| 13 Jan 2026 |
 Randy Eckenrode | Maybe the same for anything not in the stdenv. | 13:55:38 |
| Randy Eckenrode | * | 13:55:48 |
| Randy Eckenrode | Right now though the goal is to get to the point where the packaging is complete, then it can be refined. | 13:56:09 |
| 14 Jan 2026 |
| Randy Eckenrode | Oh, I see. I dropped the lib output from swiftc but forgot to replace placeholder "lib" with placeholder "out". Whoops. | 01:13:17 |
| Randy Eckenrode | https://www.macrumors.com/2026/01/12/google-chrome-150-last-version-support-macos-12/ | 02:57:37 |
| Randy Eckenrode | * | 02:58:25 |
| 15 Jan 2026 |
| Randy Eckenrode | emily I committed your suggestion. Should I go ahead and merge when ready? | 00:43:32 |
 emily | I hit the button ^^ | 02:52:34 |
| emily | sorry for taking a while to get back to that! | 02:52:38 |
| Randy Eckenrode | No worries. Now the only extra thing I’m carrying is my KosmicKrisp branch. | 03:23:20 |
| Randy Eckenrode | (Other than the Swift work.) | 03:23:30 |
| emily | what's the status of KosmicKrisp again? | 03:35:07 |
| emily | i should check how it performs with my mpv setup at some point | 03:35:17 |
| Randy Eckenrode | Don’t think it can do DXVK yet. Performance is about 25~30% of MoltenVK. They’re prioritizing conformance first. Optimization will come later. | 03:58:30 |
| emily | ouch, okay | 03:58:48 |
 Gaétan Lepage | What should we do about compiler-rt-libc-20.1.8-x86_64-darwin who has been failing since the staging-next merge in late December?
Should we mark it as broken? | 12:45:19 |
| Randy Eckenrode | I thought there was a PR to fix it. | 14:30:28 |
| Randy Eckenrode | Maybe it was just discussed …. | 14:31:39 |
| emily | @ihar.hrachyshka:matrix.org was looking at it | 16:02:47 |
| 17 Jan 2026 |
 Ihar Hrachyshka | ehm... previous attempt to build with probable fix failed on unrelated package and I forgot about it.
https://github.com/NixOS/nixpkgs/pull/480889
will run the build through the night again and report back on the PR how it went. | 02:59:43 |
| Randy Eckenrode | This is annoying. Swift uses getMainExecutable from LLVM to get the location of the compiler binary, which resolves symlinks. That means swiftc symlinked into the swift package will look for the stdlib in the swiftc package instead where it doesn’t exist. 😓 | 03:58:27 |
|  @folliehiyuki:envs.net left the room. | 09:11:47 |
|  robbie-w joined the room. | 15:22:29 |
| Randy Eckenrode | https://forums.swift.org/t/zlib-support-in-the-swift-ecosystem/83792/11 | 15:38:22 |
 alexfmpe | can one use gcc instead of clang on aarch64-darwin?
tried nix-build -E "with import ./. {}; pkgs.hello.override { stdenv = pkgs.gccStdenv; }" which fails with
configure:5654: checking whether the C compiler works
configure:5676: gcc conftest.c >&5
ld: library not found for -liconv
collect2: error: ld returned 1 exit status
...
configure:5728: error: C compiler cannot create executables
See 'config.log' for more details
| 16:35:46 |
 K900 | Not well | 16:36:00 |
| Randy Eckenrode | Does the GCC stdenv not include the SDK? It should, and libiconv should be propagated. | 16:38:25 |
| Randy Eckenrode | But using GCC is a bad idea unless you have no choice. It doesn’t support availability annotations. It’s not set up to use libc++, which everything C++ on Darwin is linked against. (GCC can use it, but we don’t support that AFAIK.) | 16:40:05 |
| alexfmpe | hmm I mainly wanted it to test some stuff that is currently broken ish with clang and didn't want to investigate/fix right now | 16:41:39 |
| alexfmpe | it's more of a "would unlock progress on current task immediately" rather than a "I'll actually rely on this for anything" | 16:42:31 |
