Flakes are such an obviously good thing...but the design and development process should be better



Very well written article, informed and composed.

Here’s to hoping we ultimately form a stronger community and more robust Nix out of this.


This is a good explaination as to why Flakes are how they are. I really likes the article and it really boosted my confidence with the project which I was honestly loosing…

I think the primordial thing to revitalize the nix/os and flakes project would be to provide people with a clear and precise roadmap about the feature, define what it will and what it will not include and when certain things will arrive or stabilize.

Nix/OS always had a problem with documentation (it is there but it is very hardly discoverable) and having two extremely different systems “compete” is really suboptimal for documentation.

If you make a tutorial on something, you have the “legacy” way and the flakes way.
The problem with this is that most of the time, you take care of a certain problem as an hybrid between the two: a top level flakes to pin the nixpkgs repo you want to use, and some sub nix files that are just standard nix expressions.
This is problem as you have to learn two different systems to do one thing which is nit great in terms of the user onboarding etc…

Should we possibly make an RFC right now even if flakes are becoming somewhat stable to define the short, medium and long term vision of flakes or nix in general?


To be honest the flakes way is sometimes unusable, and it gets wrapped in the legacy way. My personal files repo is a clear example of this :slight_smile:

1 Like

I appreciated this article as well.

The article gave me helpful context on how difficult change is in our distributed Internet reality!
For outsiders, I asked why / what flakes are around that time: https://discourse.nixos.org/t/what-are-nix-flakes-and-why-should-i-care