Why I developed my own Ubuntu distribution Starbuntu

by Peter Starfinger, Munich, September 2022

My first experiences with Linux (then Knoppix with KDE 3.5.9) are now almost 20 years ago, and after a six-year test phase parallel to Microsoft Windows, I decided in 2009 to only use Linux , specifically Ubuntu. In the years that followed I tried out a few variants: Kubuntu with KDE 4, Lubuntu, Ubuntu Mate.

Even though I have never regretted my switch from Windows to Linux, but appreciated its stability, open source and its invitation to digital self-determination and personal responsibility from the very beginning, there were always smaller or larger occasions telling me about badly or sloppily programmed software to annoy. Sometimes important features were missing, sometimes I stumbled over stubborn program errors that hadn't been corrected even years later, sometimes I had to realize with disappointment that a certain piece of software I valued was no longer developed at all. For example, if the file manager actually fully supports Plug & Play, but in reality often does not recognize the inserted USB stick or incomprehensibly only displays its content after entering the user password, frustration arises. Or if the taskbar is unstable and sometimes disappears unexpectedly or suddenly loses its settings so that you can reconfigure everything again, that doesn't add to the joy. Of course you hope for the next upgrade, but often in vain. It amazes me how persistently some bugs are allowed to annoy people for years. One wonders what some software developers actually do.

One more thing struck me more and more, namely the increasing self-importance of the Ubuntu developers, which concerns on the one hand the silent mothballing of valued software and on the other hand the installation of superfluous, resource-wasting software. Why do they ban, e.g., supposedly for security reasons, the proven gksu package as a practical GUI for the terminal command sudo and instead oblige the user to use the much more cumbersome command pkexec because it is addicted to configuration. Or why do they suddenly want to force the snap container software ideology on the user with all its disadvantages in terms of runtime, space requirements and security? Not to mention the tiresome Unity dictatorship!

Against this background, I decided in autumn 2019 to create my own distribution based solely on the Ubuntu core, namely Starbuntu to develop. On the one hand, pure curiosity drove me, and the challenge appealed to me. On the other hand, I wanted to avoid the undesirable developments mentioned, to be as conscientious as possible when selecting available software, and to write my own programs where the software didn't convince me or wasn't available at all. My main motivation was and is related to myself and not to the community: finally no longer having to make rotten compromises, but simply implementing what I want from a good desktop environment myself. You can only do that in your own workshop, not on someone else's product!

After setting up the X Window System (at that time there was still no alternative) the first problem arose: which window manager to use? I had tried many, conservative as well as exotic, with or without frills (e.g. dockable, wobbly or exploding and imploding windows), more or less stable. In the end I decided on OpenBox, the world champion in the disciplines of simplicity and stability, because I didn't want a bird of paradise, I wanted reliability!

The next cardinal question was choosing a good file manager. The decisive factor here was the leeway I had to be able to "pimp" it according to my ideas. After a long search, I chose the ROX-Filer, a wonderful, lightweight, super-fast file manager that can be expanded almost at will according to one's own taste, which initially requires a lot of configuration work, but later pays for it twice or three times over. In my opinion, the highly bred competitors do not offer great modulation capability. On top of that, the ROX-Filer also comes with a configurable desktop management.

Most of the taskbars (panels) are offered in combination with a whole desktop environment: the MATE panel, the LXDE panel, the XFCE panel, etc. So they were out of the question for me because I wanted to be independent when choosing a taskbar. Here, too, I was guided by the two criteria of simplicity and stability. Of course, it should also look appealing, but without superfluous features like a 3D design or jumping icons that react to the mouse pointer. So I finally decided on Tint2, an absolutely error-free and reliable, but also appealing bar, and have never regretted this choice.

Some decisions for a suitable software were almost without alternative, e.g. the Libreoffice package or the Calibre ebook manager, others were based on personal experience or preference and are certainly debatable. An overview of Starbuntu's on-board software (GUI) can be found here.

Now, a self-developed Linux distribution would not necessarily have a right to exist if it were limited to a more or less random compilation of well-known software, perhaps decorated with a nice design. It would lack its own "farmyard smell", the unique selling point which is not only found in an individual design (themes, background images, etc.) but above all in a self-written software that fills gaps.

The need for such software became apparent to me in many places, some of which are examples:

I think it was only through this and many other programming tasks that Starbuntu matured into a distribution in which simplicity, clarity, lucidity, practicability, but also beauty are realized.

Note: I would never have been able to implement the many GUI dialogs in my programs without Victor Ananjevsky's excellent GTK+ dialog program yad (yet another dialog). My eternal thanks go to him for that.

Peter Starfinger