You don’t have to release anything. Most of my flakes are on private storage in my homelab, including my homelab configuration, and I don’t feel any obligation to contribute anything upstream right now.

that is true! there’s a ton of Nix stuff I’m keeping entirely personal now too. my projects fall into roughly three categories:

• awful.systems infrastructure and its offshoots. this is the stuff that really needs to be released, because the administration of this instance should be as public and replicable as possible.
• projects that use Nix as a build and environment setup tool. these should see the light of day too since they have value outside of Nix, but unfortunately most of them don’t build without Nix, and (because Nix is technically excellent) there’s usually no easy replacement for it. this particularly includes some of my hardware projects: I use Nix to keep modified embedded firmware, cross compilation tools, and userland images in sync, and it fulfills a similar role (from the ground up) for my lambda calculus reducer project, but there it manages HDL dependencies too. as far as I know there’s no tool* that does what Nix does here.
• projects that pertain to how I use and deploy NixOS systems. these shouldn’t see the light of day. these include personal deployments, usability libraries, embedded system flakes, and flakes that deploy reusable Nix appliances. the biggest part of this is the NixOS sub-distro that I use on my desktops and laptops; it features a doom emacs UI, a ton of fixes to make EXWM work more reliably, and a few other services that make NixOS work generally more like a lisp machine (and which make elisp work more like Nix). a lot of these systems have upstreamable fixes I haven’t bothered with. again, this is an area where it feels like there’s no* real substitute for NixOS.

[*] guix would work fine (and for the sub-distro it’d have an advantage in that everything would speak Lisp), but given the sheer fucking number of GNU shitheads I’ve seen supporting Jon, switching from Nix to Guix feels a lot like moving out of your abusive parents’ house so you can rent out your abusive parents’ guest house

and the fridge in the guest house has a lock on the freezer so you can’t have ice cream, but don’t worry, nobody will notice (for now) if you remove the lock with some bolt cutters

I’m willing to be wrong about guix — I really want a reasonable out from the current Nix ecosystem — but this doesn’t feel like a healthy choice

Agreed that releasing stuff isn’t necessary, especially stuff propping up the ecosystem.

Unfortunately, I think the rest of your statements are exactly inverted: the nixpkgs repo is pretty difficult to fork (moves fast, needs expensive CI/caches to properly operate), and while we may still have the nix expression language (and hey, lix is a good implementation of it!), I’m getting more and more convinced that it is not such a blessing.

The phd thesis though, that one is pretty good (currently reading it for realsies); lots of good ideas in it, regardless one’s thoughts about the expression language (: