07:34 <craige> angerman, hamishmack - this thread has lurched back to life: https://discourse.nixos.org/t/broken-haskell-packages-should-be-annotated-why-they-are-broken/6176/7 and I was wondering what the barriers / objections there were to adopting haskell.nix or is that just an ugly can of worms I'm prying open?

08:49 <angerman> craige: well, unless someone from nixpkgs comes and says they *want* haskell.nix (which would bring in IFDs, though I have my doubts), i don't thin it makes much sense.

08:50 <angerman> craige: my view is that haskell.nix provides a lot of features that are useful to people who deal a lot with haskell for their own good, but for something like nixpkgs (which has very different goals and needs with its haskell infra) it might not be the best option.

08:50 <angerman> nixpkgs wants to provide software packages, and well there is software written in haskell, e.g. pandoc, git annex, ... so you need a way to easily bring that into nixpkgs.

08:52 <angerman> people who deal with lots of haskell at work, like you and me, we have different needs, and need to support a wide set of configurations for various haskell projects we build. Also our software evolves much faster than what nixpkgs has to deal with. I think the nixpkgs haskell infra is good for nixpkgs purposes, but it's fundamentally flawed and horribly terrible for anyone who actually wants to develop haskell.

10:05 <michaelpj> EncodePanda: a non-exhaustive list of things that could be issues: 1) you've got a user-local ~/.config/nix/nix.conf which overrides it. 2) the daemon needs to be restarted to pick up the global config