!9IQChSjwSHXPPWTa:lix.systems

Lix

1105 Members
Lix user channel. Feel free to discuss on-topic issues here and give each other help. For matrix.to links to the rest of the Lix channels, see: https://wiki.lix.systems/books/lix-organisation/page/matrix-rooms295 Servers

Load older messages


SenderMessageTime
16 Nov 2025
@piegames:flausch.socialpiegamesIn an ideal world, we'd get dedicated custom string types with sanity checking and user-defined in pure Nix via macro-like features. In this world, be happy if we get annotated string types so that you can run a linter on it07:07:02
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)Redacted or Malformed Event16:02:31
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)Redacted or Malformed Event16:03:45
@neindochoh:catgirl.cloudneindochoh! / Seti (they/them)thanks for sharing this hint! :) this makes things like neovim configuration inside nix files soo much more readable for me.18:17:26
@emma:rory.gayEmma [it/its] Q: is there a way to check if all outputs in a flake actually build? 19:15:40
@emma:rory.gayEmma [it/its] because nix flake check only checks if they eval, not if they actually build succesfully 19:19:07
@522_:catgirl.cloud522 it/its ⛯ΘΔuhhh... since (custom) checks are really just arbitrary derivations, can you have a check that depends on all the outputs?19:19:46
@samasaur:matrix.orgsamasaurthat is the typical "solution" i see people use19:20:04
@emma:rory.gayEmma [it/its]i guess?19:20:08
@samasaur:matrix.orgsamasaursome flake output that depends on actually building the outputs you want to test19:20:22
@emma:rory.gayEmma [it/its]thanks for pointing me to the fact that checks even exist19:24:13
@emma:rory.gayEmma [it/its] found this snippet on github, that seems to do the trick: checks = pkgs.lib.attrsets.unionOfDisjoint { /* Actual checks */ } self.packages; 19:24:26
@bmillwood:matrix.org@bmillwood:matrix.org left the room.19:31:00
@weethet:catgirl.cloudWeetHet

Re: zulip on nix4 cli

Do we actually need anything except nix2-cli? I think it's already pretty much okay. The worst part is the fileish syntax which can be replaced/extended with flakeref-style one and realistically not much else is needed

We can add nix-run natively if we want by locking onto meta.mainProgram and bin/out outputs or keep it as a separate program like the one I wrote and do we actually need anything else?

20:10:27
@weethet:catgirl.cloudWeetHetWe can make it so that when pure-eval is enabled the directory containing the file/the file itself get copied to the store so that it works properly if we don't wanna do pure eval properly or we can do proper path isolation and don't copy to store at all, we don't need new CLI for that as well20:13:39
@weethet:catgirl.cloudWeetHet If we really need, making nix smth an alias to nix-smth is also possible 20:14:11
@weethet:catgirl.cloudWeetHet The biggest issue for me is not having a unified function (fetchTree) so that the functionality of not caring if the reference is a path or a link would be available in -E/--expr mode 20:18:02
17 Nov 2025
@olivia:computer.surgeryolivia changed their profile picture.00:23:17
@stephen:crabsin.spacen3tcat changed their profile picture.01:20:47
@antifuchs:asf.computerantifuchs recently, my builds in CI (over residential internet, ofc) keep getting aborted because downloading from cache.nixos.org yields a short read on compressed input, which results in errors like error: some substitutes for the outputs of derivation '/nix/store/4v4yw7ip7ir7bcg3x24f5hrl8ql7rg3p-alsa-firmware-1.2.4-zstd.drv' failed (usually happens due to networking issues); try '--fallback' to build derivation from source. is there a way to have the build retry these substitution downloads, instead of having it attempt to build the world? 02:12:25
@raitobezarius:matrix.orgraitobezarius
In reply to @antifuchs:asf.computer
recently, my builds in CI (over residential internet, ofc) keep getting aborted because downloading from cache.nixos.org yields a short read on compressed input, which results in errors like error: some substitutes for the outputs of derivation '/nix/store/4v4yw7ip7ir7bcg3x24f5hrl8ql7rg3p-alsa-firmware-1.2.4-zstd.drv' failed (usually happens due to networking issues); try '--fallback' to build derivation from source. is there a way to have the build retry these substitution downloads, instead of having it attempt to build the world?
Are you sure this is not the curl H2 bug? Apparently fixed in nixpkgs unstable
02:29:55
@antifuchs:asf.computerantifuchsOooh is there one? Then it’s probably that02:49:40
@antifuchs:asf.computerantifuchsThanks!02:49:49
@shalokshalom:kde.orgShalokShalom joined the room.07:52:28
@zeldahreal:matrix.orgZelda 🏳️‍⚧️ [she/her] joined the room.16:23:49
@zeldahreal:matrix.orgZelda 🏳️‍⚧️ [she/her] changed their display name from Zelda to Zelda 🏳️‍⚧️ [she/her].16:25:28
18 Nov 2025
@mzerodev:matrix.orgZoe Z something that I think is probably a bug, but it's cursed and underspecified enough that it's hard to say: exportReferencesGraph of let drv = unsafeDiscardStringContext pkg.drvPath; in appendContext drv { ${drv} = { path = true; }; } is the same as exportReferencesGraph of pkg.drvPath, but I think a more reasonable behaviour would be for it to only include the drvs and not the outputs, like nix-store -qR drvPath 01:00:18
@quadradical:federated.nexusQuadRadical (Ping) changed their profile picture.01:59:42
@raitobezarius:matrix.orgraitobezariuscan you open an issue so we can discuss it and see what should be done02:06:39
@raitobezarius:matrix.orgraitobezariushm i completely forgot about adding LIX_VERSION stuff inside 2.94.0 ugh02:06:48

Show newer messages


Back to Room ListRoom Version: 10