| 18 Feb 2026 |
 Alyssa Ross | Not causing warnings for some fraction of our users, in the absence of technical arguments against making the change that would silence the warnings sounds fine to be. | 15:09:40 |
| Alyssa Ross | * | 15:09:49 |
 Grimmauld (any/all) | Nixpkgs has a lot of dumb nix lang edgecases. I think the effort of making nixpkgs a bit less demanding on weird language edge cases is a good idea, ultimately making reimplementations of the nix language more feasible in the future. | 15:22:13 |
 K900 | No, but Lix can deprecate valid footguns | 15:42:00 |
| K900 | And removing them from nixpkgs seems worth it | 15:42:06 |
| 19 Feb 2026 |
| K900 | So uh | 10:38:55 |
| K900 | Pipewire 1.6.0 just got tagged | 10:39:02 |
| K900 | Is it too late | 10:39:22 |
| K900 | It's a big feature update and also technically security | 10:40:30 |
 Vladimír Čunát | We have basically all binaries already. | 10:40:40 |
| Vladimír Čunát | But the regressions are the issue now. | 10:40:48 |
| Vladimír Čunát | * We have basically all staging-next binaries already. | 10:41:09 |
 dramforever | how bad is pipewire anyway clicks the pr oh no | 10:42:09 |
| K900 | It's like, most of Qt, at the very least | 10:49:26 |
 emily | I can probably take a brief look at Darwin if there's anything particularly pertinent | 11:06:13 |
| Vladimír Čunát | This thing has many reverse dependencies:
https://github.com/NixOS/nixpkgs/pull/488406#issuecomment-3921541704 | 11:09:21 |
| Vladimír Čunát | Then twisted is at least very unreliable, but I think the worst cases were retried many times and succeeded in the end. | 11:10:19 |
| Vladimír Čunát | And qt6.qtbase is broken on darwin:
https://github.com/NixOS/nixpkgs/pull/486498#issuecomment-3913199483 | 11:10:44 |
| Vladimír Čunát | (Let me stop; I don't want to overload you.) | 11:11:07 |
| emily | taking a look | 11:35:27 |
| emily | qtbase is failing because libX11 is getting propagated in and turning on the XCB feature which results in it trying to compile the GNOME theme and stuff | 11:54:35 |
| emily | actually no, that feature is not listed as on in the config output | 11:54:55 |
| emily | qt_internal_extend_target(Gui CONDITION UNIX AND (((QT_FEATURE_xcb OR NOT MACOS) AND (QT_FEATURE_xcb OR NOT UIKIT)) OR QT_FEATURE_wayland)
so why is this check passing. cc K900?
| 11:55:08 |
| emily | oh | 11:55:40 |
| emily | it's getting Wayland?! | 11:55:43 |
| emily | it's listed as off in the config though… but it's definitely finding something | 11:55:58 |
| emily | cmake/QtProcessConfigureArgs.cmake sets set(QT_FIND_ALL_PACKAGES_ALWAYS ON) which disables the platform guards on looking for X11 and Wayland libs | 11:58:27 |
| emily | but I don't know if we end up using that file either :) | 11:58:59 |
| emily | anyway | 11:59:38 |
| emily | I believe https://github.com/NixOS/nixpkgs/pull/477359 was the culprit | 11:59:45 |
