Maybe It Should’ve Been Mir
The Potentials
It’s no secret that X11 is decades past the point where it should have been replaced. Quite frankly, current technology now has extremely little in common with the technology that the X protocol was created for. However, the fact of the matter is that an extremely large ecosystem has grown around X11, so replacing it with something new is by no means a trivial matter. Despite this fact, there were a quite a few interesting projects that could have been successors to X11 as the default *nix desktop platform.
There’s no rule that specifically says that the replacement for a specific system has to be something brand new. There were quite a few stable, actively developed projects that could have been replacements for X11, but the two that I found most interesting are DirectFB1 and Nano-X. Both of these projects actually targeted the embedded computer market, but operated just fine on the desktop. Though they both provided similar functionality, each took a decidedly different approach from the other. DirectFB targeted the host framebuffer and provided a very thin interface on top of it. On the other hand, Nano-X took the approach of providing an extremely minimal implementation of the X11 API without using the X11 protocol underneath. Instead, Nano-X provides it’s own API for the specific purpose of using other APIs on top of it -user programs have no access to the Nano-X API.
Just because potentially suitable replacements for X11 exist does NOT mean that anyone should automatically abandon the idea of creating something new. Accordingly, many ideas were thrown around, but three projects piqued my interest after they reached the point of producing working code – Mark Thomas’s Y Windows project, Canonical’s Mir2 display server, and Xorg’s Wayland project.
Two Different Approaches
Clearly, there were enough options available to use as starting points on the path to a display system to replace X11. Or there seemed to be. When you take an objective look at the situation, you’ll see that it wasn’t really realistic to expect the embedded space solutions to be used as foundations for a new desktop GUI. In my opinion, the biggest issue is that so much additional development would have to be put into such an project that the entire ordeal would likely be similar to starting from scratch, while limiting the project to the constraints of the embedded world. So, DirectFB and Nano-X are instantly disqualified. Y Windows was Mark Thomas’s master thesis project. A small development team tried to develop the project further, but it eventually fizzled out without gaining much traction. That left us with only two serious contenders in the form of Mir and Wayland.
This is the point where working with open source while trying to actually achieve real progress gets tricky. There are two main ways to develop open source software, each with their own pros and cons. On one hand, you can build the project in-house with minimal to zero visibility into the project and release it when it’s working. On the other hand, you can try to build the project with the community having varying levels of involvement, but full transparency. For the most part, Canonical took the former approach with Mir, while Wayland was developed with the latter approach. What’s tricky about this scenario? Well, most developers know that design by committee is usually an unproductive waste of time which tends to result in a product that tends to be less than stellar.
For better or worse, the open source community rallied behind Wayland. It’s been in development for over a decade, but it’s still not a viable replacement for X11. Within that period of time, Mir has retained some form of relevance by becoming a client of Wayland. Many other sites are more than willing to debate the pros and cons of each display system. However, the primary metric that actually matters is the level of functionality that each tech package brings with it. We didn’t get to see what Mir would’ve been, but we can absolutely see what Wayland is. For about two or more decades, I’ve witness various proclamations that each year was the year of the Linux desktop. While such proclamations primarily annoyed me with their blatant disregard of the rest of the *nix community, a not so trivial level of annoyance also stemmed from the fact that the proclamation was a lie…every…single…year. Not only that, there’s also the fact of just how incredibly far away from reality those proclamations were. Currently, we’re half way through the year 2025. I can’t help wondering if the situation would have been substantially better if the community would have chosen to rally behind Mir instead…
Conclusion
Maybe it would’ve. Maybe not. But what’s clear is that Mir at least tried to be a complete solution—top to bottom—while Wayland continues to feel like an incomplete framework still waiting on someone else to build the rest. It’s not that I think Mir was perfect, but it was opinionated in ways that could’ve led to a coherent desktop stack. Instead, we’ve ended up with a fragmented ecosystem that seems perpetually stuck in a holding pattern. Wayland might technically be “the future,” but I’d argue it’s been “the future” for far too long. At this point, maybe the real question isn’t whether Mir should’ve been the answer—but whether anyone’s still willing to ask the right questions.
1 – A seemingly no longer active fork of the project exists here.
2 – Mir was eventually changed to become dependent on Wayland.