Some time ago I wrote about how OpenMandriva Lx and Mageia (both based on Mandriva Linux) have overlapping interests and how this causes the developers to unnecessarily duplicate their efforts.
The phenomenon is called forking and usually occurs when certain communities don’t get along in terms of philosophy or priorities. Forking is also a sign of freedom (in a sense), intrinsic to Linux Land. However, as Keith Curtis (the author of After the Software Wars) wrote on his blog, celebrating forking as a solely positive phenomenon could be compared to saying ‘Divorces are good for the society’. Recently, I more and more agree with that metaphor. Freedom should be sought after only to a certain degree. Overexercised freedom can bring demise to the community in the form of too finely dispersed effort…
Keith gives the example of Manjaro Linux. I believe it is a very good one. Manjaro strives to create a user-friendly experience of Arch Linux for the less computer-savvy. Unlike Keith, I used Manjaro for quite some time and was very pleased with it. What Curtis has rightfully pointed out, though, was that Arch Linux does not profit from the additional effort put into developing Manjaro, because of separate package repositories. A contrary example is Archbang. It directly builds on Arch Linux (uses its repositories), in addition supplying a number of components for better out-of-the-box experience (Opera as a web browser, Leafpad as a text editor, relatively simple console-based installer, per-configured X Window System, etc.). Thereby, we have two Arch Linux forks (though Archbang is more of a spin/respin) – one that strives to make Arch more appealing by simplifying the setup process (Archbang) and one that attempts to be ‘better than’ (Manjaro).
Debian is probably the most widely used Linux distribution, signified by the enormous number of distributions based on it. Unfortunately, many of them try to ‘re-invent the wheel’, out of the shear possibility to do so. I mean, honestly, does every major country in the world need their own Debian-based distro? Would it not be easier to provide repository mirror servers and language support instead, in addition to GUIs and initial setup for better out-of-the-box experience?
To be fair though, there is a number of distributions with a well defined goal that add, rather than subtract from existing communities. Sabayon tries to make Gentoo attractive, despite the extremely steep learning curve of its base distro. Tanglu brings forth a convenient aspect of Debian Testing by giving Debian developers the freedom to continue releasing packages into Debian Testing repositories, while Debian Testing proper is frozen prior to a new Debian Stable release.
Nevertheless, I would like to end my musings with a come-back to the title of this entry. It refers to a slightly racist joke concerning household lighting maintenance. ‘How many <put your least favorite nationality here> do you need to change a single light bulb?’ The answer is: A hundred – 1 holds the light bulb in place and the rest rotates the whole building to screw the faulty one out and then a new one in.