So I’m curious. I read in Arch is for experienced users topic that one should choose a distro on the basis of what you’re going to use it for. I can only think of one I would recommend (and that’s tentatively), which is Ubuntu Studio if and only if you are into music production. But it’s just a configuration of Ubuntu.
In all the distros I’ve ever used I’ve only used them for some programming, some GIS work (QGIS and Grass), LaTeX, git, playing a synthesiser or two from my midi keyboard, zoom, browsing the web, shopping.
At work I needed a real-time Linux for a robot control system, which allowed me to choose which cores to run my code on exclusively i.e. deterministic real-time. So I suppose that’s an example.
Am I missing out? Does it matter which Linux I use? (And if it matters, does it matter that it matters? – Douglas Adams)
For desktop use, almost every mainstream distro will run what you listed. The differences are mainly: package freshness, release cadence, and how much you want to babysit it.
None of that is about what you use it for, it’s about how you like to maintain the system.
You picked the distro for robot control so yes there are such cases.
For security, pentest and network admin work you technically can build a Kali equivalent on Arch or Debian, but you’d spend a weekend doing what Kali ships preconfigured. Same logic for REMnux on the malware analysis side.
Servers with a hard support contract. RHEL or Ubuntu LTS, because someone is paying for the SLA, not because the kernel is different.
Anything where the hardware vendor only ships one set of drivers. NVIDIA Jetson, certain industrial boards, some SBCs. The distro is whatever the BSP supports.
Older or very low-RAM hardware. Not really a “use case” but a constraint that narrows choices.
So I’d say for 90% of what most of us do, distro choice is aesthetic and habitual. Whether that matters or that it doesn’t matter, I’ll leave to Adams.
I’d add to the differences above, a distribution’s stance on proprietary licenses and software. Some distributions (e.g., Debian, Fedora) draw a harder line inclusion or not of non-free software. There’s always ways around it of course, but using non-free software such as drivers is easier on distros such as Ubuntu or Linux Mint.
That is question of means and purposes. It matters when certain Linux distribution fits one’s purposes best. It does not matter until one is subject to a corporate policy or other kind of enforced discipline.
HOWEVER … that statement is incomplete and therefore misleading!
You must
first choose what you plan to do,
identify the tools with preferred GUI interractions,
and only then
choose the Distro(s) on which those tools are verified functional in the way you want them to be!
Otherwise, the consequence is likely Distro-hopping … in search of things working the way you want!
The specific domain of “Real-time Linux” is one where I question the use of the word “Linux” itself, since the Official Kernel is not coded in a structure dictated by real-time requirements.
For real-time, you quite literally need to “strip out” and “replace” key process structures to correctly work as real-time, which is scanning multiple input channels, simultaneously, transforming/processing the values simultaneously, then acting based on priority in a defined-sequence within an absolute-limit timed cycle.
That is not how “standard Linux” is coded or operates.
If you don’t try different distros you will never know whether they suit your needs.
I disagree. At least in the last 5-10 years or so, the capabilities of different distros have converged, even if the look-and-feel hasn’t. Provided the distro has friendly support, I think a newcomer could choose any they want. For a long while at work, I used Scientific Linux (CERN variant of Red Hat) and following that CentOS before it got discarded. With those I found support was good but very formal. I used CentOS as a central server for postgreSQL. So some Linux’s were (still are?) very business oriented without so much going on as the latest popular distros.
