| 8 Sep 2022 |
 k0kada (he/him) | I think the reason why nobody realized yet that our current Flakes output are kinda bad and bizarre talks a lot about our user base (probably folks that are heavily invested in Nix/NixOS/Home-Manager) | 16:59:35 |
 ribosomerocker | well, from my point of view it says that not many people actively update their system while using it :P | 17:00:18 |
| k0kada (he/him) | In reply to @mon:tchncs.de
i mean, with non-flakes it's as easy as
doom-emacs = pkgs.callPackage (builtins.fetchTarball {
url = https://github.com/nix-community/nix-doom-emacs/archive/master.tar.gz;
}) {
doomPrivateDir = ./doom.d;
right? since that's usable in both home.packages and systemPackages? Without testing this looks right, because our default.nix is the function that makes a NDE derivation | 17:00:40 |
| ribosomerocker | though that might've just been because everyone I'm around just never got NDE to work and it always failed to function for them | 17:01:01 |
| k0kada (he/him) | However in Flakes, since we don't expose this function, things gets kinda ugly | 17:01:05 |
| k0kada (he/him) | I think it is something like nde = pkgs.callPackage (import "${inputs.nix-doom-emacs}" { }) | 17:02:02 |
| ribosomerocker | huh. well, i suppose exposing that function in the Flake would be easy | 17:02:45 |
| k0kada (he/him) | In reply to @mon:tchncs.de
huh. well, i suppose exposing that function in the Flake would be easy Yeah, but it will also be non-standard | 17:03:20 |
| k0kada (he/him) | Kinda like our hmModule | 17:03:27 |
| ribosomerocker | why would that be non-standard? a lot of the packages i'm using do that | 17:03:51 |
| k0kada (he/him) | BTW, since we are already discussing big refactors
ckie (they/them) WDYT about nixpkgs-fmt in place of nix-fmt?
| 17:04:00 |
| ribosomerocker | like hyprland and webcord flakes | 17:04:04 |
| k0kada (he/him) | (I hate nixfmt formatting) | 17:04:10 |
| k0kada (he/him) | * BTW, since we are already discussing big refactors
ckie (they/them) WDYT about nixpkgs-fmt in place of nixfmt?
| 17:04:15 |
| ribosomerocker | i think alejandra is the big name these days | 17:04:23 |
| k0kada (he/him) | In reply to @mon:tchncs.de
why would that be non-standard? a lot of the packages i'm using do that There is no standard to expose functions in Flakes yet | 17:04:45 |
| k0kada (he/him) | Even output.lib is non-standard | 17:04:55 |
| k0kada (he/him) | * Even outputs.lib is non-standard | 17:05:02 |
| ribosomerocker | In reply to@k0kada:matrix.org
There is no standard to expose functions in Flakes yet ah, yeah nothing i use exposes functions | 17:05:24 |
| k0kada (he/him) | BTW, package/packages should expose derivations, not functions either (so our usage is non-standard too) | 17:05:32 |
| ribosomerocker | except yknow, nix-doom-emacs which uses a function for package | 17:05:34 |
| ribosomerocker | yeah i know lol | 17:05:42 |
| ribosomerocker | hmmm... how about an overridable package? isn't that close to what your default.nix does or no? | 17:06:34 |
| k0kada (he/him) | So what I am thinking to do is:
outputs.packages will expose a proper full derivation of NDE (probably useless because it will use a default config or something, but I think it serves as a playground for people to test)outputs.mkNixDoomEmacsDrv (name TBD) will expose the default.nix as a proper functionoutputs.hmModule will stay since there is no standard for this
| 17:07:27 |
| k0kada (he/him) | In reply to @mon:tchncs.de
hmmm... how about an overridable package? isn't that close to what your default.nix does or no? Actually this is a good point | 17:07:48 |
| k0kada (he/him) | As long this doesn't cause another weird IFD I can try | 17:08:30 |
| ribosomerocker | i don't see a reason why it does :P | 17:08:43 |
| ribosomerocker | * i don't see a reason why it would :P | 17:08:51 |
| k0kada (he/him) | (Probably not because nixpkgs.lib.mkOverridable should exist) | 17:08:51 |
| k0kada (he/him) | Ah oh, forget what I said | 17:09:56 |
