FreeBSD is Not Linux
Bob Bruce, the founder of FreeBSD Mall, has a complaint. "FreeBSD doesn't get the credit it deserves because, in surveys, it is lumped under UNIX or, more often, under Linux," he says. "People think that if it's open source, and it's an operating system, it must be Linux."
Why is Linux better known than BSD? Why do the two get confused by the non-technical press? Compared with commercial operating systems like Windows, Mac OS X, and Solaris, both BSD and Linux present a more complicated picture. As is the nature of open source developments, each has multiple distributions. The Linux site at Linux.
Even so, Linux is seen as a more coherent whole, a clearer entity that is easier to write about and discuss. Linux distributions have a common Linux kernel-Linux 2.
And so BSD is less publicly identified with the open source movement, and is less visible in the U.
Not that NetBSD, OpenBSD, and Wind River's BSD/
Bob Bruce says that "NetBSD has a big edge in embedded systems, because it runs on more platforms than any other operating system, bar none. There's also Wasabi Systems which is set up to do NetBSD on embedded systems, and just got $2 million in funding to go forward. Wind River is also focusing in on embedded systems, and BSD/
If an operating system is best known by its prominent installations, FreeBSD supporters point to two customers. The first is Yahoo!, which runs most of its webservers on FreeBSD. The second is Apple, whose Darwin-the core foundation technology for Mac OS X and Mac OS Server-is based on the Mach 3.
The biggest complaint about Free BSD-and BSD in general-is the relative scarcity of applications. Hubley writes that FreeBSD must battle complaints that there are "few third party applications and no native applications." Bob Bruce counters that "because FreeBSD is Linux binary compatible, anything that runs on Linux runs on FreeBSD as well. For example, the FreeBSD Netscape browser isn't updated as often as the Linux browser, so I just run the Linux browser on FreeBSD."
The FreeBSD project began in 1993, growing out of the "Unofficial 386BSD Patchkit," an intermediate snapshot of the operating system created by Bill Jolitz. After Jolitz withdrew support, the project's name was changed to "FreeBSD," and the coordinators--Jordan Hubbard, Nate Williams, and Rod Grimes--contacted Walnut Creek CDROM and its owner, Bob Bruce, with the idea of distributing the OS in CD-ROM format. "Without Walnut Creek CDROM's almost unprecedented degree of faith in what was, at the time, a completely unknown project, it is quite unlikely that FreeBSD would have gotten as far, as fast, as it has today," wrote Hubbard.
In 2000, Walnut Creek CDROM merged with BSDi, the company formed by members of the Computer Systems Research Group (CSRG) at the University of California at Berkeley. In April 2001, Wind River Systems, Inc. purchased the software assets of BSDi, which included BSD/
"FreeBSD didn't fit in with Wind River's business model," Bruce recalls. "So they called me and other companies as well-Daemon News, for one-looking for a successor organization. I was the one they picked, mainly because I had run the unit before and knew how the systems worked. The people that were transferred with the unit were people who originally worked for me."
When I spoke with Bruce, his reemergence as the primary commercial keeper of FreeBSD had only just happened and he was still trying to figure out what do next. His immediate plans were to give his employees, who have seen three changes in ownership in the past 18 months, some stability. "In three months or so, after we've gotten through a release cycle and gotten all the systems going, we'll take a look at more ambitious things. Right now, we're one of the driving forces of getting the releases out, we're doing a FreeBSD toolkit update, and an update of the FreeBSD Handbook. There's also other projects-we're contracting with someone to develop a graphical installer for FreeBSD."
0 slated to debut in November
Although FreeBSD Mall has regained its role as the principal commercial source for FreeBSD CD-ROMs and OS support, the brain trust for future development remains the FreeBSD Project. Warner Losh, a member of the eight-person core team that constitutes the project's governing body, is a senior software developer for Timing Solutions in Boulder Colorado, where he works on device drivers for timing devices and control programs.
Losh doesn't think having competing BSD open source projects has hurt overall acceptance. Perhaps it has even helped. "In large part, the competition between the three projects has raised the bar for all the projects, because we are constantly competing with NetBSD or OpenBSD to see who has the best hardware support or the latest cool feature. And the fact that we can easily share code between the distributions helps everybody."
He says that FreeBSD's principal claim to fame remains its optimization for the Intel platform. "The systems you deploy with FreeBSD on Intel tend to run better and faster than the other distributions of BSD. FreeBSD has the largest number of third-party applications. And finally, we have the most extensive security reporting mechanism in place of the FreeBSD distributions. Combined, these provide a compelling reason to run FreeBSD versus NetBSD or OpenBSD."
The FreeBSD project maintains two "branches": the current branch where ongoing development takes place, and the stable brand, the source for new releases. When a new major release takes place, the project creates a new stable branch. That will happen again with release 5.
Losh estimates the project has some 300 FreeBSD committers-developers with permission to write to the FreeBSD repository. "They are all over the world, with probably 30 to 40 developers in Japan, and another 10 to 15 developers in the rest of Asia. Over half the developers are in the U.
One of Losh's roles is to coordinate between Japanese and U.
Losh says that worldwide, three groups of people are involved with FreeBSD development. One group uses FreeBSD in their job, and typically makes incremental improvements. He puts himself partly in that category. The other class are people who enjoy BSD, and contribute for the pure joy of it. "I'm also in that group," he says. A third group involves people doing localization. "That's definitely true in Japan." Indeed, FreeBSD, and perhaps BSD in general, has received a warmer welcome in Japan than in the U.
Sidebar: Jun-ichiro "itojun" Hagino on NetBSD
While Warner Losh is serving as a cultural bridge between FreeBSD and Japanese, Jun-ichiro "itojun" Hagino is doing the same between NetBSD and America. itojun is both a core researcher on the KAME Project and a member of NetBSD Project's core group. He says NetBSD has more than 200 developers supporting the OS, which is known for its wide number of ports. These include mainstream processors like the Intel i386 and Motorola 68k series on the Macintosh; unexpected machines, like the Sony Playstation 2 and the defunct Sega Dreamcast; lower-power chips like the StrongARM, and RISC processors of many stripes.
"NetBSD has clean bus architecture framework, which helps us share most of the driver code across our supported platforms," he said, speaking in English by telephone and in e-mail correspondence. "Common device driver source code is used for i386, Alpha and PowerPC, for example, and porting drivers between NetBSD and OpenBSD is easy. Porting to and from FreeBSD or BSD/
OS is harder, due to driver framework differences. We of course try to follow POSIX and other UNIX standards."
itojun says that NetBSD is second to FreeBSD in popularity among BSD distributions. "Because of its clean bus architecture, NetBSD attracts many people interested in doing new implementations and experiments using non-Intel CPUs." For example, itojun's own company, IIJ (Internet Initiative Japan), uses the operating system on its family of small routers, which incorporate an Hitachi SuperH processor. "NetBSD will continue to attract people doing high-performance development. We hope to see more people using on desktops, as well as more cooperation with other BSD groups." Does that mean BSD will eventually be unified under a common distribution. "I hope so, but I don't think it's possible. At this point, the distributions are way too different."