DEV Community

Discussion on: What would you like to see in a developer-centric Linux distro?

Collapse
 
yannik_sc profile image
Yannik_Sc

My opinion to this:

  1. It should contain the minimum required to run it. I don't want runtimes and tools on my system that I will never use, so keep it slim.
  2. Ubuntu as a base? Are you serious? I don't want to setup the again for each bigger update. Ubuntu is known to destroy itself for bigger updates so a clear No here as well
  3. The DE: KDE is fine but I still think that most developer will, at some point, switch to some kind of a a tiling WM, so using KDE makes less sense then giving the user the choice to select their own DE
  4. VMs: this would mean pre installed VirtualBox? When there are no VMs you don't need to install it, as said, keep it slim, I can install it by my own if I need it
  5. Not mentioned above but still a decision to make: X12 or Wayland? My personal opinion: X11. Wayland is already almost 10 years old, and it's still lacking tons of features and it has many (tiny) issues when you get into some things

In the end I don't think that it makes too much sense as all developers have their own preferences. At least not in this fixed architecture where you have a distro that comes with ... a better approach would be to use just a custom installer with selections, and make it possible to save and restore the selection.
And if it's meant to survive updates, keeping it up to date and keeping the setup for a long time, don't use Ubuntu. It will always crash. Use Fedora as base or something. It's much more stable.

Collapse
 
stereoplegic profile image
Mike Bybee

I appreciate the input, but I also need this to be clear: Above all else, the primary stakeholder will always be me. If you like what I put out there, great, and I welcome suggestions. If you can't get past my decisions, you're free to move on, taking anything you do find useful from my implementation with you.

  1. This will not ever be a minimal distro, but it's also not going to be Ubuntu Studio (at least not out of the box, though I install most of its and KXStudio's packages for my own use on top of my dev stuff).
  2. Dead serious. I'll take an APT collision over an RPM collision any day. The only times in 12 years of Ubuntu (with plenty of other distros as well for the past 17) that I've seen a borked Ubuntu upgrade was from bad use of bad 3rd party repos (or some bad Ubuntu-based distro - Linux Mint included - who didn't properly handle its own repos) or some dummy SysAdmin not understanding breaking changes (e.g. I saw lots of easily fixable apache 2.2-2.4 upgrades). In Fedora and RHEL (and Suse/OpenSuse, and...) on the other hand, I've seen nightmarish failures even with zero third party repos installed. And again, that's not even my primary driver (hardware support). I've seen plenty of models Fedora wouldn't even boot on, let alone support things like WiFi.
  3. Again, I'm the primary stakeholder, and I despise tiling; however, as with most things modern KDE, it's as easy to add as "Get New [Thing]" (with multiple implementations), without giving up the massive usability of a modern and über-configurable UI. And again, others are free to base what I release on some other DE/WM, whether they want to call it a "community spin" or just rebrand completely.
  4. I mean KVM, which is how you really should be running VMs in modern Linux; however, VBox repo will be added by default (you'll have to install it and the addons yourself). For Linux guests, in addition to those options, Docker will be installed and LXC/LXD will be installable.
  5. I agree wholeheartedly on X11. I wish it had some of the benefits of Wayland when it works, but Wayland still has a lot more issues than mere papercuts.