KDE and GNOME have decided that because they're not big and complicated enough already, they might work better if they have their own custom distributions underneath. What's the worst that could happen?
A talk from this year's KDE conference, Akademy 2024, looks like it's going to become real. The talk, by KDE developer Harald Sitter, was entitled An Operating System of Our Own, and the idea sounds simple enough: Sitter proposed an official KDE Linux distribution. Now the proposal is gathering steam and a plan is coming together for an official KDE Linux - codenamed "Project Banana."
It's important to note at this point that there already is an official showcase KDE distro, which is called KDE Neon. Neon comprises the latest version of the KDE Plasma desktop, pre-installed on top of a recent LTS version of Ubuntu.
In case that sounds like it's too easy, it is: there are no fewer than four different editions of KDE Neon available for download, all for x86-64. This is entirely in keeping with how the KDE project as a whole works: for instance, if you search the KDE Applications website for "text editor", you'll find three: Kate, KWrite, and Nota. If you search for "file manager", you'll find four; and "web browser", three. Within the desktop, there are multiple start-menu tools, multiple app-switching panel-button bars, and so on. Even the "About" option on the "Help" menu is duplicated: one tells you version information (in one of two version formats, either decimal-number based or date-based), and one tells you about KDE as a whole.
KDE enthusiasts see nothing wrong with this, and as far as the Reg FOSS desk can tell, consider this a perfectly reasonable and proportionate response to trying to accommodate the different needs of different users. Even here in The Register comments sections, the author has received apparently completely serious suggestions that a viable solution to the problem of too many incomplete, semi-functional desktops is to write another, better desktop.
The KDE Linux proposal also suggests that the new distribution should have multiple variants from its inception:
It is not yet clear to us if KDE Linux will take the place of KDE Neon, or be an alternative to it. (We need scarcely point out that there are several existing projects with very similar scopes to KDE Neon itself, such as Kubuntu, the official Ubuntu flavor with KDE Plasma, and Tuxedo OS, which like Pop!_OS is offered for free by a hardware vendor, but can be used on any other PC as well.
The plan is that KDE Linux will resemble Valve's Steam OS 3 in several ways: based on Arch Linux, using Btrfs, immutable, and with two root partitions that update each other (as Chrome OS does). The idea is that by moving away from the fixed two-year release cycle of Ubuntu's LTS editions, the distro will be free to incorporate newer components as needed by the KDE desktop, and so deliver a more stable, reliable experience.
Not to be left out, a developer from the GNOME project, Adrian Vovk, has a corresponding proposal: to take GNOME OS mainstream. In his modest title, to build A Desktop for All:
GNOME OS is an existing distribution, but with a very different purpose to KDE Neon. While the purpose of Neon is as a technology demonstrator, in its current form, GNOME OS is more of a technology testbed.
Its title of "Nightly" is a clue here: GNOME OS is rebuilt daily, in line with the principles of Continuous Integration/Continuous Delivery. The recommended way to run GNOME OS is in a VM using GNOME Boxes, but in principle it can also be installed on bare metal - so long as it uses UEFI and doesn't need any additional drivers.
Rather than being based on any existing distro, GNOME OS is constructed largely from scratch every time, using Apache Buildstream, which assembles all the components needed to support that day's build of the GNOME desktop from components drawn from Freedesktop SDK.
This vulture has tried to get GNOME OS running several times, but without success, possibly because we don't routinely use GNOME or Boxes. (Different machines in the Irish Sea wing of Vulture Towers use Ubuntu Unity, MX Linux, Alpine Linux, and several other OSes and distros, but apparently nothing that is close enough.
There are signs that Adrian Volk has been interested in such an idea for some time. He has a previous project, carbonOS, which had comparable goals: an immutable distro, self-updating and with automatic rollback, built from the metal up to support an unmodified GNOME desktop.
The idea is not a bad one at all, but as with KDE Linux, there are already several efforts underway to create things that are very similar: immutable, failure-tolerant distros for non-technical users. Fedora Silverblue is one, and Endless OS has similar goals - although it does modify the GNOME desktop, but only slightly and less with each release. We looked at Endless OS 5 last year and Endless OS 6 back in May.
Reporting on the world of Linux and FOSS is often frustrating, and one small aspect of this is a strong tendency towards fractal levels of self-parody. The problem is exacerbated by rigorous adherence to Poe's Law.
The GNOME project was created in response to KDE, and in important ways they continue to influence one another still, after more than a quarter of a century.
In some ways, KDE is very European: it was built by a group of German developers, using the Norwegian Qt toolkit, in the C++ language designed by Danish developer Bjarne Stroustroup, whose object-oriented influence was the Norwegian Simula programming language [Norwegian language].
Unix, of course, was an American project, and while Linux was created in Finland, in its early days its growth was driven by American projects such as Slackware, Red Hat, and Debian. Red Hat and some other more traditional folks looked upon KDE with some suspicion in the early days, partly because Qt wasn't pure FOSS back then, and also because traditional Unix folks tend to prefer plain old C, for all its many faults, over C++. Some of the founding principles of GNOME were to avoid Qt (or re-implementing Qt), and to avoid making people use C++.
Given the propensity of FOSS folks to fork projects with very little provocation, any of these factors in isolation could have been enough. In combination, they have led to projects with very different cultures, as caricatured in 2002.
Both these projects have laudable goals, and there is nothing to criticize here. We too want to see easier Linux distros for non-technical folks. There's every reason why mature desktop projects, with substantial libraries of well-integrated apps, would want flagship platforms to show off their latest and greatest versions.
GNOME is arguably the most powerful force in the Linux desktop space. GNOME has a clearly defined release cycle, and both Fedora and Ubuntu synchronize with it. That doesn't automatically mean that the distro's goals are the same as the desktop's goals, though. Even so, we have reservations about the wisdom of trying to stabilize a CI/CD driven OS stack for public use. It disregards all the effort that multiple distros have put into making stable end-user OSes and reinvents it. It's rewriting the whole OS stack, and total rewrites are famously one of the Things You Should Never Do.
KDE was that dominant once: around the turn of the century, most of the leading distros, such as Mandrake Linux, Caldera OpenLinux, Lindows/Lispire/Freespire, and Corel LinuxOS, all shipped with KDE. KDE does not have a fixed release cycle, though, leading to problems synching up with fixed-release-cycle distros. As such, choosing the basis of a rolling-release distro such as Arch Linux makes sense. What the KDE Linux project page says it seeks to achieve, configuring Arch to make it more reliable and resilient, is a great idea.
One of the single most-heard complaints outsiders have about the world of Linux distributions is that there are too many of them, which makes it too hard to choose. A common result of being faced with an embarrassment of riches is analysis paralysis.
When there are too many competing standards, inventing a new standard to cover everyone's use cases is not a good answer. But evaluating the relevance of XKCD 927 in this instance, add an extra zero on the numbers.
Edge-exclusive tool promises 'second set of eyes' for browsing
Microsoft's OS sure loves throwing your creds at remote systems
How much AI does one subscriber need?
re:Invent Calling everything SageMaker is confusing - but a new name would have been worse says AWS
How to tell a customer they're an idiot without telling them they're an idiot
Damage likely limited to those running bots with private key access
Louisiana facility's three natural gas turbine plants to churn out 2,262 MW