| 11 Dec 2025 |
 bandithedoge | flake-parts is good though | 13:55:08 |
 teo (they/he) | it encourages people to use flakes though which is bad | 15:05:49 |
 neobrain | What do people like about flake-parts? Saw it mentioned a couple of times, but I find it difficult to extract anything meaningful out of their docs.
The perSystem stuff seems like a nice (if easily replaceable) utility, and besides that the docs mention modularity but then never actually explain what's different from vanilla flakes. | 15:22:09 |
| neobrain | * Why do people actually use flake-parts? Saw it mentioned a couple of times, but I find it difficult to extract anything meaningful out of their docs.
The perSystem stuff seems like a nice (if easily replaceable) utility, and besides that the docs mention modularity but then never actually explain what's different from vanilla flakes. | 15:23:42 |
 leona | you can actually spread outputs over multiple files with imports = [ ... ] which is way more complicated in raw flakes | 15:33:36 |
| leona | (you can achieve similar with e.g. lib.evalModules but that's more complex for my brain at least) | 15:34:50 |
| neobrain | Is imports specific to flake-parts? | 15:41:15 |
| neobrain | * Is imports specific to flake-parts?
I think I only used it when defining my Colmena hive, so I assumed it's a general nix thing | 15:42:03 |
| neobrain | * Is imports specific to flake-parts?
I think I only used it when defining my Colmena hive, so I assumed it's a general nix thing since I didn't need to pull in flake-parts explicitly | 15:42:22 |
| bandithedoge | the module system feels natural to anyone who's used nixos or home-manager | 17:04:28 |
| bandithedoge | and it actually feels like a specification with type checking instead of flake weirdness | 17:06:20 |
| bandithedoge | pretty sure it's part of lib.evalModules | 17:06:47 |
 Katalin 🔪 | it is | 17:37:18 |
| Katalin 🔪 | one of recent projects I used bare evalModules and it’s pretty nice | 17:39:00 |
| Katalin 🔪 | to build the flake outputs, that is | 17:39:31 |
| bandithedoge | honestly i'd love if nixpkgs had a single function that makes flakes module-based | 17:43:42 |
 aloisw | In reply to @jappie:jappie.dev
srid has 1 1-line commit in flake-parts as far as I can see 💀
https://github.com/hercules-ci/flake-parts/commit/34884706608919433ac07f09da18b32f6fde5dae shlevy has more though. | 19:15:31 |
 hexa | was only temp-banned iirc | 19:18:08 |
| hexa | https://github.com/NixOS/nix/pull/14753 | 19:19:11 |
| hexa | can't wait until this trickles down from https://github.com/DeterminateSystems/nix-src/pull/291 | 19:19:35 |
| aloisw | That gotta be a troll using LLM to generate the code, right? | 19:26:57 |
 KFears (burnt out) | Ronathan Jinger lmao | 19:27:35 |
| hexa | Per Twitter conversation, no. | 19:28:08 |
 Rutile (Commentator2.0) feel free to ping | 
Download image.png | 19:41:28 |
| Rutile (Commentator2.0) feel free to ping | just a single question: why | 19:41:33 |
 Charles | identifiers can contain - and cannot start with a number | 19:44:33 |
| neobrain | oh, that's an interesting perk. I'll have to try and see if I can find any good resources on it | 19:46:44 |
| Charles | what's not clear to me though is how to actually reference the value contained in the variable "1-a" | 19:46:53 |
| Rutile (Commentator2.0) feel free to ping | "and cannot start with a number"
if you look closely: "1-a" = 0 worked :D | 19:49:04 |
 rosssmyth | 
Download image.png | 19:49:08 |
