Sender | Message | Time |
---|---|---|
5 Jul 2025 | ||
18:09:42 | ||
In reply to @charles:computer.surgery I mean which one is the correct? nix --experimental-features 'nix-command flakes' build ./nix#nixosConfigurations.UwU.config.system.build.toplevel Or nix --experimental-features 'nix-command flakes' build /mnt/home/lag/nix#nixosConfigurations.UwU.config.system.build.toplevel | 20:02:56 | |
20:03:58 | ||
it doesn't matter where the files are | 20:06:28 | |
just as long as they're the right files | 20:06:38 | |
because if both of those point to a copy of the exact same nix expressions then the result will be the exact same because that's kind of the point of nix | 20:07:41 | |
In reply to @charles:computer.surgeryOk | 21:14:04 | |
6 Jul 2025 | ||
08:31:34 | ||
Any idea how to do this without IFD?
| 11:54:17 | |
How do you notice IFD? | 12:00:32 | |
Wouldn't self.callPackage "${patchedLix}/package.nix" cause IFD? | 12:00:54 | |
Ah yes it does | 12:04:20 | |
does the patch affect the derivation? | 12:16:10 | |
if not, then why the two derivations? | 12:16:16 | |
if yes, then it is inherently IFD | 12:16:24 | |
Oh yeah, I can just do this:
| 12:24:51 | |
Thanks | 12:24:53 | |
How is it different thought ? You are still doing callPackage from your inputs.lix | 12:33:22 | |
lix is an input, it's already eval-time | 12:35:31 | |
17:17:22 | ||
having some issues trying to switch over to lix from my current setup. i followed the instructions on the install page but when i go to rebuild, it just kinda... sits there and does nothing? i hear my cpu fan spin up like its doing something but theres no output from the terminal. anyone had this issue before? | 21:37:07 | |
never mind i fixed it? i think? i had to remove the setting i had for my download buffer | 22:17:09 | |
7 Jul 2025 | ||
02:40:42 | ||
i think so? i was not aware that it doesn't at least | 02:41:37 | |
that looks like a lix-module bug but I have no idea why that is happening. we should definitely just default to unknown rather than blowing up | 02:43:22 | |
@ersei: good:
bad:
I have /no/ idea why that happened though, it appears that it somehow downloaded that tarball without recording the correct headers? | 02:47:00 | |
either way we should just make the lix-module more resilient to that | 02:47:15 | |
done | 02:51:47 | |
the lix team is unlikely to port this themselves as they are not especially in want of more flakes features themselves, and our git fetcher looks vastly different. if someone more enthused about flakes wants to write an implementation of the feature (which sounds reasonable enough, though the performance is guaranteed to be awful for anything but tiny repos, so I don't know who would actually need this feature who wouldn't be better served by other options like not using flakes, using lix flake-compat, or not using submodules), they can do so. You can tag me to review it, assuming it has tests. | 02:56:09 | |
that's a great question. sounds like a bug given i think we just use upstream boehmgc? it will probably be fixed once we kill 2.90. | 02:57:48 |