| 24 Sep 2025 |
 emily | did you figure this out? :) | 18:27:04 |
| emily | I like to drop GHCs to relax after hard work. | 18:27:14 |
 K900 | No, I can't find it now | 18:48:22 |
| K900 | I don't know what the fuck that was | 18:48:25 |
| emily | it seems like it's not anything in the release jobset at least. even with allowUnfree | 20:49:48 |
| 25 Sep 2025 |
 Vladimír Čunát | When I look at nixpkgs-unstable, i686 ghc and pandoc are not in cache.nixos.org. | 05:04:50 |
| Vladimír Čunát | $ nix-env -p /tmp/nix-profile -f. -qas -A pkgsi686Linux.ghc | cat
--- ghc-9.6.7
| 05:06:12 |
| emily | I was thinking it might be via some non-free thing, but I guess if anything substantial required people to build GHC themselves we'd hear about it | 15:00:46 |
 sterni | mysterious I was convinced that we build something that pulls in GHC on i686-linux | 17:28:26 |
| sterni | Maybe it was flac? https://github.com/NixOS/nixpkgs/pull/437543 | 17:28:52 |
| K900 | Could be | 17:29:41 |
| emily | yeah, even without that PR it looks like it'd be harmless to drop on i686-linux | 18:35:53 |
| emily | since we'd just get no man pages for something that is surely only being used as a component of stuff to host on an x86_64-linux system that can get its own flac manpages | 18:36:23 |
| 26 Sep 2025 |
 ElvishJerricco | cursed knowledge of the day: the linux kernel depends on systemd because of rsync: https://github.com/NixOS/nixpkgs/pull/446259 | 06:55:26 |
| ElvishJerricco | (hi I'm trying to at least a little bit minimize the rebuild count of systemd) | 06:56:47 |
| Vladimír Čunát | gcc 15 seems to be giving up NixOS 25.11, but glibc upgrade not yet 😅 | 11:32:44 |
| Vladimír Čunát | Maybe I'd prefer not to rush it in the last allowed week either, given the concerns (e.g. around executable stack). | 11:39:25 |
 Marie | For gcc 15 we would also need this, since gcc 15 cross compilers are broken right now, because our ancient patch doesn't apply anymore
https://github.com/NixOS/nixpkgs/pull/445308/ | 13:31:48 |
| K900 | Beautiful | 13:32:54 |
 ghpzin | Could you ref gcc15 PR, so people know there is a fix for pkgsCross/pkgsStatic | 13:35:22 |
| ghpzin | Could you ref gcc15 PR, so people know there is a fix for pkgsCross/pkgsStatic
| 13:35:29 |
| ghpzin | Could you ref gcc15 PR, so people know there is a fix for pkgsCross/pkgsStatic
(kind of hard to find these, already had rebase of that patch in my fork) | 13:36:12 |
| ghpzin | Could you ref gcc15 PR (https://github.com/NixOS/nixpkgs/pull/440456), so people know there is a fix for pkgsCross/pkgsStatic
(kind of hard to find these, already had rebase of that patch in my fork) | 13:36:51 |
| Marie | sure | 13:38:50 |
| ghpzin | Maybe somebody already made PR to update musl bootstrap files, but it is also impossible to find. | 13:55:23 |
| ghpzin | Maybe somebody already made PR to update musl bootstrap files (to fix pkgsMusl), but it is also impossible to find. | 14:00:17 |
| emily | GCC 15 also needs some Darwin stuff I have locally | 18:31:27 |
| emily | probably 26.05 makes the most senes at this point | 18:31:34 |
| emily | 💜 to ghpzin for all the fix PRs though | 18:31:39 |
 Sami Liedes | Ah, I may have found the room! I've been patching a bit stuff in nixpkgs which I think probably also applies to gcc 15. (Specifically what I've been fixing is packages building with GCC 14 and -D_GLIBCXX_DEBUG, which requires things that declare themselves as random access iterators to have operator<=; but I read in some bug report that gcc 15 is also similarly picky about some iterator operators.)
Is there a branch somewhere where the proposed gcc 15 patches are collected? I could switch to that and start fixing for it with -D_GLIBCXX_DEBUG, so I would likely at the same time fix them for gcc 15.
| 18:32:26 |
