I’ve read an article today, in ZDnet, mourning how desktop distributions seem to wax and wane. It really made me think about how to properly convey what I feel to be the root issues with desktop adoption of Linux (and the wider CS industry), and why I think most people are very, very wrong about it.
Most people, normal people, they do not care about shiny. They don’t care if their email client or word processor is written in C++, Rust, JavaScript, Python, Ruby, or FORTRAN. They just want something that is simple to use and allows them to accomplish their tasks. It feels like most programmers and designers have lost that concept. I’ve met people in this industry who thought that they could sell people their products based on metrics like memory usage and number of packets transferred. Very excited people tell me how they use MapReduce and AngularJS and MongoDB and MariaDB (sometimes all four at once), like they deserve an award for Most Use of All Available Frameworks And Patterns Released This Decade. There’s a great article I read recently that puts in no uncertain terms: You are not Google. Even Google isn’t Google. Following these “trends” may make your product seem “trendy” to other developers, but all you are doing is wasting your time while leaving your users under-satisfied. Focus more on your software and how it can help people, and less on how it can help your résumé.
Meanwhile, I’ve seen libre software developers put up on their homepages that they now offer Flatpak images and are very excited to tell you that. Nobody knows what that means outside of developer circles! Do you think your grandparents care that your word processor is available in Flatpak or Snap? Do they even know what that means? In addition to this, Flatpak and Snap are anti-solutions. They don’t solve any real problems; they only serve to create new ones. What if a user or distribution wants to use ALSA instead of PulseAudio? What if a user or distribution uses an alternative to systemd? What if a user wants to use sound from their existing session at all? What about CPU architectures that aren’t supported by all developers, like ARM, POWER, or RISC-V? Focus more on your software and how it can help people, and less on following the trendy new shiny.
My personal hope for this industry is that we gain stronger distributions with better focus, and that developers leave packaging to the distribution packagers. This would allow us to have a few different desktop distributions. Distributions could be working together on a common frameworks, like KDE, while having specific plans and goals in mind. Gentoo and Arch for the tinkerers; Gentoo with more focus on source building, Arch with more focus on a tightly integrated system. Adélie and Fedora for the general public; Adélie with more focus on stability and portability (X11, PPC/ARM), Fedora with more focus on incubating new systems. Obviously there would be distributions that would pop up with unique goals; their work could be integrated later in to a main distribution that is aligned with it.
I feel like this could really be a way forward towards more wide adoption of Linux, on the desktop and beyond. I hope we can work together towards it.
(As a small aside that I wanted to note: I hope “The Year of the Linux Desktop” never comes. I hope that we usher in an Era of the Linux Desktop.)
The Cat Fox Life 