Originally Published: Monday, 27 March 2000 Author: Tom Dominico, Jr.
Published to: featured_articles/Featured Articles Page: 1/1 - [Printable]

Strength in Diversity

Concerns about the possible "fragmentation" of Linux are nothing new. However, they have recently been revived in several high-profile articles, and were the subject of a recent article here on Linux.com. Some IT analysts are predicting doom for Linux, with memories of Unix forks fresh in their minds.

   Page 1 of 1  

Concerns about the possible "fragmentation" of Linux are nothing new. However, they have recently been revived in several high-profile articles, and were the subject of a recent article here on Linux.com. Some IT analysts are predicting doom for Linux, with memories of Unix forks fresh in their minds. They claim that the existence of many distributions, each fighting for market share, will be the death of Linux. Personally, I believe that diversity is our greatest strength.

Many Distributions != Fragmentation

I feel that the presence of multiple Linux vendors is a strength, not a weakness. The battle for market share is causing them to target different sectors of the Linux market. For example, Corel and Caldera have targeted Windows users, with features such as easy partitioning and a familiar-looking desktop. SuSE, with their 6-CD set, is known for throwing in everything but the kitchen sink, and for being a nicely polished, all-around distribution. Meanwhile, Red Hat seems to be focusing on the corporate market. Debian, the only non-commercial distribution, is making a name for itself as being technically superior. There are still others such as Slackware, Stampede, Storm, and so on, each bringing something unique to the table.

The beauty of this is that Linux is not "one size fits all." You can choose the distribution that suits your needs. (Funny how that concept of "choice" seems to keep popping up, as if it were very important to the nature of Linux.) The best part of all, though, is that everyone gets the same stable, powerful OS, no matter what distribution they choose. The kernel is the same across distributions (perhaps not the same version, but essentially the same, for our purposes).

This is something that you won't find on any other platform. For example, Windows users have traditionally had to choose from a "consumer" version (with a very unstable kernel), or a "workstation" version, with better stability, but not tailored for consumers. Supposedly this will change in the future, but I'm not holding my breath. It seems to me that Linux is able to provide a combination of tailored features and stability that no other OS can match. I'd call that sort of diversity a strength, not a weakness.

The Linux Standard Base

The LSB is a subject of some contention, though it has yet to produce any real results. I don't really have a problem with standardizing directory structures, mostly because it would help those who write documentation. My concern is that the standard be chosen on technical merit alone, not just because it is in use by the biggest or most popular vendor.

Similarly, I don't see the harm in defining a standard definition of a minimal Linux distribution, as long as it is just that: minimal. It must only serve as a "starting point," and should not make choices as to which toolkits to use (Gtk+ vs. Qt, etc.). I don't see the need for a standard package format. As long as things get installed in the right places, there shouldn't be any problem. I believe in allowing as much personal choice as possible, and that includes choosing between RPMs, DEBs, and so on.

I think that it may be necessary to establish a minimal LSB, if only to quell the fears of Linux incompatibility. Whether or not it is a real issue, it has taken on a life of its own. In order to foster widespread corporate adoption of Linux, we have to be sensitive to those issues. Technically speaking, I don't see how binary incompatibility is an issue with an open-source operating system. When you've got access to an application's source, you can always modify it as needed. However, that does create a fork in the application which must be maintained, and I think it's best if we avoid those sort of situations.

A Fork in the Kernel?

Frankly, I don't see this happening. There's no real reason for this to happen, and if it did, I doubt anyone would use the new kernel. Most Linux users respect the guidance of people like Linus Torvalds and Alan Cox, and I doubt they'd suddenly switch to a rival kernel.

Proprietary Additions to Distributions

I do consider this to be a potential problem. It is very important that we encourage Linux vendors to open-source any additions they make to a distribution. If they don't, they're not staying true to the ideals of open source software, and I don't think we should give them our business until they do.

This is a rather Microsoft-like tactic ("embrace and extend"), and it has no place in the Linux world. Being able to borrow from and build upon existing code is very important to maintaining the highest quality software. It's a central reason for the success of the open source movement, and that's why we must encourage all Linux vendors to open source any proprietary code.


I hope that I've laid to rest some of the concerns about Linux viability. At the same time, I hope that I've raised a few other issues that I feel to be important to the continued success of Linux. It's becoming increasingly difficult for the nay-sayers out there to bash Linux, as its popularity continues to soar.

We are standing on the verge of widespread, mainstream adoption by both corporations and individuals alike. Let's keep the momentum going by resolving these outstanding issues, and by clearing up any FUD we encounter with fact-based advocacy. Viva La Revolution!

Tom Dominico (tomd@linux.com) is a programmer, database administrator, and recent Linux convert. Cursed with insomnia, he spends his sleepless nights chatting on IRC, tweaking his Linux workstation, and reading everything he can get his hands on.

   Page 1 of 1