| 18 Dec 2025 |
 @rosssmyth:matrix.org | Yes, but it is not our part that breaks! | 15:59:49 |
| @rosssmyth:matrix.org | Also the force feedback motor in the control stick of a certain Caterpillar tractor. Can't remember which model. | 16:00:32 |
 emily | FWIW, you can always self-host GitHub runners on your own hardware, like uh, your laptop, and there are third-party providers for faster runners like https://namespace.so/ that I believe are cheaper than GitHub's. though maybe not any more, GitHub are lowering prices: https://resources.github.com/actions/2026-pricing-changes-for-github-actions/ (…while also trying to charge for self-hosted runners admittedly) | 16:01:15 |
| emily | all that said, if you can make multilib work to the point of being able to target i686 with the standard x86_64-linux environment, I think we'd be happy to accept that | 16:01:39 |
| @rosssmyth:matrix.org | Anyways, yes there are two laptops that are very load bearing to all of our firmware. Though not as much anymore since I have been moving things to Meson+Nix as I work on them. | 16:01:40 |
| emily | we do have some ancient "multilib" stdenv stuff though I doubt it works very well | 16:01:54 |
| @rosssmyth:matrix.org | I did see a toggle for multilib buried in there | 16:02:21 |
| emily | see all-packages.nix around wrapCCMulti = … and beyond, pkgs/test/cc-wrapper/multilib.nix… I expect it's pretty bitrotted | 16:03:05 |
| emily | I wouldn't worry too much about breaking changes to embedded stuff though | 16:03:44 |
| emily | the reasons it needs changes are the reasons nobody uses it | 16:03:56 |
