Systematically find Nix packages of NLnet projects

[fricklerhandwerk]

The end goal is to have a comprehensive overview of NLnet projects that can be accessed, run, or hosted more or less with one command. Creating that overview is tracked in https://github.com/ngi-nix/ngipkgs/issues/165

The problem is that there are on the order of 1000 projects on https://nlnet.nl/project/. Many are defunct, some of them packaged in one of the repositories in https://github.com/ngi-nix (but often out of date), some upstream, some in Nixpkgs, some have NixOS modules, some only in pull requests.

The overview presentation should capture as many of them as possible, which requires knowing where they are. Ideally we'd also know where to focus our energy.

Task

Make a helper tool that outputs a table with the following for each project listed on https://nlnet.nl/project/:

• Whether it's Nix-packaged on any of the following:
  • Upstream
  • Nixpgks (unstable)
  • NGIpkgs
  • Any of the https://github.com/ngi-nix repositories
• Whether it also has a NixOS service module
• The version packaged
• The most recent version
• Number of GitHub stars (or equivalent)
• Activity heuristics (e.g. number of non-automatic commits in the past 6 months)

Ideally we could use some of that information to programmatically add references to working packages/modules.

The part addressing the numerous separate repositories to be migrated into this monorepo is tracked in https://github.com/ngi-nix/ngipkgs/issues/12

[fricklerhandwerk]

Here's the overview so far: https://docs.google.com/spreadsheets/d/1Cn2s62DszrpO81f5tV7sEqLn817rhxU7b5074W7XRoM produced by the groups led by @mightyiam. @wamirez is currently working on cleaning up and migrating the issues from a private repository used in the past years into this issue tracker. This will provide more insight into what work has already been done and which packages to prioritise. It will require some more effort to deduplicate all that information.

From the table we can already estimate that ~430 projects could potentially be packaged to begin with, between 80 and 200 already have packages of largely unknown quality and state of maintenance, and there's a number of alive projects that have a significant user base which aren't packaged with Nix at all.

For a bit of context, we're currently following these priorities:

• Finish ongoing work
• Migrate existing packages that aren't suitable for Nixpkgs into the monorepo, get them under CI, and increase code quality, test coverage, and consumer convenience
• Build up the collection of supported projects, regardless of where they are packaged
• Add new packages, prioritising EDA (open hardware design) tools, federated social media, and self-hosting facilities