| 21 Nov 2021 |
 sterni | ghcHEAD is in a bit of a state as well, but I don't care too much atm, may be more interesting to replace it with ghc921 for now | 23:45:19 |
| sterni | also the platform list in release-cross.nix is a bit weird, we should maybe sync it with lib.systems.examples | 23:45:48 |
 alexfmpe | In reply to @sternenseemann:systemli.org
you'll find the 8.10.7 failures by searching for haskellPackages did you mean cross builds of ghc or misc haskell packages? | 23:47:52 |
| alexfmpe | I seem to remember ghc 9.x being broken on mingw upstream | 23:50:22 |
| alexfmpe | https://gitlab.haskell.org/ghc/ghc/-/issues/18143 | 23:51:09 |
| alexfmpe | and voila https://hydra.nixos.org/build/158165265/nixlog/1 | 23:51:37 |
| alexfmpe | oh hmm two related PRs landed one week ago, that seems worth trying | 23:53:09 |
| 22 Nov 2021 |
| alexfmpe | indeed, the current ghcHead build goes further now and hits some missing term.h error which seems to be fixed with an ncurses dep | 00:23:00 |
| alexfmpe | build still going, here's hoping it actually works now | 00:23:27 |
| alexfmpe | sterni: https://github.com/NixOS/nixpkgs/pull/146959 | 02:57:06 |
| sterni | In reply to @alexfmpe:matrix.org
did you mean cross builds of ghc or misc haskell packages? cross compilation of haskellPackages.hello is tested which also requires us to build a haskell cross compiler | 11:18:41 |
| sterni | thanks a lot for the patch! | 11:19:06 |
| sterni | John Ericson: interesting long term project could be improving the story with cc-wrapper and multi target compilers or in nixpkgs in general, given that GHC is projected to become multi target in the future as well | 11:19:47 |
 John Ericson | sterni: alexfmpe yeah I really hope we can push through the GHC splitting and GCC splitting by, say, 22.05 | 16:33:51 |
| alexfmpe | John Ericson I think you're familiar with the upstream happenings - does it make sense to try and get the ghc fixes added to 9.0 & 9.2 as patches or something? otherwise those will just stay broken | 16:53:04 |
| alexfmpe | I mean, I assume upstream won't backport cross-compiling things | 16:53:49 |
| John Ericson | yeah a problem a lot of my recent (this fall) GHC work depends on sylvain's finally making the compiler target agnostic messing with the primops | 16:54:44 |
| John Ericson | and that cannot be backported very easily because it breaks things that use GHC.Prim | 16:55:08 |
| alexfmpe | oh hmm I meant "backporting" only on nixpkgs | 18:14:05 |
| alexfmpe | in case it was simple and the upstream backport didn't happen because it was cross related only | 18:14:33 |
| alexfmpe | but if it's hard to backport upstream, then it'd be harder on nixpkgs probably | 18:14:44 |
| alexfmpe | "eventually cross compiling" it is | 18:15:10 |
|  smbarber left the room. | 20:14:08 |
| 23 Nov 2021 |
|  Server Stats Discoverer (traveler bot) left the room. | 00:55:08 |
|  Pierre-Olivier Rey joined the room. | 12:51:45 |
| Pierre-Olivier Rey | Hello, I'm just beginning with nix for a school project and I have i'm trouble to have a devShell with clang able to compile with -m32 (we generate code using llvm backend) ; do you know how to do it in a flake?
Sorry if the question seems a bit odd :') | 12:54:05 |
 symphorien | you should probably use pkgs.pkgsi686Linux.clangStdenv | 12:54:53 |
| symphorien | see this for more context about using a different stdenv: https://nixos.wiki/wiki/Using_Clang_instead_of_GCC#With_nix-shell | 12:55:15 |
 Alyssa Ross | sternenseemann: there's nothing left to do on https://github.com/NixOS/nixpkgs/pull/144749, right? | 20:11:35 |
| sterni | Alyssa Ross: at least nothing that's obvious to me! | 21:08:15 |
