@aral Honestly, I really dislike their approach, as far as I understand it? It's like it's trying to do *kind of* the same thing as Nix/NixOS are, but instead of providing a robust foundation with building blocks for everybody to use in novel ways (or hell, even building on Nix directly), they implement a really monolithic model with vendored dependencies and "the OS" as a chunk of the system with a privileged status?
Like, I can understand those sort of design choices being made in a corporate environment where there's one authoritative distributor of The System, but it just feels like it's throwing away so much potential in terms of 'democratizing' system management - like, it's trying to treat the symptoms, not the systemic problem underlying them.
@joepie91 Well, this is a distribution by Red Hat aka IBM, so that’s not too surprising.
That said (and my misgivings about IBM are in the public record), it seems like a “works out of the box”, solid, and blazing fast distribution so far.
But yes, your points are entirely valid and it’s a good thing that Nix/NixOS exist for those who want to do those things.
@aral That's the thing, though - the value in Nix addressing the problem systemically isn't really for those who use it *directly* (in fact, the current UX is pretty terrible!), but rather for those downstream from people building on top of it, IMO.
That's why it saddens me that Red Hat didn't base their system on Nix - if they had, then they could *still* have produced a solid works-out-of-the-box distribution, but without the huge cliff that now exists where as soon as you want to stray outside of Red Hat-approved territory, you pretty much have to distro-hop to NixOS if you want to retain the stability.
Whereas on a Nix-based system (whether directly or even just implementing the same basic model!) it'd just be a matter of "using the same system primitives that the distro vendor used, for my own thing, and assuming that that'll work correctly".
@joepie91 Ah, right, gotcha.
@aral Okay, second attempt at expressing what I mean, more succinctly this time:
It's like the design objective behind the Nix model is "building a more robust foundation for users to work with their system in the way that works best for them", whereas the design objective behind the Fedora model is "releasing a more reliable product for people to consume".