The Very Long Night of FreeBSD

From AJS.COM

The Very Long Night of FreeBSD

by Aaron Sherman <ajs@ajs.com>

This article appeared in the October 1999 issue of Daemon News.

As a Slashdot reader/poster, BSD 4.2 user back in the mists of time, current Linux advocate and a previous contributor to Daemon News, I felt compelled to respond to James Howard's article "The Real FreeBSD".

The article was directed primarily toward those who are not familiar with FreeBSD or BSD in general, for that matter. To be more specific, it was aimed at readers of Slashdot (a pro-Open Source Web site which tends to cater to and attract a heavily Linux-oriented audience). It covered eight "myths" about FreeBSD which the author felt needed to be cleared up. Mr. Howard may be correct. Those myths may be pervasive and may harm the reputation of FreeBSD, but his comments serve to undermine his stated goals almost as effectively as Wes Peters' previous column, Daemon's Advocate, July 1999, served to resolve them. To that end, I pose a few myth busters of my own in the hopes that all of us who support Free (Source Code Enabled; Open; Anti-Proprietary) Software can learn to "just get along."

"There are four wholly separate BSD derived operating systems ... This is not fragmentation of the market."

Well, actually it is. That's okay; don't come running with pitchforks and torches yet! The key here is that BSD is a fragmented market, and Linux is a fragmented market. For that matter so is Microsoft (though they have the power, and perhaps the will to reconsolidate... time will tell). The BSD world is pretty much textbook fragmentation, but in the software realm it is very hard to avoid this. Some users need speed at all costs. Some users need platform portability. Some users need security. Linux too will begin to feel this crunch now that the kernel is maturing.

Fragmentation leads to an increase in management requirements which in the proprietary software world are gargantuan, and lead to far too much wasted effort and too many reimplementations (e.g. the proprietary UNIX world). However, even when source code is not actively levied from another project, Open Source projects can benefit from the increased flow of ideas that result in having the source. This means that when FreeBSD implements a new type of network protocol, NetBSD has the freedom to look over the code and even use it if they want. Thus, fragmentation is much less damaging in the non-proprietary world of Open Source. Even the BSD and Linux camps can look at each other's code and learn valuable lessons from both successes and failures.

"An effort to replace GNU code with freely re-distributable implementations is also underway."

I have heard a lot of good arguments for and against the GNU licensing scheme. However, none of them leave any doubt that what the GPL does is allow the software to always remain redistributable. It is this very fact that leaves a bad taste in the mouth of the GPL's detractors (and just for fun, I won't tell you which side of that debate I'm on). Implying that GNU code is not redistributable is just silly, and harms the other points rather than supporting them.

This is later re-stated as "Since this is the only restriction on further redistribution, the BSDs are the only freely redistributable operating systems in common use." This too is deeply flawed, but adds the "one restriction isn't restrictive" idea, which flies in the face of previous rationalization. If we are to redefine free to mean that the software licensing may not be restrictive, then public domain software is the only free software and neither BSD nor Linux fit that metric.

"In many ways, Linux is a FreeBSD clone."

Neither system is a clone of the other. Rather, Linux and FreeBSD have a common UNIX ancestry. Linux is a reimplementation kernel-wise, but much of the rest of the system software and utilities are actually common, and either are direct descendants of or reimplementations of other operating systems' tools. If he had said "Linux is a BSD clone" I would have agreed part-way (as Linux also derives some of its ideas from SystemV UNIX and has many features of its own).

"many [FreeBSD users] are simply curious about why a new user would choose Linux over FreeBSD, despite FreeBSD's technical superiority."

I just thought I'd throw this comment out for people to mull over. Clearly, the author believes that FreeBSD is technically superior to Linux. Good for him I guess. I could point out the number of areas in which Linux has support for high-end hardware that FreeBSD does not (not to mention hordes of low-end hardware), but that's not really the point. In the words of the infamous Pythons, "This is not an argument!"

Before someone points out that FreeBSD is faster than Linux for most, if not all, of the benchmarks that have been performed, let me remind you that DOS is faster than both platforms. Speed is not the only yard-stick here, as I pointed out in relation to the fragmentation issue.

Conclusion

Here it is in all its splendor, kids: *BSD, Linux and the micro-kernel-from-MIT-of-the-week are Open Source operating systems which are all to be highly praised for daring the waters of the "Get Your Hand Off of My Stack" proprietary software world. We are doing something which will be remembered in the software history books as one of the definitive points in favor of faith in human nature. Let's not start sniping at each-other because we don't like the wording of this or that license or because someone likes their toy more than your toy. It's a big sand-box and we all have a lot to learn and teach to/from each-other.