| 24 Sep 2024 |
 Tristan Ross | It's a little scary to look at but glad it's in. | 21:13:22 |
|  ckie (they/them) changed their profile picture. | 23:19:25 |
| 25 Sep 2024 |
|  @janne.hess:helsinki-systems.de joined the room. | 10:00:48 |
| @janne.hess:helsinki-systems.de | Hi, I'm currently trying out stuff with python and the respective propagated inputs. My goal is to only propagate PYTHONPATH and not other variables like PATH. I tried that by adding a nix-support/python-dependencies file which works great but I struggle with parsing it. The current implementation which takes deps from propagated-build-inputs just calls addEnvHooks but I don't think I can do this because that would also add all pages to PATH. So I started implementing my own logic that recurses over the packages but I struggle to figure out which variables I should inspect. I am sure about propagatedBuildInputs and nativeBuildInputs but it feels like I shouldn't start my own list. Is there something in stdenv I can use to not reimplement these lists? | 10:04:28 |
 emily | Artturin: not sure we should ignore that treewide for blame as it obscures "wait, why is this using hostPlatform?" | 15:32:20 |
| emily | (as opposed to something like nixfmt which is purely presentational) | 15:32:40 |
| emily | (posted on github instead actually) | 15:34:15 |
| 27 Sep 2024 |
|  @yannis:mozilla.org changed their display name from yannis to yannis|pto back oct 16. | 15:17:29 |
| 1 Oct 2024 |
|  -_o joined the room. | 21:00:11 |
| 4 Oct 2024 |
|  @terrorjack:matrix.terrorjack.com joined the room. | 17:38:12 |
| 7 Oct 2024 |
|  Sam Lehman changed their profile picture. | 14:24:11 |
| 9 Oct 2024 |
| Tristan Ross | https://github.com/NixOS/nixpkgs/pull/335023 I hope we can get this in soon, it'll fix aarch64-linux building of the kernel | 18:15:00 |
 Alyssa Ross | I'm still really unsure about it | 19:36:37 |
| Tristan Ross | In reply to @qyliss:fairydust.space
I'm still really unsure about it In what way? It being necessary? Because it does make pkgsLLVM.linux build correctly on aarch64-linux. | 19:37:44 |
| Alyssa Ross | In the way I said in my comment | 19:39:20 |
| Alyssa Ross | I don't think there's a one way 64-bit<->32-bit mapping. | 19:39:27 |
| Alyssa Ross | * | 19:39:41 |
| Tristan Ross | Oh | 19:40:25 |
| Alyssa Ross | or at least, I'm very not confident that there is one | 19:40:26 |
| Tristan Ross | Could also detect if -m32 or -m64 is specified but aside from -target and -m, idk any other ways. | 19:41:57 |
| Tristan Ross | And this mainly only matches on the arch name so other vendor/os names won't trip up the detection at least. | 19:42:29 |
| Tristan Ross | At least the kernel only uses -target it seems for the case I ran into. | 19:43:39 |
| 10 Oct 2024 |
 linj | Sorry for cross-posting. What is the difference between stdenv.shell and runtimeShell? I want to do something like make SHELL=${runtimeShell} during build. | 02:14:12 |
| Tristan Ross | Idk either, I can figure out from grepping | 02:15:07 |
| Tristan Ross | stdenv.shell seems to be the one which can run the builder. Not sure what runtimeShell = prevStage.ccWrapperStdenv.shell; exactly means. | 02:16:59 |
| Tristan Ross | Oh, they're mostly the same | 02:17:32 |
| linj | In reply to @rosscomputerguy:matrix.org
stdenv.shell seems to be the one which can run the builder. Not sure what runtimeShell = prevStage.ccWrapperStdenv.shell; exactly means. Oh, I do not mean this runtimeShell, I mean pkgs.runtimeShell. | 02:18:16 |
| Tristan Ross | In reply to @me:linj.tech
Oh, I do not mean this runtimeShell, I mean pkgs.runtimeShell. Oh, it's an alias of runtimeShellPackage | 02:18:53 |
| Tristan Ross | But it references the actual exec | 02:19:06 |
| Tristan Ross | Basically like doing lib.getExec pkgs.runtimeShellPackage which the runtime shell package is bash | 02:19:34 |
