| 25 Mar 2026 |
 niko | And by required I don’t necessarily mean required as it stands, but also do you aim for them to be required | 17:45:58 |
 raitobezarius | If Lix cannot delegate large ranges of UIDs, it could just retract the system feature uid-range, we want to have as many users as possible with that system feature on by default | 17:47:08 |
| raitobezarius | As for the daemon and socket activation, for now (but this is actively debated based on the operational experiences), we intend to have the job of executing the sub daemon be done by a socket activation component: inetd, s6 socket activator, systemd; whichever you want
You can keep that subdaemon running and reuse the connection but Lix doesn't implement any special mechanism for that; it's your service manager who gives you the options (systemd does) | 17:49:18 |
| raitobezarius | For cgroups, we intend to make this reasonably mandatory but we are here again very careful about the operational stories, if some people bring to our attention this may be severely disruptive for valid reasons, we will leave escape hatches and review our plans | 17:50:34 |
 Display Name | Does lix solve dependency hell? | 18:06:18 |
| Display Name | (Sorry I am new to nix, I am assuming lix is backwards compatible iirc by the website) | 18:07:29 |
| Display Name | * | 18:07:43 |
 whispers [& it/fae] | what do you mean by "dependency hell"? | 18:08:30 |
| whispers [& it/fae] | the packages you build with lix will be the definitions you write, it can't magically change those | 18:08:49 |
| Display Name | Oh, my bad, sorry again I am new to this (I didn't downloaded, just wondering) | 18:10:20 |
| Display Name | That answer my question thanks | 18:11:11 |
| Display Name | (Not sarcastically of course) | 18:11:33 |
| whispers [& it/fae] | nix is ultimately a language for specifying build instructions ("derivations") and it's the job of a nix implementation (cppnix, lix, etc.) to interpret those instructions and to run builds | 18:12:19 |
| whispers [& it/fae] | * nix is ultimately a language for specifying build instructions ("derivations") and it's the job of a nix implementation (cppnix, lix, etc.) to interpret those instructions and run the builds | 18:12:38 |
| whispers [& it/fae] | so if your build instructions are for something with a lot of dependencies, you'll need all of those dependencies. usually whether something is "dependency hell" is dependent on the upstream project you're packaging | 18:13:19 |
| whispers [& it/fae] | * so if your build instructions are for something with a lot of dependencies, you'll need all of those dependencies. usually whether something is "dependency hell" is dependent on the upstream project you're packaging and how many dependencies it has | 18:13:29 |
| Display Name | Oh, okay, may I ask I heard nix is also a package manager is that correct? | 18:14:11 |
| raitobezarius | In reply to @takentakentaken:matrix.org
Oh, okay, may I ask I heard nix is also a package manager is that correct? Yes | 18:17:03 |
| Display Name | Okay | 18:17:10 |
| raitobezarius | In reply to @takentakentaken:matrix.org
(Sorry I am new to nix, I am assuming lix is backwards compatible to nix iirc by the website) I appreciate you disclosing this, if you have some time, please let us know what we should have written in our documentation or website to make this information easier to discover :) | 18:18:07 |
| Display Name | Okay | 18:19:05 |
|  delafthi joined the room. | 20:52:56 |
 piegames | EsperLily [she/her]: does https://git.lix.systems/lix-project/lix-website/pulls/75/files make the release notes more clear for you? | 21:15:20 |
 EsperLily [she/her] | that looks like it just copies the language from the release notes into the blog post? I actually think the blog post language is fine and it's the release notes language that isn't clear enough. It doesn't describe why the dynamic attributes part is considered "weird semantics" (AFAICT it behaves as I'd expect), and actually the rec-set-merges part doesn't really describe the issue properly either, I had to go look at the relevant change itself to see a proper example (in particular, it's really weird that { a = rec { b = 2; }; a.c = b; } actually behaves identically to { a = rec { b = 2; c = b; }; } even though I can understand that from a "it's just syntactic sugar" perspective) | 23:33:02 |
| 26 Mar 2026 |
| EsperLily [she/her] | I guess more generally, while I think rec-set-merges could stand to be described better, I agree that it can be surprising enough that it should be banned. But banning having dynamic attrs in a recursive set is annoying, I don't think there's anything confusing about the current behavior as I understand it, and it means if I want to add a dynamic attribute to an existing recursive set I have to merge it in separately (which can hurt code readability and also change behavior if the dynamic attribute might conflict with a non-dynamic attr as that should be an error but rewriting this as a merge operation turns it into an overwrite) | 00:05:36 |
 Lisanna | That's mildly horrifying | 04:17:11 |
| piegames | In reply to @esperlily:matrix.org
that looks like it just copies the language from the release notes into the blog post? I actually think the blog post language is fine and it's the release notes language that isn't clear enough. It doesn't describe why the dynamic attributes part is considered "weird semantics" (AFAICT it behaves as I'd expect), and actually the rec-set-merges part doesn't really describe the issue properly either, I had to go look at the relevant change itself to see a proper example (in particular, it's really weird that { a = rec { b = 2; }; a.c = b; } actually behaves identically to { a = rec { b = 2; c = b; }; } even though I can understand that from a "it's just syntactic sugar" perspective) Thank you. It's a bit difficult to boil down each deprecation to only a single sentence. Do the long from deprecation descriptions in the manual work better in that regard? | 08:17:17 |
 boop-north | what should I do, wait for next release? | 08:42:17 |
| boop-north | setup a systemd daemon? nix-shell didnt play nicely with selinux I doubt nix will | 08:42:47 |
| boop-north | https://docs.lix.systems/manual/lix/stable/introduction.html This page should answer what you are looking for, just the introduction page is a 2 minute read. | 08:46:55 |
