| 29 Jan 2026 |
 eveeifyeve | Nvm this is a quick and really easy fix. | 10:58:48 |
 Randy Eckenrode | My branch uses staging as a base, so it (almost always) does a full bootstrap from source. | 11:35:26 |
| Randy Eckenrode | sudo -i nix-env -e nix IIRC. | 11:36:26 |
| Randy Eckenrode | Usually the Nix from the installer is left in root’s profile, but you don’t need it if your Nix is managed by nix-darwin. | 11:37:06 |
| eveeifyeve | Yeah so I removed it.. | 11:37:23 |
| eveeifyeve | Nix-darwin should create a first install script that uninstalls the one from the installer. | 11:38:16 |
| eveeifyeve | We ❤️ llvm taking the longest to build in staging.... | 11:46:02 |
| eveeifyeve | * We ❤️ llvm taking the longest to build in staging.... 😆 | 11:46:11 |
| eveeifyeve | Randy Eckenrode: You were right I feel it does take 11 hours. I am at 6 hours and it's not built yet. | 13:14:57 |
| eveeifyeve | It's hit to the point of the swift-cxx-bootrap (buildphase) | 13:15:18 |
| eveeifyeve | Don't recommend anyone trying to build swift 6, but I am because I want to help out Randy Eckenrode. | 13:36:22 |
| Randy Eckenrode | The biggest help will be testing. Try building packaged Swift applications as well as ones in a dev shell. It should work. The goal is to make it a good experience for packaging and for local development. | 13:43:13 |
| Randy Eckenrode | If there are packages that need things that have been drops, I want to know, so I can investigate why. | 13:43:37 |
| Randy Eckenrode | For example, swift.swift is gone. Were you just doing that to get an unwrapped compiler? | 13:44:07 |
| Randy Eckenrode | The compiler is now unwrapped. | 13:44:29 |
| Randy Eckenrode | Using xcbuild with Swift dependencies? I want to make that work. | 13:45:21 |
| Randy Eckenrode | Using NIX_SWIFTFLAGS_COMPILE? It’s no longer supported due to Swift being unwrapped. How was that being used? | 13:46:15 |
| Randy Eckenrode | Need to link non-Swift dylibs? Does that still work? | 13:46:35 |
| Randy Eckenrode | If there are easy test cases, they can be put in passthru.tests. I’ve already updated the C++ interop test for both-way interop. | 13:47:30 |
| Randy Eckenrode | We can now use Swift from C++ in addition to C++ from Swift. | 13:48:02 |
| Randy Eckenrode | Other bonus things I hope will work is cross-compilation and using stdlibs for other targets. | 13:48:37 |
| Randy Eckenrode | Static is another one that needs testing. | 13:49:48 |
| Randy Eckenrode | I would also like to support Windows, WASM, and Android; but those are bonuses. | 13:54:47 |
 Ihar Hrachyshka | link to a bug report and/or details of what's going on at least? | 15:29:29 |
| eveeifyeve | I don't think they reported the issue directly, but I know about it hearing from ericson. | 15:30:18 |
| Ihar Hrachyshka | ericson? a man? a company? :) | 15:30:42 |
| eveeifyeve | John Ericson. | 15:31:08 |
| Ihar Hrachyshka | I guess I will proceed with bisect but...
$ git bisect visualize --no-merges --oneline
d4fd918e2cc2 libarchive: fix cygwin build
f4cd05d2e7c9 haskellPackages: mark builds failing on hydra as broken
4922225f956f haskellPackages.rhine-bayes: downgrade to match rhine
51c78494b9ff openldap: use --replace-fail instead of --replace
b770f35b4de1 openldap: fix build with structuredAttrs
79429223426a moltenvk: 1.4.0 -> 1.4.1
70f40e0a3b82 python313Packages.aiodns: 3.5.0 -> 3.6.1
50776d0e56eb apple-sdk: drop `FamilyDisplayName`
176c6512ba96 darwin.libcxx: 19.1.2+apple-sdk-15.5 -> 20.1.0+apple-sdk-26.0
Randy Eckenrode fyi
| 17:14:36 |
| Ihar Hrachyshka | anyone up to try the revert of 176c6512ba96 while I'm proceeding with the current bisect? | 17:18:01 |
|  shorden joined the room. | 17:26:56 |
