This week wraps up my speculation on how a divided Microsoft could be as bad for Linux as the antitrust trial has been good. If that isn't enough irony for you, try this: Linux got where it is today because it is open source. But if it turns out that a divided Microsoft does cause problems for Linux, it will be because Linux is open source.
The open-source nature of Linux is both its greatest advantage and its biggest liability. The advantage of open source is that you are free to take the source and improve it, as long as you contribute your changes back to the community. But if all the Linux distributions take the core Linux system and improve it as they see fit, they fragment Linux.
The best way to prevent this fragmentation is to respect the final authority of the core kernel developers and the developers of key libraries such as glibc and accept what they deliver as the standard. But if everyone follows this advice, what you have isn't truly open source: it's a tacit agreement to treat Linux as though it were under a licence that requires you to submit your changes to a central governing authority for approval. Ironically, that sort of arrangement normally starts riots in the open-source community.
This dichotomy speaks directly to the two complaints I hear most often from ISVs when it comes to porting enterprise-scale server applications software to Linux. First, ISVs don't want to develop for and support several Linux distributions. They want either to focus on a single dominant distribution or to help promote a standard with which all distributions will comply. They also need Linux to mature faster to support features their software already exploits with other versions of Unix.
The charter of Linux Standard Base (LSB) is to develop and promote a standard platform so that ISVs can support more than just the dominant Linux distribution. But LSB documents only a minimal standard as defined by the development community. Currently, it does not directly address the issue of improvements to Linux.
So how does an ISV get the improvements it wants into Linux? Ask the development community, which usually tells the ISV to submit a patch to the kernel developers.
There are two things wrong with that answer. First, these companies aren't inclined to get into the business of Linux kernel development. It's not that the ISVs don't have the resources (as in money, talent, or both) to devote to kernel development. The problem is even if they dedicate the resources to provide patches, there is no guarantee that the core development group will accept them. And they can't influence the decisions of the development community through commercial pressure.
This is a difficult pill for ISVs to swallow, because they're used to getting the improvements they want from Unix vendors simply by approaching the vendors with their needs. Server platform vendors are eager to please ISVs because more applications support means more market share for their platform.
The obvious solution -- and the one that seems to be playing out right now -- is for the ISV to pick a commercial Linux distributor and try to get its needs met from an organisation that is motivated by profits. Because Linux is open source, distributors are free to patch the Linux kernel, GNU utilities, and whatever else is necessary to please the ISVs. Competing commercial Linux distributions will try to offer the same improvements to woo the ISVs. But because no single commercial distribution wants to standardise on another commercial distribution as a base, every distributor will make its own improvements on its own schedule. Fragmentation results, making Linux a less attractive platform.
There is one other alternative for ISVs: they can get behind Windows 2000 as the future enterprise server platform. Right now vendors aren't eager to promote Windows 2000 at the server because Microsoft would use a dominant position to clobber competitors, including the very ISVs who helped Windows 2000 win at the server. But splitting up Microsoft and adequately policing the two companies could remove the danger, making Windows a more attractive platform -- perhaps more attractive than Linux. The bottom line is if Microsoft is split quickly, the next two years could be very trying times for Linux and open source.
Nicholas Petreley is the founding editor of LinuxWorld (http://www.linuxworld.com). Reach him at firstname.lastname@example.org.