!RROtHmAaQIkiJzJZZE:nixos.org

NixOS Infrastructure

407 Members
Next Infra call: 2024-07-11, 18:00 CEST (UTC+2) | Infra operational issues backlog: https://github.com/orgs/NixOS/projects/52 | See #infra-alerts:nixos.org for real time alerts from Prometheus.124 Servers

Load older messages


SenderMessageTime
30 Mar 2026
@hexa:lossy.networkhexanodePackages removal12:43:43
@k900:0upti.meK900Oh my fucking god12:44:41
@hexa:lossy.networkhexahttps://github.com/nix-community/nix-index/blob/master/src/listings.rs#L2912:44:45
@k900:0upti.meK900Yeah figures12:45:19
@k900:0upti.meK900Need to make that shit overridable12:45:25
@vcunat:matrix.orgVladimír Čunát I assume it doesn't matter that we drop nodePackages from the index on 25.11 as well? 13:10:04
@k900:0upti.meK900Probably not13:10:17
@vcunat:matrix.orgVladimír ČunátContext: https://github.com/nix-community/nix-index/pull/300 and around13:10:30
@hexa:lossy.networkhexadeploying pluto13:11:41
@k900:0upti.meK900I am tempted to actually make that shit overrideable properly13:11:48
@hexa:lossy.networkhexaok, nix-index on pluto is working again13:20:47
@vcunat:matrix.orgVladimír Čunát

I am tempted to actually make that shit overrideable properly

If I wanted to properly do this, this list of extra attributes should be somewhere in nixpkgs instead. Though that might be overengineering.

13:31:29
@k900:0upti.meK900recurseIntoAttrs = "maybe"13:34:55
@k900:0upti.meK900/j13:34:58
@k900:0upti.meK900https://github.com/nix-community/nix-index/pull/30114:55:57
@emilazy:matrix.orgemilyI think we already have a thing for this15:16:32
@emilazy:matrix.orgemilypackages-config or something15:16:37
@emilazy:matrix.orgemilyit's one of those things where we have five slightly different versions of the same thing15:16:51
@emilazy:matrix.orgemilyhttps://github.com/NixOS/nixpkgs/blob/master/pkgs/top-level/packages-config.nix15:17:22
@emilazy:matrix.orgemilycan it be driven by this?15:17:27
@emilazy:matrix.orgemilyotherwise it's just inconsistency between various things trying to achieve the same goal15:18:58
@rick:matrix.ciphernetics.nlRick (Mindavi)I guess having it as a flag brings it a step closer to script it15:25:55
@emilazy:matrix.orgemily it's a bit orthogonal, since packages-config.nix is passed as the Nixpkgs config when instantiating the package set 18:22:20
@emilazy:matrix.orgemily therefore the functionality could actually be removed from nix-index entirely – it would just be a matter of instantiating Nixpkgs appropriately 18:22:47
@emilazy:matrix.orgemilyhttps://github.com/nix-community/nix-index/blob/master/src/nixpkgs.rs#L2918:23:36
@emilazy:matrix.orgemily
        .arg("config")
        .arg("{ allowAliases = false; }") // override default nixpkgs config discovery

would be replaced by setting NIXPKGS_CONFIG=$nixpkgs/pkgs/top-level/packages-config.nix (or passing --arg config "import $nixpkgs/pkgs/top-level-packages.config.nix"), and then the scopes stuff can go away.

18:24:49
@emilazy:matrix.orgemily (ofc haskellPackages, coqPackages, texlive.pkgs might need handling in packages-config.nix… though I wonder if all of those are actually necessary these days, e.g. haskellPackages is certainly recursed into?) 18:25:43
@emilazy:matrix.orgemily it doesn't make much sense for the info used by https://github.com/NixOS/nixpkgs/blob/7d983847f0c19127319281055f1e086a04384de5/pkgs/top-level/packages-info.nix and https://github.com/NixOS/nixos-search/blob/7e05acf78305990769523f7670150e9e33f1f8f4/flake-info/src/commands/nixpkgs_info.rs to differ from that used by nix-index 18:26:49
@emilazy:matrix.orgemily since packages.json from Hydra uses it https://github.com/NixOS/nixpkgs-channels/blob/4762fba469e2baa82f983b262e2c06ac2fdaae67/pkgs/top-level/make-tarball.nix#L106 and nix-index is doing the same essential thing 18:27:20
@emilazy:matrix.orgemily(er… how did I end up on that old repo)18:28:20

Show newer messages


Back to Room ListRoom Version: 6