New blog post: towards a simple continuous delivery mechanism for #NixOS #Nix https://leftfold.tech/posts/nixos-continuous-delivery
-
New blog post: towards a simple continuous delivery mechanism for #NixOS #Nix
https://leftfold.tech/posts/nixos-continuous-delivery -
New blog post: towards a simple continuous delivery mechanism for #NixOS #Nix
https://leftfold.tech/posts/nixos-continuous-delivery@johnny I think that for #NixOS we need an additional concept of splitting the "system" and the "application".
I did use nixos containers in the past, all in the big system configuration, this is inflexible, as you describe in your post. And I think your approach would have solved this.
Seems to me that there is this more generic need behind this: Handle the core system on the one end, and the application on the other end. And being able to update each of them individually.
-
@johnny I think that for #NixOS we need an additional concept of splitting the "system" and the "application".
I did use nixos containers in the past, all in the big system configuration, this is inflexible, as you describe in your post. And I think your approach would have solved this.
Seems to me that there is this more generic need behind this: Handle the core system on the one end, and the application on the other end. And being able to update each of them individually.
-
@novet Flakes are a mechanism for distributing Nix files, but you can't reference them without a revision lock or update them without rebuilding the system that includes them.
I think what @johbo said is about right – when we want a live deployment of an application, we don't think about that on the "core system" level, we ideally want to have a separate layer just for "end-user" applications that can be more dynamic. But with NixOS that isn't really possible.
