New distros get a lot of crap, but often they are solving a need for someone.
Take Windowmaker Live: ostensibly it’s just Debian + Windowmaker. I have seen comments saying why not just install WM on Debian? By asking that question, it’s clear the asker hasnt tried recently. There is a lot to configure, and there are lots of usability papercuts.
A custom distro allows someone to fix those problems for themself, and share those fixes with others. It’s not fragmentation, it’s just FOSS.
It would help more people to improve the installer for difficult-to-install software rather than creating an entire operating system around that software. Using the entire operating system as an installer is over the top
If you improve the installer to the point it can install any combination of software together (including incompatible versions of deps) you end up with NixOS again
Don’t know the case for this - but there are absolutely cases where the merger is blocked for some reason, and why not just fix it yourself with a distro? It hurts nobody.
Yeah I’m reading what they said and that kind of solution wouldn’t be acceptable in any industry…
Imagine if you wanted to add AC to your central heating system and they told you they need to add a second furnace in parallel to the one you already have because it’s possible to add AC to your current setup, but it’s very complicated to do so…
New distros get a lot of crap, but often they are solving a need for someone.
Take Windowmaker Live: ostensibly it’s just Debian + Windowmaker. I have seen comments saying why not just install WM on Debian? By asking that question, it’s clear the asker hasnt tried recently. There is a lot to configure, and there are lots of usability papercuts.
A custom distro allows someone to fix those problems for themself, and share those fixes with others. It’s not fragmentation, it’s just FOSS.
It would help more people to improve the installer for difficult-to-install software rather than creating an entire operating system around that software. Using the entire operating system as an installer is over the top
You’re both right
Maybe a script or deployment tool config would suffice
You’re re-inventing the Nix tool which is exactly a script that sets up all the programs and services you want to install
Except no one really uses Nix outside of Nix OS. It is slow and complicated for little reason.
Just use Ansible and an answers file
Then how come we have more packages than the AUR?
And don’t say it’s because we packaged Python and Haskell stuff since we have more non-unique packages too
Maybe one you could just download and pipe to bash. /s
ansible-pull and kickstart was what I had in mind
I figured, I was playing on the number of github repos with instructions for curl pipe bash combos.
If you improve the installer to the point it can install any combination of software together (including incompatible versions of deps) you end up with NixOS again
Much like how crustaceans have repeatedly evolved into pseudo-crabs, Linux distros tend to evolve into pseudo-nix
Don’t know the case for this - but there are absolutely cases where the merger is blocked for some reason, and why not just fix it yourself with a distro? It hurts nobody.
Yeah I’m reading what they said and that kind of solution wouldn’t be acceptable in any industry…
Imagine if you wanted to add AC to your central heating system and they told you they need to add a second furnace in parallel to the one you already have because it’s possible to add AC to your current setup, but it’s very complicated to do so…
+1, kubuntu was a lot simpler than installing KDE on Ubuntu and still having grub applications