Xlibre fork lights a fire under long-dormant X.org development

Comment Considerable new activity is happening both in the established X.org X11 server and around its new fork, Xlibre.

Suddenly, over in X11 land, everything seems to be kicking off. As we reported last week, there is a fresh fork of the X.org server called Xlibre, which seems to have stirred up lots of activity.

The project was begun by developer Enrico Weigelt, who had already featured in The Register more than once before this move. About four years ago, Linus Torvalds rebuked him for spreading anti-vaxxer misinformation on the Linux Kernel Mailing List. Weigelt surfaced again a year ago with some patches to improve refresh rate handling on multihead X11 setups, and a new tool he called the "Xorg testing ground," used to build and test the X.org server in a jail. That can now be found on GitHub, as Weigelt's account on Freedesktop.org was closed.

We must confess to a mistake here. In reporting on the new fork, we referred to him as a long-time X.org maintainer. Some folks from X.org got in touch to point out that this wasn't in fact the case. He only started committing code in early 2024, and was never a project maintainer. We do suspect he's been the single most active developer in that time period, though. We would offer some evidence of this, such as a link to his commits, but along with blocking his account, as Phoronix reports, the X.org team has been busily reverting lots of his merge requests.

Despite this burst of activity, the team has put out not one but two new releases, both of the X.org X server. On Tuesday, the project released versions 21.1.17 of the X server, and 24.1.7 of Xwayland, due to multiple security vulnerabilities that were discovered back in March, confirmed in April, and fixed this month. It seems they missed one. On Wednesday, these were superseded by xorg-server version 21.1.18 and Xwayland 24.1.8.

Meanwhile, the new Xlibre project is attracting lots of interest. This month, there have already been dozens of threads on its mailing list. (Interestingly, the list was created back in February, so Weigelt has apparently been planning this for a while.)

Back then, apart from an initial test message, the only post was one in which Weigelt announced the project's Telegram group. This is called x11dev and has nearly 500 members and many thousands of threads. On GitHub, Xlibre has been starred some 2,200 times.

We can't tell if this is despite its position on diversity, equity, and inclusion (DEI) and nonexistent Code of Conduct (ENOENT is Unix-speak for "Error: no such entity") — or because of it. Either way, it's working.

Over on the social network that is now appropriately enough called X, the Devuan project has posted in support of XLibre:

It feels like anti-Wayland sentiment and resistance are growing. Whether this is despite changes such as GNOME 49, Ubuntu 25.10, and Fedora 43 removing the ability to run GNOME using X11, or because of such changes, is unclear but we suspect the latter. In that story, we linked to a detailed document maintained by Simon Peter, creator of the AppImage packaging format and the helloSystem Mac-like FreeBSD distribution. The title of his list of objections summarizes its content: "Think twice before abandoning Xorg. Wayland breaks everything!."

It's not just him. We've also previously linked to digital artist David Revoy, who last year published a long, in-depth explanation of why KDE Plasma 6 and Wayland were unusable for him. (To be fair, the newly released Plasma 6.4 fixes some of these issues.)

Many problems remain, though. Last week, the popular KiCad PCB-layout tool posted a blog post about the program's Wayland support. For Wayland fans, it will make discouraging reading. Some of the gems include:

It also describes problems with toolbars, multiple window management, dragging things between windows, and much more. It concludes:

And recommends:

Many Wayland critics point out its failings compared to the way people used X11 in the 20th century, its lack of network transparency, for instance. That's true (although Waypipe does enable it to work over a network connection), but it misses some important points. Yes, this was possible, but not essential. The Reg FOSS desk has been installing and supporting Unix boxes since 1988 and has never once needed or used X11 over a network connection.

• KDE targets Windows 10 'exiles' claiming 'your computer is toast'
• KDE Plasma 6.4 ships with major usability and Wayland improvements
• Forked-off Xlibre tells Wayland display protocol to DEI in a fire
• OpenMamba: Eat your greens, they're good for you

Secondly, it's much less true of 21st century Linux desktop computers and their modern full-color hardware-accelerated desktops. The way modern X11 desktops work, with 3D compositing using OpenGL, doesn't work over a network either.

These days, the problem is that many of the ways that traditional desktop GUIs work is also incompatible with Wayland. This grumpy old user wants to continue using his standards-based keyboard-driven user interface, complete with matching text editor. We know that many more youthful techies don't use these things, but that does not mean it's acceptable to remove them.

These sorts of features are not just minor unimportant aids for long-time users who are stuck in their ways, they are also keystone accessibility features. In a few decades, those same techies removing uncool features such as keyboard-driven menu bars will discover that they can't see sharply any more and that they keep losing the mouse pointer, or can't see the cursor, or their fingers are too stiff to perform touchscreen gestures, or their hands tremble when they're trying to tap precisely. This happens to everyone. It will happen to you. It's called aging, and it's a horrific experience, but it's better than the alternative. And worse things than just getting old happen. Any number of ailments can take your senses or your limbs. For example, diabetes can take your eyesight – but before that, the endless blood tests numb your fingertips, and the resulting scar tissue means capacitive touchscreens stop registering those fingers. You can still use a stylus, but you can only use one of those at a time, meaning no more multi-finger gestures for you.

The very important principle of Chesterton's Fence applies here. Just because you don't use something and don't need it does not mean that nobody needs it. An alternative that's fine for you may not work for others. They may depend on the thing you're removing. Hamburger menus reduce screen clutter, but a screen-reader can read textual menu bars aloud.

A hamburger menu can't be read until it's opened, and to do that, you need to find it first. Title bars are legacy UI dating back to the Apple Lisa. But some desktops, such as MATE, let you roll up windows leaving only their title bar, improving window management. Almost all X11 desktops let you middle-click the title bar to place that window behind all the others. GNOME gets rid of title bars, so neither of these works any longer. One user's clutter is another's essential UI – or worse, you can discover that what you thought was clutter was actually useful. For example, look at the modern GNOME applications that now need two separate hamburger menus, such as GNOME Disks. Which does which? Why do they look different? What do three lines mean compared to three dots? That dull, old-fashioned, single unified menu bar suddenly doesn't seem so terrible.

The Reg FOSS desk is nearing 60, and as a result is totally indifferent to most of the shiny new features that Wayland enables. He doesn't care about high definition, high dynamic range, high and variable refresh rates, tear-free video, and so on, because he is physically unable to see such things through his expensive varifocal glasses. He never uses trackpad gestures because he avoids trackpads, but he uses all seven buttons on his wired USB mouse.

We know that many more youthful techies don't use these things – and that a vocal cohort inside the community, including younger coders, is looking to improve things – but that does not mean it's acceptable to remove them.

Meanwhile, it's all happening. Wayland gains ground; xorg-server witnesses purges of code and contributors, but also new versions, and new offshoots; and that new fork wins legions of supporters, if not as many contributors. It's all go. Change and evolution are good things. History is good, too. Long-lived, often-ported code is code that has won many of the battles of natural selection. Long live X11! ®