| 21 Aug 2025 |
emily | I would simply solve it | 20:58:41 |
emily | tbh the module system already defers a lot of that stuff in a way that makes it easier to cross-check | 20:59:09 |
Grimmauld (any/all) | sure, but how? Quantum computer evaluating all possible nix configs at once? 😆 | 20:59:25 |
Grimmauld (any/all) | i mean, some simple heuristics would already help, like test rebuilds and seeing whether test configs still eval | 21:00:07 |
Grimmauld (any/all) | but a complete eval test as we have for packages is imo impossible | 21:00:26 |
emily | I mean if you have mkIf x y | 21:02:28 |
emily | then y gets included in the intermediate results | 21:02:34 |
emily | it is more structured than completely raw language conditionals, you can see more branches at once | 21:02:55 |
emily | anyway, we don't need to test every combination, we don't for packages either | 21:03:10 |
emily | it's just NixOS tests are expensive for spot-checking | 21:03:18 |
emily | (I bet we could do a model checker thing for complete coverage. a thing that is famously not expensive) | 21:03:41 |
emily | K900: why do we need the Qt 5 thing | 21:07:43 |
emily | I guess just because Qt 5? | 21:07:52 |
K900 | Yes | 21:08:01 |
K900 | I am cooking | 21:08:03 |
K900 | Let me cook | 21:08:04 |
emily | 🍳 | 21:08:13 |
emily | drop qt 5 when | 21:08:17 |
emily | how much of it is there even left | 21:08:23 |
emily | is it intentional to have plasma5Packages = libsForQt5; | 21:10:34 |
emily | in not even aliases.nix | 21:10:36 |
K900 | Yes | 21:10:43 |
K900 | Well | 21:10:45 |
emily | used by like 5 things | 21:10:46 |
K900 | "Intentional" | 21:10:46 |
K900 | As in stupid | 21:10:48 |
emily | pkgs/applications/audio/playbar2/default.nix
35: description = "Mpris2 Client for Plasma5";
snuck under the radar… it's getting away…
| 21:11:16 |
K900 | Weirdly that's fine actually | 21:11:49 |
K900 | plasma-framework is still there | 21:11:54 |
K900 | Because for Stupid Legacy Reasons in 5 it's in Frameworks, not Plasma | 21:12:02 |