| 24 Feb 2026 |
 K900 | The flake based CLI | 09:57:42 |
 neobrain | so nix-command? | 09:58:47 |
| neobrain | * so the same as nix-command? | 09:58:54 |
| K900 | Yes | 09:58:54 |
| neobrain | * so the same as the nix-command feature? | 09:58:59 |
| neobrain | I see, thanks :) | 09:59:24 |
 sterni | In reply to @raitobezarius:matrix.org
it already managed that in the complicated cross contexts :p hm in *Platform? neat | 10:31:24 |
 raitobezarius | yeah | 10:31:38 |
| raitobezarius | some BSD stuff | 10:31:41 |
| raitobezarius | related to Python manual splicing optimizations IIRC | 10:31:54 |
|  Liyua joined the room. | 10:47:08 |
 goldstein | hi! does anyone know if dropping support for path literals in flake input urls is intentional? code like
{
inputs.a.url = ./a;
}
works on cppnix, but not on lix, and I’m wondering if that’s intentionally deprecated or just accidentally dropped
| 13:22:48 |
| goldstein | * hi! does anyone know if dropping support for path literals in flake input urls is intentional? code like
{
inputs.a.url = ./a;
outputs = _: {};
}
works on cppnix, but not on lix, and I’m wondering if that’s intentionally deprecated or just accidentally dropped
| 13:23:04 |
 piegames | Uhm | 13:25:37 |
| piegames | What's the error message? | 13:25:44 |
| goldstein | error:
… while evaluating flake attribute 'url'
at /nix/store/x9y295jnchs4qlyxk9im7nqzfp1cr300-source/flake.nix:2:3:
1| {
2| inputs.a.url = ./a;
| ^
3| outputs = _: {};
error: expected a string but got a path at /nix/store/x9y295jnchs4qlyxk9im7nqzfp1cr300-source/flake.nix:2:3
| 13:26:20 |
| goldstein | * error:
… while evaluating flake attribute 'url'
at /nix/store/x9y295jnchs4qlyxk9im7nqzfp1cr300-source/flake.nix:2:3:
1| {
2| inputs.a.url = ./a;
| ^
3| outputs = _: {};
error: expected a string but got a path at /nix/store/x9y295jnchs4qlyxk9im7nqzfp1cr300-source/flake.nix:2:3
| 13:26:30 |
| goldstein | nix (Lix, like Nix) 2.95.0-pre20260223-dev_dirty
(no idea how to find out which specific commit it was, but I’m following main and have updated recently)
| 13:27:46 |
| goldstein | pretty sure it’s just this check?
https://git.lix.systems/lix-project/lix/src/commit/6b4de44461dbd872d12eca8721f24b8756f2ee99/lix/libexpr/flake/flake.cc#L172 | 13:30:02 |
| goldstein | although it was introduced way before the split, so not sure why cppnix doesn’t have the same problem | 13:31:40 |
| goldstein | I think it was just allowed in cppnix post-split, see
https://github.com/NixOS/nix/commit/3180671cabeb6a6010057770731e12761ed5666c
do we want this to work in lix too? I can try to port this patch | 13:34:27 |
| goldstein | also, more generally: what’s the policy for reporting compatibility issues? I was doing an equivalent of nix crater run, so I have a bunch of fun stuff, but I’m not sure what is intentional and what isn’t. should I just report it on the issue tracker? | 13:44:32 |
| piegames | Flakes are currently frozen until at least after the 2.95 release | 13:49:16 |
| piegames | In reply to @goldstein:tty5.dev
also, more generally: what’s the policy for reporting compatibility issues? I was doing an equivalent of nix crater run, so I have a bunch of fun stuff, but I’m not sure what is intentional and what isn’t. should I just report it on the issue tracker? Opening issues is fine, if you have a bulk then Zulip or here are good too | 13:49:50 |
| piegames | What setup are you using for your runs? | 13:49:58 |
| goldstein | I’m currently writing a post, but it’s basically “get a random flake from the internet, remove flake.lock, run nix flake lock, try to evaluate some common outputs” | 13:50:34 |
| goldstein | I’ve checked ~7600 flakes this way and am currently sorting the results | 13:50:57 |
| piegames | Nice | 13:51:54 |
| piegames | Would you be interested in contributing to https://git.lix.systems/lix-project/flaker | 13:52:04 |
| piegames | Because currently it only does parser diffing, not eval | 13:52:27 |
