| 23 Oct 2025 |
|  lavender.pet changed their profile picture. | 13:15:50 |
| lavender.pet changed their profile picture. | 13:19:05 |
| 24 Oct 2025 |
|  cjab joined the room. | 02:01:05 |
 522 it/its ⛯ΘΔ | lix module overlay without flakes question: i'm experimenting with removing flakes from my system but am running into an issue where some override accesses the lix.rev of the lix argument, which doesn't exist (i'm using npins, it's called .revision)
nixpkgs.overlays = [
(import "${sources.lix-module}/overlay.nix" {
lix = sources.lix // {
# rev = sources.lix.revision;
};
versionSuffix = "-pre-${builtins.substring 0 12 sources.lix.revision}";
})
];
in my configuration.nix gets me
error: attribute 'rev' missing
at /nix/store/bbl3a77bpxlqml9sqclz2ax2fqvgj1y3-source/overlay.nix:120:62:
119| __intentionallyOverridingVersion = true;
120| version = "2.94.0-lix-${builtins.substring 0 7 lix.rev}";
| ^
121|
(with my hacky fix for this commented out)
| 10:46:28 |
| 522 it/its ⛯ΘΔ | specifically it only happens when adding colmena to my system, since that causes nix-eval-jobs to be evaluated(which this is an overlay for) | 10:47:18 |
 Molly | i had a similar problem a while back, iirc this is a known issue. the fix which i think raito recommended was:
((import "${pins.lix-module}/overlay.nix") { lix = { outPath = pins.lix; rev = "dirty"; }; })
| 11:41:02 |
| 522 it/its ⛯ΘΔ | so basically what i did, intentionally give it the rev it wants, ig? | 11:42:59 |
| 522 it/its ⛯ΘΔ | (idk how much the rev matters here, but i might as well give it the commit sha because i do have that) | 11:43:27 |
| 522 it/its ⛯ΘΔ | is there an issue for this so i can link to it in my config? | 11:43:39 |
| 522 it/its ⛯ΘΔ | ah, https://git.lix.systems/lix-project/lix/issues/906 | 11:44:18 |
| 522 it/its ⛯ΘΔ | oh but that seems to suggest applying a patch that has been merged | 11:45:25 |
| 522 it/its ⛯ΘΔ | ah, https://git.lix.systems/lix-project/nixos-module/issues/56 seems more appropriate | 11:46:05 |
|  m0tholith joined the room. | 14:24:48 |
| m0tholith | hey everyone, im getting into using lix on nixos using the "advanced change" config, and im having trouble since i use the pipes feature. can i keep using pipes or do i have to stop using the,? | 14:26:36 |
| m0tholith | * hey everyone, im getting into using lix on nixos using the "advanced change" config, and im having trouble since i use the pipes feature. can i keep using pipes or do i have to stop using them? | 14:26:40 |
 Sofie 🏳️⚧️ (she/her) | what...?
[root@cedarix:~]# nix shell nixpkgs#git
error: source tree referenced by 'path:/nix/store/9v6qa656sq3xc58vkxslqy646p0ajj61-source?lastModified=1759381078&narHash=sha256-gTrEEp5gEspIcCOx9PD8kMaF1iEmfBcTbO0Jag2QhQs%3D&rev=7df7ff7d8e00218376575f0acdcc5d66741351ee' does not contain a '/flake.nix' file
| 14:26:59 |
 K900 | You can use the pipe-operator experimental feature | 14:27:26 |
| m0tholith | yeah i just checked the lix manual and it looks like that's what i need. kind of weird that it's named operator instead of operators like in cppnix? | 14:28:23 |
| Sofie 🏳️⚧️ (she/her) | this was to an remove rpi | 14:30:46 |
| K900 | Because we only have forward pipe | 14:33:23 |
| K900 | No reverse pipe | 14:33:27 |
| Sofie 🏳️⚧️ (she/her) | * | 14:59:33 |
| Sofie 🏳️⚧️ (she/her) | is this expected? | 15:42:39 |
| K900 | No | 15:51:18 |
 piegames | In reply to @k900:0upti.me
Because we only have forward pipe That's half-true. Because CppNix did not look at the Lix implementation when doing it | 16:29:23 |
| piegames | We currently have a backpipe operator, but it is not guaranteed to stay | 16:29:50 |
 daylin | is it likely the "forward" operator will stay? I like it very much and would be sad to see it go | 17:26:56 |
| piegames | Yes | 17:28:36 |
 lillecarl | I talked to Rickard from nixbuild.net and we were discussing GC, I mentioned my idea for time based GC using the registrationtime field of the database (and keeping it up2date) and he said that they're doing something like that at nixbuild (but more elaborate with ownership tracking and $stuff). What's Lix stance on db.sqlite schema changes? I imagine it's hard to get a new field added to ValidPaths table, it would be really nice to track when something was last required (including with "nix copy"), then nix-serve(ng) and a normal NixOS box really ticks all boxes for a good LRU cache solution OOTB | 22:37:15 |
| lillecarl | https://github.com/NixOS/nix/issues/7572 I'm not the first person to have this idea, hehe | 23:04:44 |
