| 3 Dec 2025 |
 Randy Eckenrode | There was also a change to enforce availability annotations. | 15:08:42 |
| Randy Eckenrode | But the missing symbols appear to be internal to Dovecot. Something in its build is getting confused. | 15:09:11 |
 Katalin 🔪 | fwiw there is a “mostly done just needs a little polish” update to 2.4 or whatever and I’m already using it | 20:40:34 |
| Katalin 🔪 | this is the PR, if you want to see if it works for you: https://github.com/NixOS/nixpkgs/pull/410912
I do have my own changeset that I run on my server that has diverged a bit but only in the nixos module so it shoulndn’t be necessary here | 20:48:55 |
| 4 Dec 2025 |
| Randy Eckenrode | It’s amazing that Apple managed to successfully compile Swift 6.2 for release with Xcode. I got a bootstrap compiler to build with macros enabled. It crashes building the Swift Compiler Driver. | 00:52:54 |
| Randy Eckenrode | Why would nativeBuildInputs use the dev output of a derivation? | 02:11:10 |
| Randy Eckenrode | Oh, I need to fixup nix-support. | 02:20:39 |
| Randy Eckenrode | /nix/store/7x3jpimm3qaffhdcgj23rd51gpmy4p2f-apple-sdk-26.0/Platforms/MacOSX.platform/Developer/SDKs/MacOSX26.0.sdk/usr/lib/swift/Swift.swiftmodule/arm64e-apple-macos.swiftinterface:34793:42: error: Using type 'Int' can cause metadata allocation or locks
| 04:48:13 |
| Randy Eckenrode | 34791 |
34792 | let queryVersion = (Int(major), Int(minor), Int(patch))
34793 | let major32 = Int32(truncatingIfNeeded:Int(queryVersion.0))
| `- error: Using type 'Int' can cause metadata allocation or locks
34794 | let minor32 = Int32(truncatingIfNeeded:Int(queryVersion.1))
34795 | let patch32 = Int32(truncatingIfNeeded:Int(queryVersion.2))
| 04:50:31 |
| Randy Eckenrode | That looks really familiar. I swear I fixed something like this before. | 04:50:41 |
| Randy Eckenrode | https://github.com/swiftlang/swift/pull/77780, but shouldn’t that be in Swift 6.2? | 05:09:55 |
 xored | hmm I had aerospace and jankyborders through nix-darwin, I'm considering moving to hm so I can see the config and maybe get stylix support, I disabled the service on nix-darwin and yet even after telling me the packages were uninstalled, it is launching on login, jankyborders and all, am I assuming wrong that I don't need to do manual cleanup? | 05:12:13 |
| xored | I can still see the daemons in ~/Library/LaunchAgents | 05:14:32 |
| xored | maybe it's a bug? while removing them it did ask show they were write-protected, but I didn't have to sudo or anything | 05:16:07 |
|  isabel changed their profile picture. | 16:41:41 |
 Capypara | Hi! A user for a flake of mine uses macOS. Unfortunately I don't own a Mac, so I have no idea of the nix-darwin ecosystem. They get this error when running an activation script. nixpkgs is the latest nixos-25.11 (but a week old one had the same result):
last 9 log lines:
> Running phase: patchPhase
> Running phase: updateAutotoolsGnuConfigScriptsPhase
> Running phase: configurePhase
> no configure script, doing nothing
> Running phase: buildPhase
> Traceback (most recent call last):
> File "/nix/store/5g45axkqi74cigpqpz3s9yayn80k6dv7-build-config.py", line 4, in <module>
> import yaml
> ModuleNotFoundError: No module named 'yaml'
For full logs, run:
nix log /nix/store/rs6nnmijyfhy9p4q1d1iyf2w4a2j9c44-riptide-build-config.drv
| 17:15:28 |
| Capypara | nativeBuildInputs & buildPhase:
nativeBuildInputs = [
(pkgs.python314.withPackages (python-pkgs: [
python-pkgs.pyyaml
python-pkgs.mergedeep
]))
];
buildPhase = ''
runHook preBuild
python3 \
$src \
${lib.escapeShellArg (builtins.toJSON cfg)} \
${hostsFile} \
./config.yml
runHook postBuild
'';
| 17:15:45 |
| Capypara | Anybody got an idea what the problem here could be? | 17:15:55 |
 leona | https://github.com/NixOS/nixpkgs/issues/461884 fix will come in the next 1.5 weeks probably | 17:16:56 |
| Capypara | ah okay! I thought those landed already, thanks! | 17:18:16 |
| leona | As nobody built something else, they need a full stdenv rebuild on darwin | 17:18:52 |
| Capypara | ah! I see | 17:19:08 |
| Randy Eckenrode | Darwin tries not to pull too much, but it needs Python during its bootstrap, which makes Python-related fixes into full rebuilds. | 19:23:11 |
 kfiz | Yes. | 22:25:52 |
| kfiz | I'll try this. Thanks. | 22:26:53 |
 toonn | .oO(Does Pypy change less often and is it capable enough?) | 23:07:28 |
| toonn | (Just an intrusive thought : )) | 23:07:47 |
| Randy Eckenrode | It’s a few versions behind CPython. The bootstrap story doesn’t look too bad. We would have to replace the top-level Python during the bootstrap. | 23:11:51 |
| Randy Eckenrode | There’s enough stuff that depends on CPython that updates have to go through staging anyway. It doesn’t seem worth the effort. | 23:12:29 |
| Randy Eckenrode | So PyPy, Rust, Go, Haskell, and what else all bootstrap from an upstream binary? | 23:14:01 |
