| 3 Feb 2026 |
 Sofie 🏳️⚧️ (she/her) | ah | 17:52:39 |
 lillecarl | hexa: systemctl edit --runtime nix-daemon.service 🧠💡 | 23:17:49 |
| 4 Feb 2026 |
|  tiferrei changed their profile picture. | 02:26:53 |
|  Domi set a profile picture. | 08:37:50 |
| Domi changed their profile picture. | 08:38:06 |
| Domi changed their profile picture. | 08:38:22 |
 actioninja | I'm actually completely baffled to find out that in nix math operations are parse time rewritten to magic global symbols | 12:30:13 |
 piegames | It is baffling, yes | 12:30:52 |
| piegames | Stuff you do when writing code in a garbage language that makes everything tedious and you just want to Get Things Done and take the path of least resistance | 12:31:41 |
| piegames | And so, simply adding new builtins was easier (less boilerplate) than introducing new AST nodes | 12:32:07 |
| actioninja | even with the whole "I don't want to add an infix ast node" problem I can think of saner ways of doing it, I feel like just having some magically defined operator functions with names as the operators and then shuffling to rpn would be less work and less prone to crazy edge case behavior | 12:37:35 |
| actioninja | and then you get rpn for free and the lispers can write their unreadable math | 12:38:34 |
| piegames | (C++ is bad, and one has to be mindful of how much it has improved since 20 years ago) | 12:40:46 |
| tiferrei left the room. | 14:29:46 |
 niko ⚡️ | In reply to @actioninja:4d2.org
I'm actually completely baffled to find out that in nix math operations are parse time rewritten to magic global symbols Yes, but also not always because + can also add strings together so obviously that doesn’t translate to __add | 21:45:25 |
| niko ⚡️ | Which imo is the worst part, because nix can’t even decide which way it wants to go | 21:45:39 |
| Sofie 🏳️⚧️ (she/her) | currently making a tofu/terraform module to deploy nixos devices :3 | 22:46:27 |
| 5 Feb 2026 |
| piegames | In reply to @niko:nrab.lol
Yes, but also not always because + can also add strings together so obviously that doesn’t translate to __add Okay so, + is not addition, it is string concatenation :) | 00:37:13 |
| piegames | Basically in the beginning Nix had no numbers. So when they were introduced, + got overloaded, but for the new operators primops were used instead | 00:38:51 |
| piegames | https://git.lix.systems/lix-project/lix/src/branch/main/lix/libexpr/nixexpr.hh#L619 | 00:40:27 |
| piegames | (isInterpolation used to be called forceString IIRC, but that name was even more confusing) | 00:41:07 |
 antifuchs | In reply to @piegames:flausch.social
Okay so, + is not addition, it is string concatenation :) Nice, that’s even more cursed | 01:03:54 |
 koaledu | Every day we stray further from God. | 01:06:09 |
 Emma [it/its] | hm, is there a way to parallelise a flake update? | 02:04:30 |
| Emma [it/its] | given that -j doesn't seem to be applicable here | 02:04:58 |
 Coca | I thought flakes updated concurrently already but that's probably not true | 02:15:57 |
| Emma [it/its] | they dont (can check with -v) | 02:19:21 |
| Coca | Hm, that makes npins the only pinner that's concurrent then in the nix ecosystem then I think | 02:20:56 |
| Coca | Weird nobody has done it before | 02:21:17 |
| Coca | * Hm, that makes npins the only pinner that's concurrent then in the nix ecosystem I think | 02:21:26 |
