Comment: Desktop Fragmentation

by Thorsten Leemhuis

Competition may be good for business, but competition among Linux desktops is currently so fierce that it may end up being to everyone's detriment in the medium-term.

In a recent posting on his blog entitled Staring Into The Abyss, long-time GTK+ and GNOME developer Benjamin Otte addressed the idea that GNOME is both losing market share and declining in significance. That might sound like a GNOME-specific problem, but it also has implications for other desktops.

Whether you love or hate GNOME, it was, until about 18 months ago, the dominant Linux desktop, with KDE as its main rival. Magazine articles and internet howto's didn't need a ream of 'if, then, else' structures to explain how to set up dual monitors, audio output or desktop wallpapers or to change the date or time. Its dominant position has meant that GNOME has set the standard in many areas and has helped many big changes to make the breakthrough – examples include NetworkManager, PulseAudio, PackageKit, D-Bus, UPower and udisks. None of these components is perfect and some started out with a whole heap of problems, but ultimately they have resulted in tangible improvements in Linux' suitability as a home and work OS, and have subsequently found there way into KDE.

The last 18 months have, however, seen a transformation in the desktop world. It started with Canonical's switch to Unity, which dragged many former GNOME users with it. This was soon followed by the hotly-awaited GNOME version 3 and GNOME Shell. This was so different to conventional desktops that many GNOME users were reluctant to upgrade. The result was GNOME 2 fork MATE, version 1.4 of which has just been released, and GNOME Shell fork Cinnamon, with a more conventional desktop concept which quickly attracted a lot of fans. Many GNOME 2 users without the stomach for GNOME 3 or Unity also switched to Xfce, further raising its profile.

Rather than two, there are thus now six desktops enjoying the favour of users. This sort of competition is certain to result in the odd useful interface concept, but this is sure to be outweighed by the disadvantages which this level of competition and fragmentation will create. The already spartan testing time available to beta testers will now be spread across multiple desktop environments. Greater coordination will be required between developers working on components used by multiple desktops.

This will make major restructuring work in particular more complicated, as this will now consume developers' time and patience. Even when it was just GNOME and KDE, collaboration sometimes left much to be desired. And should collaboration between multiple desktops prove less difficult than feared, there will still be plenty of things which will be reinvented for each desktop, tying up more developer time.

That sort of energy could certainly be usefully employed solving more serious problems, of which, in the Linux desktop world, there are plenty, as is evident whenever you watch a Linux novice try to take his or her first steps on a Linux desktop. Fragmentation means that resolution of such problems will take a little longer. If I still believed in the long-standing Linux on the desktop pipe dream, I would posit that it has once again receded even further into the distance.

Agree? Disagree? Have your say in the forum.