Trouble ahead for Linux?

To fork or not to fork, that isn't the question. Some people have suggested that Linux is a dangerous operating-system choice because Linux might fork into incompatible versions. Give me a break.
Written by Steven J.Vaughan Nichols, Contributor on

Of course Linux is forking. It's always forking. But--important lesson ahead--it doesn't matter. Why? Because Linux is a Gnu Public License open-source OS. And that means that everyone gets to see and use your changes.

In Linux, good changes are incorporated into the one true, Linus Torvalds blessed Linux kernel, and other changes aren't. If a change doesn't make the canonical code, it withers on the vine. In other open-source projects, like Apache and Samba, there's no Linus, but it works the same way. The good get in, the bad are tossed on the junk pile.

Think about it. That's why there's a stable kernel for the Linux companies to build their commercial operating-system packages around and a development version for testing out new ideas.

Of course, the Linux companies add some of their own stuff to the overall operating system. But--whether it's functionality, like the clustering that TurboLinux adds to TurboCluster, or the chrome of a user-friendly GUI, such as Corel's--the fundamental code, the real operating system, remains the same.

While having Linus around to keep the system on the straight and narrow course is great; Linux is a lot bigger than one person. If Linus, God forbid, were to get hired by Microsoft tomorrow, Linux would get along without him.

It would do so because there are hundreds of open-source programmers out there, who want--indeed, need--to have Linux stay on one standard base. While efforts like the Linux Standard Base movement will help, even if it continues its slow pace, the Linux developer community would find some way of keeping the code straight. They really don't have any choice in the matter.

If you don't understand the open-source model or know your computer history, let me give you the top five reasons why Linux isn't going to fork its way into disaster. Number Five: The Unix Lesson Unix got the reputation as an also ran operating system because it was--and is--a closed-source OS with many competing vendors. Think about it. Today, when you buy Unix, you're not really buying Unix; you're buying AIX, Solaris or Tru64.

Closed source can work when one company--a Microsoft with Windows or a DEC with VMS--calls all the shots. Closed source with multiple vendors is suicide. Here, forking really will zap you. You need to look no further than the Unix graveyard to see the tombstones of forgotten companies, like Interactive Systems and UHC, and witness the fate of most that tried the proprietary product/multiple vendor approach.

Number Four: Can Forking Really Be That Bad When Microsoft Does It?

Let's see, I count more than a half a dozen significantly different current versions of Windows: three W2Ks, three NTs, Windows 95, Windows 98 and Windows CE. There are fundamental incompatibles between all four major variants. Funny, I don't see any stories predicting Microsoft's fall because of forking. Bill Gates is still the richest man on Earth, right?

Number Three: Open Source

Now, you can fork in a significant way in open source if you don't remove fundamental compatibility. For example, emacs and Xemacs, the Swiss army knives of Unix word processors, split have very different code trees. From the user's perspective, there's little practical difference and both programs have file format compatibility.

The BSDs, on the other hand, have preserved essential compatibility between BSD/OS, FreeBSD, NetBSD and OpenBSD, but they're perceived as being significantly forked. In addition, the BSD developers efforts have been split between three (given that BSD/OS and FreeBSD will merge successfully) different projects. In this, the BSDs, with their much smaller market share, serve as a useful cautionary lesson to Linux businesses.

Last, but not least, open-source culture is opposed to significant forking. For more on that, see such fundamental open-source texts as Eric Raymond's Homesteading the Noosphere (LINK: http://www.tuxedo.org/~esr/writings/homesteading).

Number Two: Do You Want Applications With That?

This is really simple. Multiple systems mean fewer--a lot fewer--applications. Fewer applications means fewer customers. Fewer customers means incompatible Linux vendors would end up saying, "Do you want fries with that?"

Number One: Linux Vendors Are Sane. United they stand, divided they fall. I know that. They know that. And, that's the end of the story.

Are there Unix skeletons in the Linux cupboard? Let Jesse Berst show you what the implications could be for your company. Go to AnchorDesk UK to read the news comment.

Editorial standards