Linux Comes to T-Engine

Steven J. Searle

Web Master, TRON Web

On March 18, 2003, and in the days that followed, foreign-language news articles dealing with T-Engine began to appear all over the World Wide Web. English, Russian, and French, they all had something to say that was very important, but not about T-Engine. Rather it was about the future of real-time, embedded Linux. As a result of requests from various corporate members in the T-Engine Forum, the premier definer of real-time Linux, MontaVista Software Inc., anounced that it would port its real-time Linux implementation to the T-Engine open development platform. To that end, it was also announced that a working group to support this effort would be established inside the T-Engine Forum to handle the technical matters and draw up a specification. Thus in the near future, standard interfaces for Linux on T-Engine are going to be decided, which will provide Linux middleware developers and their customers with an environment on which to run their applications.

What took place on the Internet bulletin boards shortly afterward was a case of deja vu reminiscent of the fall of 2001 when TRON Project Leader Ken Sakamura was jointly awarded a prize for his contributions to computer infrastructure development along with Richard Stallman and Linus Torvalds by the Takeda Foundation. Then, as recently, Linux-related Internet bulletin boards were filled with people asking: "what's TRON?" As usual, people missed the essential point. Linux on T-Engine, which is officially referred to "T-Linux," is no big deal; it's just one more middleware environment on T-Engine, where it will be competing with T-Java, for example. What's more important is T-Engine itself and the standardization it brings to embedded systems development. The fact that massive standardization is taking place in this field at the same time that we are moving from PC-centric computing to ubiquitous computing based on real-time embedded technologies was totally lost on most people.

There are some people, both inside and outside the Linux community, who believe that Linux and TRON are in competition with one another, but this is not the case. In fact, even real-time Linux for embedded use and TRON OS are not in competiton with each other. They are based on different computing models (the former uses a time-sharing multitasking scheme, while the latter uses a priority-based multitasking scheme), their kernels are not the same size (real-time Linux is measured in megabytes, while TRON OS is measured in kilobytes), and their task switching times are very different (that of Linux is measured in milliseconds, while that of TRON OS is measured in microseconds). If Linux is in competition with anything, it's FreeBSD. However, since FreeBSD is an open source operating system development project, it can't be run out of business; and since it is based on a computing model similar to Linux, application programs can be freely ported from one operating system to the other.

There were, fortunately, a few comments on the Linux-related bulletin boards by thinking people. One person, who was obviously knowledgeable about BTRON, said that it would be nice to bring TRON Code and the TRON Multilingual Environment to the Linux operating system. Bravo! A great idea that I support wholeheartedly. Actually, I proposed something along these lines way back in the early 1990s in a meeting with TRON researchers at the University of Tokyo, but I was told that it would be difficult if the lower layers of a non-TRON operating system, e.g., UNIX, didn't have the resources necessary to support the TRON Multilingual Environment. Well, here we are in the early 2000s, and at last a UNIX-like operating system is finally going to get engineered into a TRON-based development platform. This will make all sorts of things possible, and I hope that it will lead to a lot of cross fertilization between the two open, royalty-free operating system movements.