| 4 Nov 2025 |
|  whispers (it/fae) left the room. | 20:25:00 |
| whispers (it/fae) joined the room. | 20:25:10 |
| 7 Nov 2025 |
|  seaweed salad changed their display name from LARPing tech Simps like Hacker News users to seaweed salad. | 00:07:30 |
| 8 Nov 2025 |
 zygohistomorphic glyph | is there a good way to have colmena pick up on submodules when using flakes? decided to move my secrets management to a submodule but it has proven to be a huge pain | 22:49:15 |
| zygohistomorphic glyph | 'nix build' supports submodules via nix build .?submodules=1 so I tried passing '--nix-option submodules 1' but that doesn't help. | 22:50:51 |
| 9 Nov 2025 |
| zygohistomorphic glyph | i ended up solving it by using flake-compat for the hive. couldn't get rid of flakes entirely since i rely on agenix-rekey | 13:30:54 |
| 14 Nov 2025 |
|  Florian set a profile picture. | 18:41:01 |
| 17 Nov 2025 |
|  Leon joined the room. | 00:52:52 |
| Leon | Is there a way to avoid hard-coding any particular system architecture into a flake-based Colmena hive? I supply an explicit nodeNixpkgs override for every node (with system set to that node's particular architecture), but I still need to supply a value for colmena.meta.nixpkgs nonetheless. With pure flake eval, builtins.currentSystem is not available for that. Right now, I'm using some random value which I don't have binfmt setup for (powerpc64le-freebsd) to make sure that nothing actually depends on that value, but that feels more like a hack than anything else. | 01:07:09 |
| Leon | I'm a little suprised that Colmena doesn't use the de-facto standard flake output format of colmenaHive.${system} = ... (which would solve that problem). Could also be helpful for writing flakes that use IFD to generate the hive attribute set and can be evaluated cross-platform. Is there any particular reason why it doesn't work like that? | 01:10:27 |
| 18 Nov 2025 |
|  elisaado joined the room. | 20:24:46 |
