 and $70, depending on various options.

One of the nice things about the commercial version of KDE Studio is that it integrates Trolltech's free Qt Designer program. Qt Designer lets you paint Qt widgets into forms, much like any other tool. It generates an XML file, which you then convert into C++ source and header files. According to Trolltech, you can also integrate Qt Designer into Microsoft's Visual Studio for Windows. Not surprisingly, you can also integrate Qt Designer into Emacs (in case you're among the uninitiated, the Emacs editor is famous for its ability to be anything from an email client to a flight simulator).

In short, you don't have to be a hardcore geek to start learning and using Qt. You can be a Windows pansy who is petrified of the command line and be just as productive.

The college-kid factor

Now consider the fact that Linux gained much of its popularity because that's what all those energetic college students used and learned at school, then promoted both overtly and covertly at work. Programmers tend to stick to the tools they learn best and even become religious about defending them. If the universities start cranking out graduates with good Qt expertise as well as Linux expertise, it seems inevitable that we'll see a boom in Qt-based applications on Windows, Linux, and a full range of appliances.

Speaking of appliances, Qt/Embedded is the shrinky-dink version of Qt. Instead of using the memory-hungry X Windowing System for a graphical environment, Qt/Embedded uses something called frame buffers for its graphics. That makes it easy to create small and fast applications for Linux-based devices like PDAs, mobile phones, or various other types of Internet appliances. The cool thing about Qt/Embedded is that the difference in programming for embedded systems, Windows, or Linux is largely invisible to the programmer. It seems like it would be quite easy to build a single Qt-based application that could run equally well on Linux, Windows, or a handheld device.

Naturally, that doesn't mean you can just crank out Qt or KDE applications for the PC and expect them to be appropriate for a handheld device, even if you were careful not to use any features unique to a PC. Handheld devices have much smaller visible screens and tend to require more intuitive user interfaces. But the point is that Trolltech's Qt makes it easy to specialize on a single powerful toolkit and exploit both your knowledge and chunks of code for use across PC desktops and embedded systems.

Compaq has already brought to market a version of its iPAQ palmtop computer running embedded Linux from a company called LISA Systems, and Trolltech's Qt Palmtop Environment, which includes a set of PDA applications and an MPEG music player. According to LISA, its embedded Linux has better power management features than Windows CE, which is important to Compaq since the iPAQ tends to suck down battery juice as if it were lemonade.

The iPAQ is sure to elicit oohs and ahs, but I'm more anxious to see a full-size Internet appliance based on Linux and Qt/Embedded. If you start to see devices like that roll out later this year or sometime next year, that will solidify Qt as the de facto standard toolkit. The appliance market is sure to dwarf PC sales in the future, making Qt the most desirable platform for building applications. Windows can't interfere with that kind of growth because you can compile Qt applications to run on Windows as well. It's hard to beat the adoption of good applications when they run on Windows, Linux, and Internet appliances because there's no learning curve in switching from one to another.

And that's why Qt has the edge for the future over GTK and GNOME. Sure, there's Ximian and Eazel with their inscrutable business plans. And then there's Sun's endorsement of GNOME for its OpenOffice project (which is half the size of StarOffice 5.2, half as useful, and advancing at the speed of cold molasses). But even if you disagree with my assessment of Ximian, Eazel, and Sun, I am unaware of any formal effort to get GTK into computer science departments. Even if someone were to start such a program, GTK lacks the most significant advantages of Qt when it comes to portability to Windows and embedded systems. While GTK has frame buffer support for embedded systems and Windows compatibility in the works, it's way behind Qt in both categories.