| 26 Mar 2026 |
 nasso | could i do something like:
a nasso/nixos-config flake with my system-wide configuration, no home-manager, nothing user-specific a nasso/nixos-config-personal flake, pulls nasso/nixos-config, adds my personal home-manager config and has primaryUser = "nasso", and when i darwin-switch it only affects my personal user account a nasso/nixos-config-work flake (in a private repo), pulls nasso/nixos-config, adds my home-manager config for my work user account, it sets primaryUser = "<work account username>", and when i darwin-switch it only affects that user account
| 17:57:52 |
| nasso | this would let me make eveyrthing public except for nasso/nixos-config-work, which i dont necessarily want to share (yet) | 17:59:10 |
| nasso | and this sounds very modular and cool but idk i never seen any setup like that | 17:59:29 |
 Randy Eckenrode | That should work, but your source flake will need to export your config in a way the other flakes can consume it. | 17:59:31 |
| nasso | yea | 17:59:37 |
| nasso | it wouldn't necessarily be usable as a standalone nixos config, i might name it something like nasso/nixos-config-base or something idk | 18:00:21 |
| Randy Eckenrode | Personally, I’d just export it as a module you could enable in the dependent flakes. I’m not sure what the state of the art is for flake reuse because I don’t particularly like frameworks like that. | 18:00:33 |
| nasso | oh yeah You're Absolutely Right | 18:00:58 |
| nasso | i dont even need a flake for it | 18:01:03 |
| nasso | just a default.nix with whatever i want | 18:01:21 |
 K900 | @WeetHet thanks | 19:12:33 |
| K900 | The qtdeclarative fix worked | 19:12:38 |
| K900 | Merging to staging now | 19:12:42 |
 emily | uh | 19:13:53 |
| emily | K900: can you do the revert instead? | 19:14:00 |
| K900 | I did the revert | 19:14:09 |
| K900 | Which is the fix | 19:14:13 |
| emily | there's no reason to replace sigtool with rcodesign here | 19:14:15 |
| emily | ah ok | 19:14:16 |
| K900 | I have not seen the other fix | 19:14:18 |
| emily | I thought you meant the rcodesign one | 19:14:20 |
| K900 | No I have not seen that and I am not qualified to merge it | 19:14:29 |
| emily | which, maybe we should rcodesign in general but I think it's not very sustainable/worthwhile right now without a compatible CLI | 19:14:36 |
| Randy Eckenrode | Can we use a wrapper that’s good enough? | 21:54:18 |
| emily | probably, although since it doesn't expose the signature-stripping operation we'd presumably just ignore those commands | 21:57:05 |
| emily | at which point it's sorta the same as "teach sigtool to ignore those commands" | 21:57:12 |
| Randy Eckenrode | Using rcodesign would allow us to support some of the scenarios that sigtool doesn’t like signing bundles. | 22:01:25 |
 Sarah Clark | If you haven't looked at zh.fail in a while, the number of Darwin breakages is a thing to behold https://zh.fail/ | 22:15:51 |
| 27 Mar 2026 |
 Ihar Hrachyshka | not sure how to interpret it. is it good or is smth really broken? | 01:07:16 |
 viraptor | It's good. Number of broken packages falling. | 01:08:45 |
