| 12 Jun 2021 |
 toonn | That's worse than the local lottery : ) | 19:38:01 |
 LnL | yeah so in practice it means the jobset will wait until the queue is free | 19:38:50 |
| LnL | main problem currently is probably a staging-next eval + a staging-next for 21.05 | 19:41:02 |
| LnL | https://monitoring.nixos.org/grafana/d/MJw9PcAiz/hydra-jobs?orgId=1&refresh=30s&viewPanel=2&from=now-7d&to=now | 19:43:19 |
| LnL | it would help quite a bit if the M1 machines where also marked as x86 compatible builders | 19:45:04 |
| LnL | or are there too many problems with doing that? | 19:45:18 |
| toonn | Thanks for the graph. Looks like 2-3 more days? | 19:46:03 |
| toonn | No idea. | 19:46:09 |
| toonn | I was thinking that too : ) | 19:46:14 |
| LnL | at least half of the apple hardware is idle at the moment (and the ones which are not years old) | 19:46:22 |
| toonn | They'd have to prioritize aarch64 jobs ofc. | 19:46:28 |
| LnL | not sure what hydra does with priorities between platforms, but that wasn't really a problem with i686 before AFAIK | 19:48:00 |
|  magpi joined the room. | 23:55:38 |
| 13 Jun 2021 |
| toonn | If anyone's familiar with missing framework re-exports I'd appreciate a hint, https://git.io/JZbRT : ) | 12:23:02 |
| LnL | what's the context here? SecurityFoundation doesn't reexport anything from what I can tell | 14:38:37 |
| LnL | nix log /nix/store/dn63dxzz06k8g39lnvankk5schq8b97v-apple-framework-SecurityFoundation | 14:39:36 |
 thefloweringash | I’m assuming this is from a branch that uses 10.13 frameworks? The substitutions are built from the frameworks structure file, so it should be sufficient to ensure the line for SecurityFoundation includes SecurityFoundation = { Security /* … */ }; | 14:42:49 |
| thefloweringash | There’s also a little bit of mess here that’s my fault. The darwin-stubs are pre-mangled for nix substitution with the substitutions nferred from dependencies by name. but to use the apple sdk directly I made a different tool that doesn’t require pre-mangling and uses prefix rewrites that are built from passthru data on the dependencies | 14:45:21 |
| LnL | doesn't seem all that bad to me to have non existing references unless things are configured properly | 14:46:43 |
| LnL | and yeah if this is SecurityFoundation itself I'd expect this has to state the dependency https://github.com/NixOS/nixpkgs/blob/0a47fada9e774976b4e0ef92880e3da282b07782/pkgs/os-specific/darwin/apple-sdk/frameworks.nix#L100 | 14:47:34 |
| thefloweringash | yeah, both ways check the existence of all re-exported frameworks. it's just a little messy to do things differently | 14:51:04 |
| LnL | ah cool | 14:51:49 |
| toonn | Should I regenerate that file somehow or add stuff manually? | 15:26:43 |
| LnL | As far as I know that's not auto generated, but might have changed | 15:51:19 |
| toonn | LnL: I have my SDK bump branch building again. What test do you tend to use to verify everything works? | 19:17:20 |
| toonn | I guess I'll add all the new apple-source-releases one by one. | 19:17:36 |
| LnL | if you want to do some more extensive testing trying things from different ecosystems is usually a good idea | 19:24:03 |
| LnL | there's also the darwin-tested job which has some specific things included https://hydra.nixos.org/job/nixpkgs/trunk/darwin-tested#tabs-constituents | 19:24:49 |
| toonn | LnL: Different ecosystems? Like pythonPackages, etc.? | 19:28:21 |
| toonn | And do you know why apple-source-releases/default.nix lists everything explicitly rather than importing the generated `macos-10.whatever.nix`? | 19:29:08 |
