| 27 Apr 2025 |
 trofi | It's embedded at https://github.com/NixOS/nixpkgs/blob/1ee8b713821882c66f5ecfde5c4e5874b1cb5e2f/pkgs/development/libraries/glibc/default.nix#L113 | 05:25:57 |
| trofi | Specifically glibc ever gets build by a handicap xgcc from early stages during the bootstrap:
$ nix-store --query --graph $(nix-instantiate -A glibc) |& grep -P -- '.*gcc.*->.*glibc'
"qn5vjz1dxigfax34am4k4zxcx1y3q3v0-bootstrap-stage2-gcc-wrapper-14.2.1.20250322.drv" -> "q9rikjkwdf1qnlypwxr78jfmh9q99k1g-glibc-2.40-66.drv" [color = "blue"];
"x455my7bxwf4p7wngvwi4m23lf1np2vx-xgcc-14.2.1.20250322.drv" -> "q9rikjkwdf1qnlypwxr78jfmh9q99k1g-glibc-2.40-66.drv" [color = "red"];
| 05:30:08 |
| trofi | * It's not ideal, but it's intentional by nixpkgs. | 05:40:28 |
|  haylin joined the room. | 05:56:33 |
| 28 Apr 2025 |
|  chvp left the room. | 14:28:33 |
| 30 Apr 2025 |
 aleksana 🏳️⚧️ (force me to bed after 18:00 UTC) | There isn't any consumer for these micro-architectures https://github.com/NixOS/nixpkgs/blob/master/lib/systems/architectures.nix | 12:08:25 |
| aleksana 🏳️⚧️ (force me to bed after 18:00 UTC) | How are they supposed to be applied to a package or stdenv? | 12:08:54 |
| aleksana 🏳️⚧️ (force me to bed after 18:00 UTC) | I'm hitting a similar situation for loongarch64, where I'm advised to raise march level by default, but I still want to preserve the possibility to build with lower march level. See doc https://github.com/loongson/la-softdev-convention/blob/master/la-softdev-convention.adoc#10-operating-system-package-build-requirements | 12:10:29 |
 Sandro 🐧 | I use it like this https://github.com/NuschtOS/nixos-modules/blob/main/modules/simd.nix#L20 | 12:42:23 |
|  sheeldotme joined the room. | 14:47:37 |
| aleksana 🏳️⚧️ (force me to bed after 18:00 UTC) | Okay, so any usage of lib.systems.architectures.features? | 14:56:46 |
 Tristan Ross | In reply to @aleksana:mozilla.org
Okay, so any usage of lib.systems.architectures.features? Not that I'm aware of but I'll have to look at what it is first. | 15:00:14 |
| aleksana 🏳️⚧️ (force me to bed after 18:00 UTC) | It doesn't look like an exhausive list of either gcc features or /proc/cpuinfo | 15:03:42 |
| aleksana 🏳️⚧️ (force me to bed after 18:00 UTC) | Also what is the difference between lib.systems.platforms and lib.systems.examples? Why is armv7a-android in platforms while aarch64-android in examples? | 15:59:36 |
| 1 May 2025 |
|  Rosuavio changed their display name from Rosario Pulella to Rosuavio. | 20:09:28 |
| 3 May 2025 |
 Randy Eckenrode | Any thoughts on changing the target check to treat arm64-apple-darwin and arm64-apple-macosxY.X as equivalent? Swift and SwiftPM heavily use the latter. It’s apparently the way that setting the deployment target is done. | 17:06:05 |
 Alyssa Ross | I think we have to pick one | 17:09:21 |
| Alyssa Ross | Assuming I'm correctly understanding what you mean by "target check"? | 17:10:12 |
| Randy Eckenrode | The one that spams lots of warnings when you clang-wrapper -target <some triple>, and the triple is different from what the wrapper was built for. | 17:11:16 |
| Alyssa Ross | Oh that sounds fine | 17:11:46 |
| Alyssa Ross | Assuming LLVM treats them the same | 17:12:00 |
| Randy Eckenrode | You can change the deployment target a bunch of different ways. The way Swift does it (and when invoking Clang) is via the triple. The wrapper tries to set the target via -mmacos-version-min=. I think there’s another way it can also fall back to do. | 17:12:03 |
| Randy Eckenrode | So maybe Darwin can settle on following Swift’s lead and consolidate all those. Probably a 25.11 thing. | 17:12:38 |
| Randy Eckenrode | It would also fix the following warnings when SwiftPM invokes Clang.
Warning: supplying the --target arm64-apple-macosx10.13 != arm64-apple-darwin argument to a nix-wrapped compiler may not work correctly - cc-wrapper is currently not designed with multi-target compilers in mind. You may want to use an un-wrapped compiler instead.
| 17:14:42 |
| Randy Eckenrode | * It would also fix the following warnings when SwiftPM invokes Clang.
clang: warning: overriding '-mmacos-version-min=11.3' option with '-target arm64-apple-macosx10.13'
| 17:15:07 |
| 4 May 2025 |
| aleksana 🏳️⚧️ (force me to bed after 18:00 UTC) | Two PRs aiming to fix "cross compilation" between different march
https://github.com/NixOS/nixpkgs/pull/403549
https://github.com/NixOS/nixpkgs/pull/403960 | 07:08:19 |
 Morgan (@numinit) | Randy Eckenrode: Hey there, Darwin question for you. ponyc does this and we have to patch:
https://github.com/NixOS/nixpkgs/blob/master/pkgs/by-name/po/ponyc/fix-darwin-build.patch
However, hydra is complaining about:
ld: warning: directory not found for option '-L/nix/store/6fg11q5s4xpgfpvps1hy7w9vjhfl14mx-Libsystem-11.0/lib'
ld: library not found for -lSystem
Feel like I'm just missing something stupid here.
| 21:58:08 |
| Randy Eckenrode | Don’t use darwin.Libsystem. It’s an empty stub with nothing in it. It’s scheduled for removal too. If you need to find the location of libSystem.tbd, it’s in $SDKROOT/usr/lib. | 22:00:16 |
| Morgan (@numinit) | Gotcha, that explains it | 22:01:12 |
| Morgan (@numinit) | Thanks, I'll give that a shot | 22:01:27 |
