I’m pretty happy with the top-level goal of doubling the number of active contributors. The Fedora user base is growing, and that means we can grow the project itself (more users means a greater pool of people willing pitch in) and that we need to (because otherwise we won’t be able to keep up with the demand). I think it’s achievable, too — and the things we’ll need to work on (mentorship, infrastructure, community tooling, ease of use for packagers, new ways of distributing software within the Fedora umbrella…) will also benefit all of us existing contributors.
I’d also like to talk about secondary goals — other things that aren’t under that focused umbrella that nontheless are important for our strategy. We should also discuss non-goals — things we’re intentionally not focusing on (or even leaving behind). So, this post is one of several promised follow-ups along that line. Specifically, let’s consider the idea of Improving Fedora Project and Fedora Linux mindshare in the Linux ecosystem.
Of course, some of this is a natural requirement for achieving the “double contributors” goal. But, how much do we want it to be a goal in itself? What areas should we focus on, and why?
I can see many places where we’re posed for growth, if we invest in it:
Fedora Server and Fedora Cloud working groups have, in the past couple of years, renewed community energy. With the changes over in CentOS, people are looking at different options in the free/DIY Linux server space.
This has several dimensions: of course, people looking to actually run workloads (in dev, test, or production!); but also, there’s a direct connection to future CentOS Stream and to RHEL Server, and soon to Amazon Linux as well. While those downstreams allow some involvement, this is the place to get involved for big changes and really influencing the future.
Fedora Workstation… I almost don’t have to say anything here, because I think the success is self-evident, as is the potential for more. Lots of great execution on new and improving desktop technologies is appealing to the enthusiast crowd.
And, Fedora Silverblue. The absolute numbers are still low, but the buzz and interest continue to be high. It’s our very own “I use Arch, btw.” So, the question is: can we turn that into a mainstream offering? And… should we?
Fedora CoreOS… it’s been a long road, but I think we actually have something interesting and special for container-focused infrastructures. We haven’t really done a big push to reach the people for whom the original CoreOS was interesting… but we could.
Fedora IoT: really nice technology sharing a lot with CoreOS and Silverblue but with unique features as well. I think we’ve really been hampered by lack of reliably-useful just-works hardware. But… maybe things will get better on that front with the next-gen Jetson hardware and ARM’s SystemReady IR. And we have interesting things going on with robotics (ROS) and… I personally am pretty excited about Fedora IoT with Home Assistant.
Other desktop spins — I know many of us love them! As I was putting together my 35 Fedora Release in 35 minutes talk for Open Source Summit, I was strongly reminded that “spins” and “remixes” are deep in our DNA. There’s a reason we encoded this idea in the mission statement, after all. I think we have a lot of room to build, here — and in particular, room to help others with interesting ideas try them out. I’d love for desktop projects which want to put together their own complete showcase to have Fedora be the obvious place.
Packaging, containers, and software distribution. Of course, “packaging up stuff and making it easily available to users” is core to what we do. But, in the modern computing world, distros aren’t the center. If we want that aspect to continue, we’ll need to adjust so we’re adding clear value. It’s worth considering if that’s even worthwhile. (I think it is, but… it will take some changes.)
Academia and student groups — we’ve talked about this for a long time… it seems like a natural area for, well, all of the above to take hold. We’ve tried to run some objectives around the idea, but none have really struck the mark. Should we make that part of this effort?
Leadership in community health — we’re far from perfect, but we’ve been working hard with our refreshed Code of Conduct, active DEI team, and increasing support for mentorship. Fedora has shown leadership in the workings of open source projects… how can we build on that?
Around the world in different languages — our translations teams have long been a Fedora strength. Building on that might be a goal in itself (there’s a separate topic coming about accessibility and inclusion, where this definitely relates), but … maybe it’s also a general mindshare area?
All of that is a lot – probably too much to constitute a focus. Are there ways we can bring it all together? Are there things we should decide to leave out? And, of course, there definitely are things which I haven’t mentioned — which of those should we make sure to include? What should we target, and how will we know when we’re succeeding?
I think academia and student groups are a great focus. I’m using Fedora (Silverblue) for my CS classwork right now, and between Flatpak IDEs and Toolbx, it’s easier than ever to set up and maintain development environments.
Along these lines, TeX is something I’m hoping to explore, and that is a natural for academic research.
I’m more than a little biased here, but absolutely there is opportunity for growth in academia and student groups. One possibility could further partnership with Jupyter. We have a good start Jupyter and data science in Fedora - Fedora Magazine and our Scientific spin. JupyterHub container images are currently based on Ubuntu and for many students, this is their first exposure to Linux. Having a Fedora based offering could be useful. I would also be willing to help get that going, if it’s of interest to others.
To bring it all together, probably divide and conquer will be necessary to prioritize big hitters by user persona - existing user base and environment of Workstation and server.
I’m not sure what level of granularity we have on Fedora Linux user data, so I would take an example of R&D where groups of users have their OS build for product testing, porting, and kernel testing recommended by the IT team. For new developers joining the dev team, the first impression will be what Linux build they have.
Our strategy in this respect will be how to increase adoption for industrial use cases and find a light house. I took an example of product development as one I can see where I work (Samsung).
There are many SmartTV, industrial drone brands, and Telco that have certain preferences for Linux OS in testing and development. We’re talking about circa over 50K engineers per large enterprise in consumer electronics and telco.
This is a big one for me. The work we’ve done to collaborate with other distro communities, as well as upstreams, in the past year or so is important. One, it allows us to connect with contributors who might want to work across multiple distros (for example, a current FESCo member is also on the openSUSE board). Two it helps our reputation among an audience already pre-disposed to Linux. We shouldn’t think of it as “let’s take users from other distros”, because that’s not what I’m saying. It’s more of a “let’s make ourselves an option in the minds of people who are currently in the Linux ecosystem, but not in Fedora”.
Secondarily, but related, our work with Amazon—and their work with us!—have the potential to really cement our role as a leader in the Linux distribution space. Silverblue fits that as well. I’d love to see that become our default workstation offering in the next few years. It’s a paradigm shift for a lot of people, but with great benefit.
I’ll stoke the fires of the BIOS thread and say that legacy hardware support should be an explicit non-goal. I don’t want to drop support for otherwise functional hardware just because, but when it holds us back (either due to lack of support for new features or lack of time from maintainers), we should admit that it’s okay to drop it. I think it’s great that some distributions maintain support for very old hardware. I don’t think that every distro needs to, and in particular not Fedora.
Indeed, it’s related. The environments I maintain are used for social science as well as “harder” sciences and share a common software base between them.
It’s it’s helpful, here’s one of the public, open source’d container bases for Jupyter that I’ve been working on. Jupyter upstream uses Ubuntu, and there are a number of reasons why basing on Fedora might be advantageous, especially since Fedora packages RStudio and RStudio-server so these don’t need added separately, as they are in this repo base. Also, maintaining R compile environments can be a big pain. I’m using Jenkins to do continuous integration/testing on these and downstream images, but it would definitely be great if we could get this stuff packaged and regularly maintained in Fedora and offer Fedora based images for these kind of environments.