We took our first impressions of Fedora on a bare metal install on the HP Dragonfly Elite G1, seen here on a Moft Z folding laptop stand.

Enlarge / We took our first impressions of Fedora on a bare metal install on the HP Dragonfly Elite G1, seen here on a Moft Z folding laptop stand.

Jim Salter

Today’s Linux distro review is one I’ve been wanting to do for quite some time—Fedora Workstation. Fedora is one of the heavyweight desktop distros of the Linux world, with a vibrant community and a strong presence at every open source convention I’ve ever attended. (Remember physically attending events? Ars remembers.)

I never felt particularly drawn to Fedora myself, because it’s a bleeding-edge distro—one targeted to the very newest software, possibly at the expense of stability. That’s not what I personally want in an operating system—I fix broken things professionally; I’d prefer not to fix them personally any more than I have to.

But as one of the few distros using the next-generation Wayland display server by default, Fedora made me very curious indeed. Although the screenshots taken throughout this review are of virtual machines, my first installation of Fedora Workstation (ever!) was bare metal, on the HP Dragonfly Elite G1.

A chat with Fedora project leader Matthew Miller

Before getting started with Fedora, I put out some feelers on Twitter to see if I could get a project leader to answer some questions about the distro. Matthew Miller was gracious enough to respond at some length. (The interview has been lightly edited for length and readability.)

Ars: I’m particularly interested in—and clueless about!—Wayland. I’ve been hearing it’s the super amazing next generation of window manager for years, but I’ve never actually used it. What should I look for to see it to my best advantage?

Miller: OK, so, I’m going to have to get a little geeky on this one. But first let me give the perhaps-surprising answer for most people: when Wayland is working as it should, you won’t even notice it. Nothing flashy or new or amazing at all—just better performance and a smoother experience.

So, Wayland is simply a… more lightweight design [display server] that connects applications more directly with the hardware. It can also better take advantage of modern graphics cards, allowing for flicker-free animation and tear-free video. And, it allows applications to run on the same display but be isolated from each other—the X11 design actually makes it so any running application can snoop on key-presses intended for another application. This makes it safer to run sandboxed applications that might be from a less-trusted source or at least to contain a compromise from spreading to another application.

Additionally, since Wayland is where the development energy is, new features like hardware acceleration for Firefox and improvements to high DPI rendering are only available there. It’s not that these couldn’t theoretically be done in X11 (with more effort), it’s just that no one is doing that work.

All this means Wayland is definitely the way forward, and we made Wayland the default four years ago in Fedora 25. But really, the improvements are all under the hood. You definitely see enthusiasts for whom this is a big deal, but that’s like car enthusiasts talking about a new engine design. For most users, you’ll see the equivalent of better gas mileage and smoother performance, but it won’t be something you actually directly care about—and that’s a good thing.

Ars: I’d love to hear you talk about how Fedora sits in the Red Hat and overall Linux ecosystem—do you have a direct upstream?

Miller: Fedora is one of those few distributions which is at the head of a river, to stretch the upstream/downstream analogy. In fact, I’m going to go ahead and keep stretching—the spring at the start of a river doesn’t create water; that water comes from rains and snow melt and so on.

Likewise, most of the software that goes in the Fedora distribution isn’t written by the Fedora Project, but rather in thousands of other projects like the Linux kernel, systemd, the Wayland graphics stack, Firefox, the GCC compiler, and so on. Those are our upstreams, and we collect and integrate them into the flow that is the Fedora distribution.

Ars: How direct is the downstream—if that’s an accurate description—to CentOS and Red Hat Enterprise Linux?

Miller: Very direct! Every few years—now officially three—Red Hat takes a release of the Fedora OS and branches it into an alpha version of Red Hat Enterprise Linux. We’re actually working to make this even more transparent.

In parallel with the in-progress Fedora 33 release, we have a build environment that mimics that used for RHEL (see https://fedoraproject.org/wiki/Changes/ELN_Buildroot_and_Compose), and the goal is for that to eventually flow smoothly into CentOS Stream, and that branch of CentOS Stream will then become RHEL 9.0.

CentOS Stream as a mid-stage is new. Traditional CentOS Linux used to be simply a rebuild of the sources of RHEL, and very far downstream. So we’re looking forward to this new closer relationship to Fedora as it develops!

Ars: Can you tell me about best practices for end user maintenance of a Fedora system?

Miller: I guess I have two answers for this. First, if you’re someone who likes to tinker with your system, feel free to do so. You may break things, but that’s how you learn, and that’s why we have a great community who will help you in that journey. Bring your problems to https://ask.fedoraproject.org!

Second, though, if you just want to use your system… no problem. Fedora’s various options like Fedora Workstation are meant for end users, and deep maintenance isn’t necessary. Make sure to apply updates on a regular basis and you should have many years of hassle-free Linux.

Ars: Does that change any with long-term use? What about upgrades?

Miller: A few years ago, I frequently saw people ask for us to make Fedora either have a long-term support edition or switch to a rolling-release model—where updates flow continuously, with no whole-version upgrades.

This struck me as curious, because these are nearly opposite approaches. When I dug into it, the real concern is generally that people are afraid of upgrades: they seem like a lot of work with a lot of potential for disruption.

So, we’ve tried to address that directly. Long-term support is very expensive to implement—and Fedora is primarily volunteers who have limited time.

A rolling release sounds good, but Linux distributions integrate so much upstream software that it’s inevitable that eventually some big change will affect you. So a rolling release really means that those upgrade surprises can come at any time, and we don’t want to surprise our users in a negative way.

We decided to keep to our six-month release schedule and focus on making updates smooth. This focus has worked out: I upgraded my main laptop from Fedora Workstation 31 to 32 while I had lunch, and I’ve seen many people reporting similar experiences. It’s basically: push the button, come back to an updated system.

We do maintain an overlap between supported releases, so if you’re not ready for change, you can skip a release and upgrade once a year. But generally, we’ve worked to make it be no big deal to stay up to date—while leaving in your hands exactly when you want to make the update.

Ars: Can you tell me about the distro’s overall project goals and/or mission statement?

Miller: I hope you find our Mission and Foundations page useful on this. Fedora is a community project, and our goal is to enable interested teams to provide specific Linux distro solutions for areas they care about.

The most popular of these is Fedora Workstation, our main desktop offering—the team working on this wants to build a polished and well-engineered desktop operating system that brings the power of open source to anyone who wants it.

We also have Fedora CoreOS, for next-generation compute clusters and container-based workflows in the server room. And Fedora IoT takes some of the technology developed for CoreOS and brings it to edge computing and smaller ARM devices that might power your home automation or data-gathering and processing in the field.

But we also encourage much more niche uses really tailored for specific cases. For example, new with Fedora 32 we have a Computational Neuroscience Lab. It turns out that a lot of volunteers in Fedora work in this field, and they decided to work together to make a collection of the important open source software they use for their work so that other scientists can get productive quickly. Fedora Jam is the exact same thing—but for an audience of musicians and audio enthusiasts.

I’m obviously very happy to see big successes for Fedora Workstation, Fedora CoreOS, and Fedora IoT—the recent Lenovo announcement is huge!—but ultimately I measure success for Fedora as a project in how well we do at fostering these various communities of interest and how we build a collaborative environment, so each team can grow their own user communities as part of Fedora as a whole.

Read More