From The Editor

Mobile Wishes and Linux Dreams

A favorite pasttime of mobile device owners and developers is the topic of operating system superiority. However, unlike the desktop wars of the early 90's between OS/2, Windows, and the Macintosh, the leading handheld OSs were all designed with different goals in mind resulting in most of these superiority discussions ending in a draw. In fact, an interesting sidebar tangential to the topic of OS superiority would be to discuss whether it is necessary to even have one dominant operating environment. In a distributed, networked world, the operating system is being marginalized to a point where as long as your computer can "speak" a few specific dialects (HTTP, HTML, XML, TCP/IP, and Acrobat for instance), noone around you much cares which OS you're using. This line of thinking has been pushed to varying degrees of success by industry luminaries such as Larry Ellison of Oracle. In the end, I think we trivialize end user requirements with the notion that all client software is unnecessary and that all data can be formatted for display in a standardized manner. Admitting that client application development is not a thing of the past opens up the playing field a bit by placing the partial success or failure of a platform in developers' hands.

If developers are, in fact, an important factor in the success of an OS, I have to believe that a huge advantage is offered by any technology promising a smooth transition from the desktops and server of today to the mobile, wireless platforms of tomorrow. Looking around at the competitors lining up to do battle for this new market space, I see an interesting assortment of choices. The Palm Computing Platform is popular among the early adopters (that's what we all are termed at this point until these technologies become truly ubiquitous...) yet it asks you to learn yet another entirely new API and GUI toolkit. Java is being positioned to enter this space but it's already been pushed back from PC clients to back-end servers; if the performance and portability problems limited it on the desktop, how on earth do we expect it to dominate handhelds? Psion/Symbian has also generated strong developer emotions but...new tools, new API. Windows CE...well, when Microsoft nails down what THEY want to do with this OS, I'm sure we will revisit that topic.

Clearly, there is an opportunity here for an inexpensive, portable, stable OS with an existing large developer base to step in and generate a significant amount of interest. Linux is obviously the best candidate to lead this charge, and recently a possible candidate has stepped forward led by Transmeta and coincidentally enough, Linux's creator: Linus Torvalds. If you didn't follow the big announcement on January 19, Transmeta (a super-secretive startup based in Santa Clara, CA) recently presented a very interesting new microprocessor named Crusoe along with several related technologies.

Essentially, Crusoe is a "smart" microprocessor that uses Transmeta's Code Morphing technology to "morph" standard instructions (such as x86 instructions) to the much simpler Crusoe instruction set for execution on their VLIW (Very Large Instruction Word) engine. If this works as promised, up to 80% of the transistors (and therefore power) required of a standard microprocessor can be eliminated and replaced with software. The promise of an extremely low-power/high-performance architecture is the Holy Grail of mobile computing. In conjunction with Crusoe, Transmeta has also announced Mobile Linux, which is basically the standard Linux 2.3 kernel with special power management drivers; Transmeta refers to this power management as their LongRun technology. Instead of the off/on cycling common among most mobile computing chips (such as the Motorola Dragonball processor used by Palm organizers), LongRun allows power to be regulated by the dynamic control of both processor voltage and clock frequency resulting in even larger power savings to the device. Clearly, this is a technology that we will be following closely!

The use of Mobile Linux with the Crusoe architecture could be the "killer application" that places Linux at the forefront of mobile developers' minds. Linux already has the marked advantage of being very low-cost (essentially free with optional support costs) and it sports a rapidly growing developer base. The ability to build code libraries once and port them to a wide range of platforms is a very strong argument in its favor; unfortunately, the one thing that Linux is lacking at this stage may be the most important factor of all: a champion. IBM has championed Linux on the server yet has opted to license the Palm OS and Windows CE for their own mobile computing products. Linux has grown through a grass-roots movement of developers and administrators looking for a low-cost, powerful, extensible operating system. The end result of this has been a large network of developers but very few companies that have championed the product and marketed it agressively. Transmeta has taken some criticism from industry watchers who cite Transmeta's lack of current customers as evidence that the technology is more vapor than real. If it, in fact, works as advertised, will that be enough to overcome the early lead other vendors have taken in this market niche?

In short, the questions to ask at this early stage of the game are these:

• How important is it for a handheld OS to offer the advantages of robustness and technical superiority versus the advantages of simplicity and connectivity?
• Does developer mindshare equate to market success if all competitors are starting on equal footing?
• Conversely, does market share success drive developers to your platform and ensure future success?

I'd certainly be interested in knowing what you, the readers, think. I think we can ignore market share numbers for just a bit longer. The race hasn't started in earnest yet but I think the competitors are all present and accounted for. In fact, they are approaching the starting line.